diff --git a/setup.py b/setup.py index fe21851..ec39d15 100755 --- a/setup.py +++ b/setup.py @@ -1,72 +1,72 @@ #!/usr/bin/env python3 # Copyright (C) 2015-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -from setuptools import setup, find_packages - -from os import path from io import open +from os import path + +from setuptools import find_packages, setup here = path.abspath(path.dirname(__file__)) # Get the long description from the README file with open(path.join(here, "README.md"), encoding="utf-8") as f: long_description = f.read() def parse_requirements(name=None): if name: reqf = "requirements-%s.txt" % name else: reqf = "requirements.txt" requirements = [] if not path.exists(reqf): return requirements with open(reqf) as f: for line in f.readlines(): line = line.strip() if not line or line.startswith("#"): continue requirements.append(line) return requirements setup( name="swh.scheduler", description="Software Heritage Scheduler", long_description=long_description, long_description_content_type="text/markdown", python_requires=">=3.7", author="Software Heritage developers", author_email="swh-devel@inria.fr", url="https://forge.softwareheritage.org/diffusion/DSCH/", packages=find_packages(), setup_requires=["setuptools-scm"], use_scm_version=True, install_requires=parse_requirements() + parse_requirements("swh"), extras_require={"testing": parse_requirements("test")}, include_package_data=True, entry_points=""" [console_scripts] swh-scheduler=swh.scheduler.cli:main [swh.cli.subcommands] scheduler=swh.scheduler.cli:cli """, classifiers=[ "Programming Language :: Python :: 3", "Intended Audience :: Developers", "License :: OSI Approved :: GNU General Public License v3 (GPLv3)", "Operating System :: OS Independent", "Development Status :: 5 - Production/Stable", ], project_urls={ "Bug Reports": "https://forge.softwareheritage.org/maniphest", "Funding": "https://www.softwareheritage.org/donate", "Source": "https://forge.softwareheritage.org/source/swh-scheduler", "Documentation": "https://docs.softwareheritage.org/devel/swh-scheduler/", }, ) diff --git a/swh/scheduler/__init__.py b/swh/scheduler/__init__.py index a0b72f2..2d3892f 100644 --- a/swh/scheduler/__init__.py +++ b/swh/scheduler/__init__.py @@ -1,68 +1,67 @@ # Copyright (C) 2018 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from typing import Any, Dict - # Percentage of tasks with priority to schedule PRIORITY_SLOT = 0.6 DEFAULT_CONFIG = { "scheduler": ( "dict", {"cls": "local", "args": {"db": "dbname=softwareheritage-scheduler-dev",},}, ) } # current configuration. To be set by the config loading mechanism CONFIG = {} # type: Dict[str, Any] def compute_nb_tasks_from(num_tasks): """Compute and returns the tuple, number of tasks without priority, number of tasks with priority. Args: num_tasks (int): Returns: tuple number of tasks without priority (int), number of tasks with priority (int) """ if not num_tasks: return None, None return (int((1 - PRIORITY_SLOT) * num_tasks), int(PRIORITY_SLOT * num_tasks)) def get_scheduler(cls, args={}): """ Get a scheduler object of class `scheduler_class` with arguments `scheduler_args`. Args: scheduler (dict): dictionary with keys: cls (str): scheduler's class, either 'local' or 'remote' args (dict): dictionary with keys, default to empty. Returns: an instance of swh.scheduler, either local or remote: local: swh.scheduler.backend.SchedulerBackend remote: swh.scheduler.api.client.RemoteScheduler Raises: ValueError if passed an unknown storage class. """ if cls == "remote": from .api.client import RemoteScheduler as SchedulerBackend elif cls == "local": from .backend import SchedulerBackend else: raise ValueError("Unknown swh.scheduler class `%s`" % cls) return SchedulerBackend(**args) diff --git a/swh/scheduler/api/client.py b/swh/scheduler/api/client.py index 15d8e54..8a36d05 100644 --- a/swh/scheduler/api/client.py +++ b/swh/scheduler/api/client.py @@ -1,24 +1,24 @@ # Copyright (C) 2018-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from swh.core.api import RPCClient -from .serializers import ENCODERS, DECODERS from .. import exc from ..interface import SchedulerInterface +from .serializers import DECODERS, ENCODERS class RemoteScheduler(RPCClient): """Proxy to a remote scheduler API """ backend_class = SchedulerInterface reraise_exceptions = [getattr(exc, a) for a in exc.__all__] extra_type_decoders = DECODERS extra_type_encoders = ENCODERS diff --git a/swh/scheduler/api/server.py b/swh/scheduler/api/server.py index 7a9d8f8..3a6c173 100644 --- a/swh/scheduler/api/server.py +++ b/swh/scheduler/api/server.py @@ -1,155 +1,154 @@ # Copyright (C) 2018-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import logging import os from swh.core import config from swh.core.api import JSONFormatter, MsgpackFormatter, RPCServerApp from swh.core.api import encode_data_server as encode_data from swh.core.api import error_handler, negotiate - from swh.scheduler import get_scheduler from swh.scheduler.exc import SchedulerException from swh.scheduler.interface import SchedulerInterface -from .serializers import ENCODERS, DECODERS +from .serializers import DECODERS, ENCODERS scheduler = None def get_global_scheduler(): global scheduler if not scheduler: scheduler = get_scheduler(**app.config["scheduler"]) return scheduler class SchedulerServerApp(RPCServerApp): extra_type_decoders = DECODERS extra_type_encoders = ENCODERS app = SchedulerServerApp( __name__, backend_class=SchedulerInterface, backend_factory=get_global_scheduler ) @app.errorhandler(SchedulerException) def argument_error_handler(exception): return error_handler(exception, encode_data, status_code=400) @app.errorhandler(Exception) def my_error_handler(exception): return error_handler(exception, encode_data) def has_no_empty_params(rule): return len(rule.defaults or ()) >= len(rule.arguments or ()) @app.route("/") def index(): return """ Software Heritage scheduler RPC server

You have reached the Software Heritage scheduler RPC server.
See its documentation and API for more information

""" @app.route("/site-map") @negotiate(MsgpackFormatter) @negotiate(JSONFormatter) def site_map(): links = [] for rule in app.url_map.iter_rules(): if has_no_empty_params(rule) and hasattr(SchedulerInterface, rule.endpoint): links.append( dict( rule=rule.rule, description=getattr(SchedulerInterface, rule.endpoint).__doc__, ) ) # links is now a list of url, endpoint tuples return links def load_and_check_config(config_file, type="local"): """Check the minimal configuration is set to run the api or raise an error explanation. Args: config_file (str): Path to the configuration file to load type (str): configuration type. For 'local' type, more checks are done. Raises: Error if the setup is not as expected Returns: configuration as a dict """ if not config_file: raise EnvironmentError("Configuration file must be defined") if not os.path.exists(config_file): raise FileNotFoundError("Configuration file %s does not exist" % (config_file,)) cfg = config.read(config_file) vcfg = cfg.get("scheduler") if not vcfg: raise KeyError("Missing '%scheduler' configuration") if type == "local": cls = vcfg.get("cls") if cls != "local": raise ValueError( "The scheduler backend can only be started with a 'local' " "configuration" ) args = vcfg.get("args") if not args: raise KeyError("Invalid configuration; missing 'args' config entry") db = args.get("db") if not db: raise KeyError("Invalid configuration; missing 'db' config entry") return cfg api_cfg = None def make_app_from_configfile(): """Run the WSGI app from the webserver, loading the configuration from a configuration file. SWH_CONFIG_FILENAME environment variable defines the configuration path to load. """ global api_cfg if not api_cfg: config_file = os.environ.get("SWH_CONFIG_FILENAME") api_cfg = load_and_check_config(config_file) app.config.update(api_cfg) handler = logging.StreamHandler() app.logger.addHandler(handler) return app if __name__ == "__main__": print('Please use the "swh-scheduler api-server" command') diff --git a/swh/scheduler/backend.py b/swh/scheduler/backend.py index 84cda7c..5160269 100644 --- a/swh/scheduler/backend.py +++ b/swh/scheduler/backend.py @@ -1,734 +1,733 @@ # Copyright (C) 2015-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import json import logging +from typing import Any, Dict, Iterable, List, Optional, Tuple, Union from uuid import UUID from arrow import Arrow, utcnow import attr -import psycopg2.pool -import psycopg2.extras - -from typing import Any, Dict, Iterable, List, Optional, Tuple, Union from psycopg2.extensions import AsIs +import psycopg2.extras +import psycopg2.pool from swh.core.db import BaseDb from swh.core.db.common import db_transaction from .exc import StaleData from .model import ( - Lister, ListedOrigin, ListedOriginPageToken, + Lister, PaginatedListedOriginList, ) logger = logging.getLogger(__name__) def adapt_arrow(arrow): return AsIs("'%s'::timestamptz" % arrow.isoformat()) psycopg2.extensions.register_adapter(dict, psycopg2.extras.Json) psycopg2.extensions.register_adapter(Arrow, adapt_arrow) psycopg2.extras.register_uuid() def format_query(query, keys): """Format a query with the given keys""" query_keys = ", ".join(keys) placeholders = ", ".join(["%s"] * len(keys)) return query.format(keys=query_keys, placeholders=placeholders) class SchedulerBackend: """Backend for the Software Heritage scheduling database. """ def __init__(self, db, min_pool_conns=1, max_pool_conns=10): """ Args: db_conn: either a libpq connection string, or a psycopg2 connection """ if isinstance(db, psycopg2.extensions.connection): self._pool = None self._db = BaseDb(db) else: self._pool = psycopg2.pool.ThreadedConnectionPool( min_pool_conns, max_pool_conns, db, cursor_factory=psycopg2.extras.RealDictCursor, ) self._db = None def get_db(self): if self._db: return self._db return BaseDb.from_pool(self._pool) def put_db(self, db): if db is not self._db: db.put_conn() task_type_keys = [ "type", "description", "backend_name", "default_interval", "min_interval", "max_interval", "backoff_factor", "max_queue_length", "num_retries", "retry_delay", ] @db_transaction() def create_task_type(self, task_type, db=None, cur=None): """Create a new task type ready for scheduling. Args: task_type (dict): a dictionary with the following keys: - type (str): an identifier for the task type - description (str): a human-readable description of what the task does - backend_name (str): the name of the task in the job-scheduling backend - default_interval (datetime.timedelta): the default interval between two task runs - min_interval (datetime.timedelta): the minimum interval between two task runs - max_interval (datetime.timedelta): the maximum interval between two task runs - backoff_factor (float): the factor by which the interval changes at each run - max_queue_length (int): the maximum length of the task queue for this task type """ keys = [key for key in self.task_type_keys if key in task_type] query = format_query( """insert into task_type ({keys}) values ({placeholders}) on conflict do nothing""", keys, ) cur.execute(query, [task_type[key] for key in keys]) @db_transaction() def get_task_type(self, task_type_name, db=None, cur=None): """Retrieve the task type with id task_type_name""" query = format_query( "select {keys} from task_type where type=%s", self.task_type_keys, ) cur.execute(query, (task_type_name,)) return cur.fetchone() @db_transaction() def get_task_types(self, db=None, cur=None): """Retrieve all registered task types""" query = format_query("select {keys} from task_type", self.task_type_keys,) cur.execute(query) return cur.fetchall() @db_transaction() def get_or_create_lister( self, name: str, instance_name: Optional[str] = None, db=None, cur=None ) -> Lister: """Retrieve information about the given instance of the lister from the database, or create the entry if it did not exist. """ if instance_name is None: instance_name = "" select_cols = ", ".join(Lister.select_columns()) insert_cols, insert_meta = ( ", ".join(tup) for tup in Lister.insert_columns_and_metavars() ) query = f""" with added as ( insert into listers ({insert_cols}) values ({insert_meta}) on conflict do nothing returning {select_cols} ) select {select_cols} from added union all select {select_cols} from listers where (name, instance_name) = (%(name)s, %(instance_name)s); """ cur.execute(query, attr.asdict(Lister(name=name, instance_name=instance_name))) return Lister(**cur.fetchone()) @db_transaction() def update_lister(self, lister: Lister, db=None, cur=None) -> Lister: """Update the state for the given lister instance in the database. Returns: a new Lister object, with all fields updated from the database Raises: StaleData if the `updated` timestamp for the lister instance in database doesn't match the one passed by the user. """ select_cols = ", ".join(Lister.select_columns()) set_vars = ", ".join( f"{col} = {meta}" for col, meta in zip(*Lister.insert_columns_and_metavars()) ) query = f"""update listers set {set_vars} where id=%(id)s and updated=%(updated)s returning {select_cols}""" cur.execute(query, attr.asdict(lister)) updated = cur.fetchone() if not updated: raise StaleData("Stale data; Lister state not updated") return Lister(**updated) @db_transaction() def record_listed_origins( self, listed_origins: Iterable[ListedOrigin], db=None, cur=None ) -> List[ListedOrigin]: """Record a set of origins that a lister has listed. This performs an "upsert": origins with the same (lister_id, url, visit_type) values are updated with new values for extra_loader_arguments, last_update and last_seen. """ pk_cols = ListedOrigin.primary_key_columns() select_cols = ListedOrigin.select_columns() insert_cols, insert_meta = ListedOrigin.insert_columns_and_metavars() upsert_cols = [col for col in insert_cols if col not in pk_cols] upsert_set = ", ".join(f"{col} = EXCLUDED.{col}" for col in upsert_cols) query = f"""INSERT into listed_origins ({", ".join(insert_cols)}) VALUES %s ON CONFLICT ({", ".join(pk_cols)}) DO UPDATE SET {upsert_set} RETURNING {", ".join(select_cols)} """ ret = psycopg2.extras.execute_values( cur=cur, sql=query, argslist=(attr.asdict(origin) for origin in listed_origins), template=f"({', '.join(insert_meta)})", page_size=1000, fetch=True, ) return [ListedOrigin(**d) for d in ret] @db_transaction() def get_listed_origins( self, lister_id: Optional[UUID] = None, url: Optional[str] = None, limit: int = 1000, page_token: Optional[ListedOriginPageToken] = None, db=None, cur=None, ) -> PaginatedListedOriginList: """Get information on the listed origins matching either the `url` or `lister_id`, or both arguments. """ query_filters: List[str] = [] query_params: List[Union[int, str, UUID, Tuple[UUID, str]]] = [] if lister_id: query_filters.append("lister_id = %s") query_params.append(lister_id) if url is not None: query_filters.append("url = %s") query_params.append(url) if page_token is not None: query_filters.append("(lister_id, url) > %s") # the typeshed annotation for tuple() is too strict. query_params.append(tuple(page_token)) # type: ignore query_params.append(limit) select_cols = ", ".join(ListedOrigin.select_columns()) if query_filters: where_clause = "where %s" % (" and ".join(query_filters)) else: where_clause = "" query = f"""SELECT {select_cols} from listed_origins {where_clause} ORDER BY lister_id, url LIMIT %s""" cur.execute(query, tuple(query_params)) origins = [ListedOrigin(**d) for d in cur] if len(origins) == limit: page_token = (origins[-1].lister_id, origins[-1].url) else: page_token = None return PaginatedListedOriginList(origins, page_token) task_create_keys = [ "type", "arguments", "next_run", "policy", "status", "retries_left", "priority", ] task_keys = task_create_keys + ["id", "current_interval"] @db_transaction() def create_tasks(self, tasks, policy="recurring", db=None, cur=None): """Create new tasks. Args: tasks (list): each task is a dictionary with the following keys: - type (str): the task type - arguments (dict): the arguments for the task runner, keys: - args (list of str): arguments - kwargs (dict str -> str): keyword arguments - next_run (datetime.datetime): the next scheduled run for the task Returns: a list of created tasks. """ cur.execute("select swh_scheduler_mktemp_task()") db.copy_to( tasks, "tmp_task", self.task_create_keys, default_values={"policy": policy, "status": "next_run_not_scheduled"}, cur=cur, ) query = format_query( "select {keys} from swh_scheduler_create_tasks_from_temp()", self.task_keys, ) cur.execute(query) return cur.fetchall() @db_transaction() def set_status_tasks( self, task_ids, status="disabled", next_run=None, db=None, cur=None ): """Set the tasks' status whose ids are listed. If given, also set the next_run date. """ if not task_ids: return query = ["UPDATE task SET status = %s"] args = [status] if next_run: query.append(", next_run = %s") args.append(next_run) query.append(" WHERE id IN %s") args.append(tuple(task_ids)) cur.execute("".join(query), args) @db_transaction() def disable_tasks(self, task_ids, db=None, cur=None): """Disable the tasks whose ids are listed.""" return self.set_status_tasks(task_ids, db=db, cur=cur) @db_transaction() def search_tasks( self, task_id=None, task_type=None, status=None, priority=None, policy=None, before=None, after=None, limit=None, db=None, cur=None, ): """Search tasks from selected criterions""" where = [] args = [] if task_id: if isinstance(task_id, (str, int)): where.append("id = %s") else: where.append("id in %s") task_id = tuple(task_id) args.append(task_id) if task_type: if isinstance(task_type, str): where.append("type = %s") else: where.append("type in %s") task_type = tuple(task_type) args.append(task_type) if status: if isinstance(status, str): where.append("status = %s") else: where.append("status in %s") status = tuple(status) args.append(status) if priority: if isinstance(priority, str): where.append("priority = %s") else: priority = tuple(priority) where.append("priority in %s") args.append(priority) if policy: where.append("policy = %s") args.append(policy) if before: where.append("next_run <= %s") args.append(before) if after: where.append("next_run >= %s") args.append(after) query = "select * from task" if where: query += " where " + " and ".join(where) if limit: query += " limit %s :: bigint" args.append(limit) cur.execute(query, args) return cur.fetchall() @db_transaction() def get_tasks(self, task_ids, db=None, cur=None): """Retrieve the info of tasks whose ids are listed.""" query = format_query("select {keys} from task where id in %s", self.task_keys) cur.execute(query, (tuple(task_ids),)) return cur.fetchall() @db_transaction() def peek_ready_tasks( self, task_type, timestamp=None, num_tasks=None, num_tasks_priority=None, db=None, cur=None, ): """Fetch the list of ready tasks Args: task_type (str): filtering task per their type timestamp (datetime.datetime): peek tasks that need to be executed before that timestamp num_tasks (int): only peek at num_tasks tasks (with no priority) num_tasks_priority (int): only peek at num_tasks_priority tasks (with priority) Returns: a list of tasks """ if timestamp is None: timestamp = utcnow() cur.execute( """select * from swh_scheduler_peek_ready_tasks( %s, %s, %s :: bigint, %s :: bigint)""", (task_type, timestamp, num_tasks, num_tasks_priority), ) logger.debug("PEEK %s => %s" % (task_type, cur.rowcount)) return cur.fetchall() @db_transaction() def grab_ready_tasks( self, task_type, timestamp=None, num_tasks=None, num_tasks_priority=None, db=None, cur=None, ): """Fetch the list of ready tasks, and mark them as scheduled Args: task_type (str): filtering task per their type timestamp (datetime.datetime): grab tasks that need to be executed before that timestamp num_tasks (int): only grab num_tasks tasks (with no priority) num_tasks_priority (int): only grab oneshot num_tasks tasks (with priorities) Returns: a list of tasks """ if timestamp is None: timestamp = utcnow() cur.execute( """select * from swh_scheduler_grab_ready_tasks( %s, %s, %s :: bigint, %s :: bigint)""", (task_type, timestamp, num_tasks, num_tasks_priority), ) logger.debug("GRAB %s => %s" % (task_type, cur.rowcount)) return cur.fetchall() task_run_create_keys = ["task", "backend_id", "scheduled", "metadata"] @db_transaction() def schedule_task_run( self, task_id, backend_id, metadata=None, timestamp=None, db=None, cur=None ): """Mark a given task as scheduled, adding a task_run entry in the database. Args: task_id (int): the identifier for the task being scheduled backend_id (str): the identifier of the job in the backend metadata (dict): metadata to add to the task_run entry timestamp (datetime.datetime): the instant the event occurred Returns: a fresh task_run entry """ if metadata is None: metadata = {} if timestamp is None: timestamp = utcnow() cur.execute( "select * from swh_scheduler_schedule_task_run(%s, %s, %s, %s)", (task_id, backend_id, metadata, timestamp), ) return cur.fetchone() @db_transaction() def mass_schedule_task_runs(self, task_runs, db=None, cur=None): """Schedule a bunch of task runs. Args: task_runs (list): a list of dicts with keys: - task (int): the identifier for the task being scheduled - backend_id (str): the identifier of the job in the backend - metadata (dict): metadata to add to the task_run entry - scheduled (datetime.datetime): the instant the event occurred Returns: None """ cur.execute("select swh_scheduler_mktemp_task_run()") db.copy_to(task_runs, "tmp_task_run", self.task_run_create_keys, cur=cur) cur.execute("select swh_scheduler_schedule_task_run_from_temp()") @db_transaction() def start_task_run( self, backend_id, metadata=None, timestamp=None, db=None, cur=None ): """Mark a given task as started, updating the corresponding task_run entry in the database. Args: backend_id (str): the identifier of the job in the backend metadata (dict): metadata to add to the task_run entry timestamp (datetime.datetime): the instant the event occurred Returns: the updated task_run entry """ if metadata is None: metadata = {} if timestamp is None: timestamp = utcnow() cur.execute( "select * from swh_scheduler_start_task_run(%s, %s, %s)", (backend_id, metadata, timestamp), ) return cur.fetchone() @db_transaction() def end_task_run( self, backend_id, status, metadata=None, timestamp=None, result=None, db=None, cur=None, ): """Mark a given task as ended, updating the corresponding task_run entry in the database. Args: backend_id (str): the identifier of the job in the backend status (str): how the task ended; one of: 'eventful', 'uneventful', 'failed' metadata (dict): metadata to add to the task_run entry timestamp (datetime.datetime): the instant the event occurred Returns: the updated task_run entry """ if metadata is None: metadata = {} if timestamp is None: timestamp = utcnow() cur.execute( "select * from swh_scheduler_end_task_run(%s, %s, %s, %s)", (backend_id, status, metadata, timestamp), ) return cur.fetchone() @db_transaction() def filter_task_to_archive( self, after_ts: str, before_ts: str, limit: int = 10, page_token: Optional[str] = None, db=None, cur=None, ) -> Dict[str, Any]: """Compute the tasks to archive within the datetime interval [after_ts, before_ts[. The method returns a paginated result. Returns: dict with the following keys: - **next_page_token**: opaque token to be used as `page_token` to retrieve the next page of result. If absent, there is no more pages to gather. - **tasks**: list of task dictionaries with the following keys: **id** (str): origin task id **started** (Optional[datetime]): started date **scheduled** (datetime): scheduled date **arguments** (json dict): task's arguments ... """ assert not page_token or isinstance(page_token, str) last_id = -1 if page_token is None else int(page_token) tasks = [] cur.execute( "select * from swh_scheduler_task_to_archive(%s, %s, %s, %s)", (after_ts, before_ts, last_id, limit + 1), ) for row in cur: task = dict(row) # nested type index does not accept bare values # transform it as a dict to comply with this task["arguments"]["args"] = { i: v for i, v in enumerate(task["arguments"]["args"]) } kwargs = task["arguments"]["kwargs"] task["arguments"]["kwargs"] = json.dumps(kwargs) tasks.append(task) if len(tasks) >= limit + 1: # remains data, add pagination information result = { "tasks": tasks[:limit], "next_page_token": str(tasks[-1]["task_id"]), } else: result = {"tasks": tasks} return result @db_transaction() def delete_archived_tasks(self, task_ids, db=None, cur=None): """Delete archived tasks as much as possible. Only the task_ids whose complete associated task_run have been cleaned up will be. """ _task_ids = _task_run_ids = [] for task_id in task_ids: _task_ids.append(task_id["task_id"]) _task_run_ids.append(task_id["task_run_id"]) cur.execute( "select * from swh_scheduler_delete_archived_tasks(%s, %s)", (_task_ids, _task_run_ids), ) task_run_keys = [ "id", "task", "backend_id", "scheduled", "started", "ended", "metadata", "status", ] @db_transaction() def get_task_runs(self, task_ids, limit=None, db=None, cur=None): """Search task run for a task id""" where = [] args = [] if task_ids: if isinstance(task_ids, (str, int)): where.append("task = %s") else: where.append("task in %s") task_ids = tuple(task_ids) args.append(task_ids) else: return () query = "select * from task_run where " + " and ".join(where) if limit: query += " limit %s :: bigint" args.append(limit) cur.execute(query, args) return cur.fetchall() @db_transaction() def get_priority_ratios(self, db=None, cur=None): cur.execute("select id, ratio from priority_ratio") return {row["id"]: row["ratio"] for row in cur.fetchall()} diff --git a/swh/scheduler/backend_es.py b/swh/scheduler/backend_es.py index ad7b6a3..e1b949c 100644 --- a/swh/scheduler/backend_es.py +++ b/swh/scheduler/backend_es.py @@ -1,268 +1,269 @@ # Copyright (C) 2018-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """Elastic Search backend """ +from copy import deepcopy import datetime # noqa import logging - -from copy import deepcopy from typing import Any, Dict from elasticsearch import helpers from swh.core import utils logger = logging.getLogger(__name__) DEFAULT_CONFIG = { "elasticsearch": { "cls": "local", "args": { "index_name_prefix": "swh-tasks", "storage_nodes": ["localhost:9200"], "client_options": { "sniff_on_start": False, "sniff_on_connection_fail": True, "http_compress": False, "sniffer_timeout": 60, }, }, } } def get_elasticsearch(cls: str, args: Dict[str, Any] = {}): """Instantiate an elastic search instance """ if cls == "local": from elasticsearch import Elasticsearch elif cls == "memory": - from .elasticsearch_memory import MemoryElasticsearch as Elasticsearch # type: ignore # noqa + from .elasticsearch_memory import ( # type: ignore # noqa + MemoryElasticsearch as Elasticsearch, + ) else: raise ValueError("Unknown elasticsearch class `%s`" % cls) return Elasticsearch(**args) class ElasticSearchBackend: """ElasticSearch backend to index tasks This uses an elasticsearch client to actually discuss with the elasticsearch instance. """ def __init__(self, **config): self.config = deepcopy(DEFAULT_CONFIG) self.config.update(config) es_conf = self.config["elasticsearch"] args = deepcopy(es_conf["args"]) self.index_name_prefix = args.pop("index_name_prefix") self.storage = get_elasticsearch( cls=es_conf["cls"], args={ "hosts": args.get("storage_nodes", []), **args.get("client_options", {}), }, ) # document's index type (cf. /data/elastic-template.json) self.doc_type = "task" def initialize(self): self.storage.indices.put_mapping( index=f"{self.index_name_prefix}-*", doc_type=self.doc_type, # to allow type definition below include_type_name=True, # to allow install mapping even if no index yet allow_no_indices=True, body={ "properties": { "task_id": {"type": "double"}, "task_policy": {"type": "text"}, "task_status": {"type": "text"}, "task_run_id": {"type": "double"}, "arguments": { "type": "object", "properties": { "args": {"type": "nested", "dynamic": False}, "kwargs": {"type": "text"}, }, }, "type": {"type": "text"}, "backend_id": {"type": "text"}, "metadata": {"type": "object", "enabled": False}, "scheduled": { "type": "date", "format": "yyyy-MM-dd HH:mm:ss||yyyy-MM-dd||strict_date_optional_time||epoch_millis", # noqa }, "started": { "type": "date", "format": "yyyy-MM-dd HH:mm:ss||yyyy-MM-dd||strict_date_optional_time||epoch_millis", # noqa }, "ended": { "type": "date", "format": "yyyy-MM-dd HH:mm:ss||yyyy-MM-dd||strict_date_optional_time||epoch_millis", # noqa }, } }, ) self.storage.indices.put_settings( index=f"{self.index_name_prefix}-*", allow_no_indices=True, body={ "index": { "codec": "best_compression", "refresh_interval": "1s", "number_of_shards": 1, } }, ) def create(self, index_name) -> None: """Create and initialize index_name with mapping for all indices matching `swh-tasks-` pattern """ assert index_name.startswith(self.index_name_prefix) self.storage.indices.create(index_name) def compute_index_name(self, year, month): """Given a year, month, compute the index's name. """ return "%s-%s-%s" % (self.index_name_prefix, year, "%02d" % month) def mget(self, index_name, doc_ids, chunk_size=500, source=True): """Retrieve document's full content according to their ids as per source's setup. The `source` allows to retrieve only what's interesting, e.g: - source=True ; gives back the original indexed data - source=False ; returns without the original _source field - source=['task_id'] ; returns only task_id in the _source field Args: index_name (str): Name of the concerned index. doc_ids (generator): Generator of ids to retrieve chunk_size (int): Number of documents chunk to send for retrieval source (bool/[str]): Source of information to return Yields: document indexed as per source's setup """ if isinstance(source, list): source = {"_source": ",".join(source)} else: source = {"_source": str(source).lower()} for ids in utils.grouper(doc_ids, n=1000): res = self.storage.mget( body={"ids": list(ids)}, index=index_name, doc_type=self.doc_type, params=source, ) if not res: logger.error("Error during retrieval of data, skipping!") continue for doc in res["docs"]: found = doc.get("found") if not found: msg = "Doc id %s not found, not indexed yet" % doc["_id"] logger.warning(msg) continue yield doc["_source"] def _streaming_bulk(self, index_name, doc_stream, chunk_size=500): """Bulk index data and returns the successful indexed data's identifier. Args: index_name (str): Name of the concerned index. doc_stream (generator): Generator of documents to index chunk_size (int): Number of documents chunk to send for indexation Yields: document id indexed """ actions = ( { "_index": index_name, "_op_type": "index", "_type": self.doc_type, "_source": data, } for data in doc_stream ) for ok, result in helpers.streaming_bulk( client=self.storage, actions=actions, chunk_size=chunk_size, raise_on_error=False, raise_on_exception=False, ): if not ok: logger.error("Error during %s indexation. Skipping.", result) continue yield result["index"]["_id"] def is_index_opened(self, index_name: str) -> bool: """Determine if an index is opened or not """ try: self.storage.indices.stats(index_name) return True except Exception: # fails when indice is closed (no other api call found) return False def streaming_bulk(self, index_name, doc_stream, chunk_size=500, source=True): """Bulk index data and returns the successful indexed data as per source's setup. the `source` permits to retrieve only what's of interest to us, e.g: - source=True ; gives back the original indexed data - source=False ; returns without the original _source field - source=['task_id'] ; returns only task_id in the _source field Note that: - if the index is closed, it will be opened - if the index does not exist, it will be created and opened This keeps the index opened for performance reasons. Args: index_name (str): Name of the concerned index. doc_stream (generator): Document generator to index chunk_size (int): Number of documents chunk to send source (bool, [str]): the information to return """ # index must exist if not self.storage.indices.exists(index_name): self.create(index_name) # index must be opened if not self.is_index_opened(index_name): self.storage.indices.open(index_name) indexed_ids = self._streaming_bulk( index_name, doc_stream, chunk_size=chunk_size ) yield from self.mget( index_name, indexed_ids, chunk_size=chunk_size, source=source ) diff --git a/swh/scheduler/celery_backend/config.py b/swh/scheduler/celery_backend/config.py index 58a57d8..de3c2ee 100644 --- a/swh/scheduler/celery_backend/config.py +++ b/swh/scheduler/celery_backend/config.py @@ -1,345 +1,342 @@ # Copyright (C) 2015-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import functools import logging import os -import pkg_resources import traceback from typing import Any, Dict import urllib.parse from celery import Celery -from celery.signals import setup_logging, celeryd_after_setup, worker_init +from celery.signals import celeryd_after_setup, setup_logging, worker_init from celery.utils.log import ColorFormatter from celery.worker.control import Panel - from kombu import Exchange, Queue from kombu.five import monotonic as _monotonic - +import pkg_resources import requests -from swh.scheduler import CONFIG as SWH_CONFIG - from swh.core.config import load_named_config, merge_configs from swh.core.sentry import init_sentry +from swh.scheduler import CONFIG as SWH_CONFIG try: from swh.core.logger import JournalHandler except ImportError: JournalHandler = None # type: ignore DEFAULT_CONFIG_NAME = "worker" CONFIG_NAME_ENVVAR = "SWH_WORKER_INSTANCE" CONFIG_NAME_TEMPLATE = "worker/%s" DEFAULT_CONFIG = { "task_broker": ("str", "amqp://guest@localhost//"), "task_modules": ("list[str]", []), "task_queues": ("list[str]", []), "task_soft_time_limit": ("int", 0), } logger = logging.getLogger(__name__) # Celery eats tracebacks in signal callbacks, this decorator catches # and prints them. # Also tries to notify Sentry if possible. def _print_errors(f): @functools.wraps(f) def newf(*args, **kwargs): try: return f(*args, **kwargs) except Exception as exc: traceback.print_exc() try: import sentry_sdk sentry_sdk.capture_exception(exc) except Exception: traceback.print_exc() return newf @setup_logging.connect @_print_errors def setup_log_handler( loglevel=None, logfile=None, format=None, colorize=None, log_console=None, log_journal=None, **kwargs, ): """Setup logging according to Software Heritage preferences. We use the command-line loglevel for tasks only, as we never really care about the debug messages from celery. """ if loglevel is None: loglevel = logging.DEBUG if isinstance(loglevel, str): loglevel = logging._nameToLevel[loglevel] formatter = logging.Formatter(format) root_logger = logging.getLogger("") root_logger.setLevel(logging.INFO) log_target = os.environ.get("SWH_LOG_TARGET", "console") if log_target == "console": log_console = True elif log_target == "journal": log_journal = True # this looks for log levels *higher* than DEBUG if loglevel <= logging.DEBUG and log_console is None: log_console = True if log_console: color_formatter = ColorFormatter(format) if colorize else formatter console = logging.StreamHandler() console.setLevel(logging.DEBUG) console.setFormatter(color_formatter) root_logger.addHandler(console) if log_journal: if not JournalHandler: root_logger.warning( "JournalHandler is not available, skipping. " "Please install swh-core[logging]." ) else: systemd_journal = JournalHandler() systemd_journal.setLevel(logging.DEBUG) systemd_journal.setFormatter(formatter) root_logger.addHandler(systemd_journal) logging.getLogger("celery").setLevel(logging.INFO) # Silence amqp heartbeat_tick messages logger = logging.getLogger("amqp") logger.addFilter(lambda record: not record.msg.startswith("heartbeat_tick")) logger.setLevel(loglevel) # Silence useless "Starting new HTTP connection" messages logging.getLogger("urllib3").setLevel(logging.WARNING) # Completely disable azure logspam azure_logger = logging.getLogger("azure.core.pipeline.policies.http_logging_policy") azure_logger.setLevel(logging.WARNING) logging.getLogger("swh").setLevel(loglevel) # get_task_logger makes the swh tasks loggers children of celery.task logging.getLogger("celery.task").setLevel(loglevel) return loglevel @celeryd_after_setup.connect @_print_errors def setup_queues_and_tasks(sender, instance, **kwargs): """Signal called on worker start. This automatically registers swh.scheduler.task.Task subclasses as available celery tasks. This also subscribes the worker to the "implicit" per-task queues defined for these task classes. """ logger.info("Setup Queues & Tasks for %s", sender) instance.app.conf["worker_name"] = sender @worker_init.connect @_print_errors def on_worker_init(*args, **kwargs): try: from sentry_sdk.integrations.celery import CeleryIntegration except ImportError: integrations = [] else: integrations = [CeleryIntegration()] sentry_dsn = None # will be set in `init_sentry` function init_sentry(sentry_dsn, integrations=integrations) @Panel.register def monotonic(state): """Get the current value for the monotonic clock""" return {"monotonic": _monotonic()} def route_for_task(name, args, kwargs, options, task=None, **kw): """Route tasks according to the task_queue attribute in the task class""" if name is not None and name.startswith("swh."): return {"queue": name} def get_queue_stats(app, queue_name): """Get the statistics regarding a queue on the broker. Arguments: queue_name: name of the queue to check Returns a dictionary raw from the RabbitMQ management API; or `None` if the current configuration does not use RabbitMQ. Interesting keys: - Consumers (number of consumers for the queue) - messages (number of messages in queue) - messages_unacknowledged (number of messages currently being processed) Documentation: https://www.rabbitmq.com/management.html#http-api """ conn_info = app.connection().info() if conn_info["transport"] == "memory": # We're running in a test environment, without RabbitMQ. return None url = "http://{hostname}:{port}/api/queues/{vhost}/{queue}".format( hostname=conn_info["hostname"], port=conn_info["port"] + 10000, vhost=urllib.parse.quote(conn_info["virtual_host"], safe=""), queue=urllib.parse.quote(queue_name, safe=""), ) credentials = (conn_info["userid"], conn_info["password"]) r = requests.get(url, auth=credentials) if r.status_code == 404: return {} if r.status_code != 200: raise ValueError( "Got error %s when reading queue stats: %s" % (r.status_code, r.json()) ) return r.json() def get_queue_length(app, queue_name): """Shortcut to get a queue's length""" stats = get_queue_stats(app, queue_name) if stats: return stats.get("messages") def register_task_class(app, name, cls): """Register a class-based task under the given name""" if name in app.tasks: return task_instance = cls() task_instance.name = name app.register_task(task_instance) INSTANCE_NAME = os.environ.get(CONFIG_NAME_ENVVAR) CONFIG_NAME = os.environ.get("SWH_CONFIG_FILENAME") CONFIG = {} # type: Dict[str, Any] if CONFIG_NAME: # load the celery config from the main config file given as # SWH_CONFIG_FILENAME environment variable. # This is expected to have a [celery] section in which we have the # celery specific configuration. SWH_CONFIG.clear() SWH_CONFIG.update(load_named_config(CONFIG_NAME)) CONFIG = SWH_CONFIG.get("celery", {}) if not CONFIG: # otherwise, back to compat config loading mechanism if INSTANCE_NAME: CONFIG_NAME = CONFIG_NAME_TEMPLATE % INSTANCE_NAME else: CONFIG_NAME = DEFAULT_CONFIG_NAME # Load the Celery config CONFIG = load_named_config(CONFIG_NAME, DEFAULT_CONFIG) CONFIG.setdefault("task_modules", []) # load tasks modules declared as plugin entry points for entrypoint in pkg_resources.iter_entry_points("swh.workers"): worker_registrer_fn = entrypoint.load() # The registry function is expected to return a dict which the 'tasks' key # is a string (or a list of strings) with the name of the python module in # which celery tasks are defined. task_modules = worker_registrer_fn().get("task_modules", []) CONFIG["task_modules"].extend(task_modules) # Celery Queues CELERY_QUEUES = [Queue("celery", Exchange("celery"), routing_key="celery")] CELERY_DEFAULT_CONFIG = dict( # Timezone configuration: all in UTC enable_utc=True, timezone="UTC", # Imported modules imports=CONFIG.get("task_modules", []), # Time (in seconds, or a timedelta object) for when after stored task # tombstones will be deleted. None means to never expire results. result_expires=None, # A string identifying the default serialization method to use. Can # be json (default), pickle, yaml, msgpack, or any custom # serialization methods that have been registered with task_serializer="msgpack", # Result serialization format result_serializer="msgpack", # Late ack means the task messages will be acknowledged after the task has # been executed, not just before, which is the default behavior. task_acks_late=True, # A string identifying the default serialization method to use. # Can be pickle (default), json, yaml, msgpack or any custom serialization # methods that have been registered with kombu.serialization.registry accept_content=["msgpack", "json"], # If True the task will report its status as “started” # when the task is executed by a worker. task_track_started=True, # Default compression used for task messages. Can be gzip, bzip2 # (if available), or any custom compression schemes registered # in the Kombu compression registry. # result_compression='bzip2', # task_compression='bzip2', # Disable all rate limits, even if tasks has explicit rate limits set. # (Disabling rate limits altogether is recommended if you don’t have any # tasks using them.) worker_disable_rate_limits=True, # Task routing task_routes=route_for_task, # Allow pool restarts from remote worker_pool_restarts=True, # Do not prefetch tasks worker_prefetch_multiplier=1, # Send events worker_send_task_events=True, # Do not send useless task_sent events task_send_sent_event=False, ) def build_app(config=None): config = merge_configs( {k: v for (k, (_, v)) in DEFAULT_CONFIG.items()}, config or {} ) config["task_queues"] = CELERY_QUEUES + [ Queue(queue, Exchange(queue), routing_key=queue) for queue in config.get("task_queues", ()) ] logger.debug("Creating a Celery app with %s", config) # Instantiate the Celery app app = Celery(broker=config["task_broker"], task_cls="swh.scheduler.task:SWHTask") app.add_defaults(CELERY_DEFAULT_CONFIG) app.add_defaults(config) return app app = build_app(CONFIG) # XXX for BW compat Celery.get_queue_length = get_queue_length diff --git a/swh/scheduler/celery_backend/listener.py b/swh/scheduler/celery_backend/listener.py index a9e3fc7..b608269 100644 --- a/swh/scheduler/celery_backend/listener.py +++ b/swh/scheduler/celery_backend/listener.py @@ -1,222 +1,220 @@ # Copyright (C) 2015-2018 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime import logging -import time import sys - -import click +import time from arrow import utcnow -from kombu import Queue - import celery from celery.events import EventReceiver +import click +from kombu import Queue from swh.core.statsd import statsd class ReliableEventReceiver(EventReceiver): def __init__( self, channel, handlers=None, routing_key="#", node_id=None, app=None, queue_prefix="celeryev", accept=None, ): super(ReliableEventReceiver, self).__init__( channel, handlers, routing_key, node_id, app, queue_prefix, accept ) self.queue = Queue( ".".join([self.queue_prefix, self.node_id]), exchange=self.exchange, routing_key=self.routing_key, auto_delete=False, durable=True, ) def get_consumers(self, consumer, channel): return [ consumer( queues=[self.queue], callbacks=[self._receive], no_ack=False, accept=self.accept, ) ] def _receive(self, bodies, message): if not isinstance(bodies, list): # celery<4 returned body as element bodies = [bodies] for body in bodies: type, body = self.event_from_message(body) self.process(type, body, message) def process(self, type, event, message): """Process the received event by dispatching it to the appropriate handler.""" handler = self.handlers.get(type) or self.handlers.get("*") if handler: handler(event, message) statsd.increment( "swh_scheduler_listener_handled_event_total", tags={"event_type": type} ) ACTION_SEND_DELAY = datetime.timedelta(seconds=1.0) ACTION_QUEUE_MAX_LENGTH = 1000 def event_monitor(app, backend): logger = logging.getLogger("swh.scheduler.listener") actions = { "last_send": utcnow() - 2 * ACTION_SEND_DELAY, "queue": [], } def try_perform_actions(actions=actions): logger.debug("Try perform pending actions") if actions["queue"] and ( len(actions["queue"]) > ACTION_QUEUE_MAX_LENGTH or utcnow() - actions["last_send"] > ACTION_SEND_DELAY ): perform_actions(actions) def perform_actions(actions, backend=backend): logger.info("Perform %s pending actions" % len(actions["queue"])) action_map = { "start_task_run": backend.start_task_run, "end_task_run": backend.end_task_run, } messages = [] db = backend.get_db() try: cursor = db.cursor(None) for action in actions["queue"]: messages.append(action["message"]) function = action_map[action["action"]] args = action.get("args", ()) kwargs = action.get("kwargs", {}) kwargs["cur"] = cursor function(*args, **kwargs) except Exception: db.conn.rollback() else: db.conn.commit() finally: backend.put_db(db) for message in messages: if not message.acknowledged: message.ack() actions["queue"] = [] actions["last_send"] = utcnow() def queue_action(action, actions=actions): actions["queue"].append(action) try_perform_actions() def catchall_event(event, message): logger.debug("event: %s %s", event["type"], event.get("name", "N/A")) if not message.acknowledged: message.ack() try_perform_actions() def task_started(event, message): logger.debug("task_started: %s %s", event["type"], event.get("name", "N/A")) queue_action( { "action": "start_task_run", "args": [event["uuid"]], "kwargs": { "timestamp": utcnow(), "metadata": {"worker": event["hostname"],}, }, "message": message, } ) def task_succeeded(event, message): logger.debug("task_succeeded: event: %s" % event) logger.debug(" message: %s" % message) result = event["result"] logger.debug("task_succeeded: result: %s" % result) try: status = result.get("status") if status == "success": status = "eventful" if result.get("eventful") else "uneventful" except Exception: status = "eventful" if result else "uneventful" queue_action( { "action": "end_task_run", "args": [event["uuid"]], "kwargs": {"timestamp": utcnow(), "status": status, "result": result,}, "message": message, } ) def task_failed(event, message): logger.debug("task_failed: event: %s" % event) logger.debug(" message: %s" % message) queue_action( { "action": "end_task_run", "args": [event["uuid"]], "kwargs": {"timestamp": utcnow(), "status": "failed",}, "message": message, } ) recv = ReliableEventReceiver( celery.current_app.connection(), app=celery.current_app, handlers={ "task-started": task_started, "task-result": task_succeeded, "task-failed": task_failed, "*": catchall_event, }, node_id="listener", ) errors = 0 while True: try: recv.capture(limit=None, timeout=None, wakeup=True) errors = 0 except KeyboardInterrupt: logger.exception("Keyboard interrupt, exiting") break except Exception: logger.exception("Unexpected exception") if errors < 5: time.sleep(errors) errors += 1 else: logger.error("Too many consecutive errors, exiting") sys.exit(1) @click.command() @click.pass_context def main(ctx): click.echo("Deprecated! Use 'swh-scheduler listener' instead.", err=True) ctx.exit(1) if __name__ == "__main__": main() diff --git a/swh/scheduler/celery_backend/runner.py b/swh/scheduler/celery_backend/runner.py index b8592d0..957c1b7 100644 --- a/swh/scheduler/celery_backend/runner.py +++ b/swh/scheduler/celery_backend/runner.py @@ -1,131 +1,131 @@ # Copyright (C) 2015-2018 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -import arrow import logging + +import arrow from kombu.utils.uuid import uuid from swh.core.statsd import statsd -from swh.scheduler import get_scheduler, compute_nb_tasks_from - +from swh.scheduler import compute_nb_tasks_from, get_scheduler logger = logging.getLogger(__name__) # Max batch size for tasks MAX_NUM_TASKS = 10000 def run_ready_tasks(backend, app): """Run tasks that are ready Args: backend (Scheduler): backend to read tasks to schedule app (App): Celery application to send tasks to Returns: A list of dictionaries:: { 'task': the scheduler's task id, 'backend_id': Celery's task id, 'scheduler': arrow.utcnow() } The result can be used to block-wait for the tasks' results:: backend_tasks = run_ready_tasks(self.scheduler, app) for task in backend_tasks: AsyncResult(id=task['backend_id']).get() """ all_backend_tasks = [] while True: task_types = {} pending_tasks = [] for task_type in backend.get_task_types(): task_type_name = task_type["type"] task_types[task_type_name] = task_type max_queue_length = task_type["max_queue_length"] if max_queue_length is None: max_queue_length = 0 backend_name = task_type["backend_name"] if max_queue_length: try: queue_length = app.get_queue_length(backend_name) except ValueError: queue_length = None if queue_length is None: # Running without RabbitMQ (probably a test env). num_tasks = MAX_NUM_TASKS else: num_tasks = min(max_queue_length - queue_length, MAX_NUM_TASKS) else: num_tasks = MAX_NUM_TASKS # only pull tasks if the buffer is at least 1/5th empty (= 80% # full), to help postgresql use properly indexed queries. if num_tasks > min(MAX_NUM_TASKS, max_queue_length) // 5: num_tasks, num_tasks_priority = compute_nb_tasks_from(num_tasks) grabbed_tasks = backend.grab_ready_tasks( task_type_name, num_tasks=num_tasks, num_tasks_priority=num_tasks_priority, ) if grabbed_tasks: pending_tasks.extend(grabbed_tasks) logger.info( "Grabbed %s tasks %s", len(grabbed_tasks), task_type_name ) statsd.increment( "swh_scheduler_runner_scheduled_task_total", len(grabbed_tasks), tags={"task_type": task_type_name}, ) if not pending_tasks: return all_backend_tasks backend_tasks = [] celery_tasks = [] for task in pending_tasks: args = task["arguments"]["args"] kwargs = task["arguments"]["kwargs"] backend_name = task_types[task["type"]]["backend_name"] backend_id = uuid() celery_tasks.append((backend_name, backend_id, args, kwargs)) data = { "task": task["id"], "backend_id": backend_id, "scheduled": arrow.utcnow(), } backend_tasks.append(data) logger.debug("Sent %s celery tasks", len(backend_tasks)) backend.mass_schedule_task_runs(backend_tasks) for backend_name, backend_id, args, kwargs in celery_tasks: app.send_task( backend_name, task_id=backend_id, args=args, kwargs=kwargs, ) all_backend_tasks.extend(backend_tasks) def main(): from .config import app as main_app for module in main_app.conf.CELERY_IMPORTS: __import__(module) main_backend = get_scheduler("local") try: run_ready_tasks(main_backend, main_app) except Exception: main_backend.rollback() raise if __name__ == "__main__": main() diff --git a/swh/scheduler/cli/__init__.py b/swh/scheduler/cli/__init__.py index 4cd81c6..eb4db7c 100644 --- a/swh/scheduler/cli/__init__.py +++ b/swh/scheduler/cli/__init__.py @@ -1,93 +1,93 @@ # Copyright (C) 2016-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information # WARNING: do not import unnecessary things here to keep cli startup time under # control import logging import click from swh.core.cli import CONTEXT_SETTINGS, AliasedGroup @click.group(name="scheduler", context_settings=CONTEXT_SETTINGS, cls=AliasedGroup) @click.option( "--config-file", "-C", default=None, type=click.Path(exists=True, dir_okay=False,), help="Configuration file.", ) @click.option( "--database", "-d", default=None, help="Scheduling database DSN (imply cls is 'local')", ) @click.option( "--url", "-u", default=None, help="Scheduler's url access (imply cls is 'remote')" ) @click.option( "--no-stdout", is_flag=True, default=False, help="Do NOT output logs on the console" ) @click.pass_context def cli(ctx, config_file, database, url, no_stdout): """Software Heritage Scheduler tools. Use a local scheduler instance by default (plugged to the main scheduler db). """ try: from psycopg2 import OperationalError except ImportError: class OperationalError(Exception): pass from swh.core import config - from swh.scheduler import get_scheduler, DEFAULT_CONFIG + from swh.scheduler import DEFAULT_CONFIG, get_scheduler ctx.ensure_object(dict) logger = logging.getLogger(__name__) scheduler = None conf = config.read(config_file, DEFAULT_CONFIG) if "scheduler" not in conf: raise ValueError("missing 'scheduler' configuration") if database: conf["scheduler"]["cls"] = "local" conf["scheduler"]["args"]["db"] = database elif url: conf["scheduler"]["cls"] = "remote" conf["scheduler"]["args"] = {"url": url} sched_conf = conf["scheduler"] try: logger.debug("Instantiating scheduler with %s" % (sched_conf)) scheduler = get_scheduler(**sched_conf) except (ValueError, OperationalError): # it's the subcommand to decide whether not having a proper # scheduler instance is a problem. pass ctx.obj["scheduler"] = scheduler ctx.obj["config"] = conf from . import admin, celery_monitor, task, task_type # noqa def main(): import click.core click.core.DEPRECATED_HELP_NOTICE = """ DEPRECATED! Please use the command 'swh scheduler'.""" cli.deprecated = True return cli(auto_envvar_prefix="SWH_SCHEDULER") if __name__ == "__main__": main() diff --git a/swh/scheduler/cli/admin.py b/swh/scheduler/cli/admin.py index 9d8fdd2..2ce0873 100644 --- a/swh/scheduler/cli/admin.py +++ b/swh/scheduler/cli/admin.py @@ -1,110 +1,110 @@ # Copyright (C) 2016-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information # WARNING: do not import unnecessary things here to keep cli startup time under # control import logging import time import click from . import cli @cli.command("start-runner") @click.option( "--period", "-p", default=0, help=( "Period (in s) at witch pending tasks are checked and " "executed. Set to 0 (default) for a one shot." ), ) @click.pass_context def runner(ctx, period): """Starts a swh-scheduler runner service. This process is responsible for checking for ready-to-run tasks and schedule them.""" - from swh.scheduler.celery_backend.runner import run_ready_tasks from swh.scheduler.celery_backend.config import build_app + from swh.scheduler.celery_backend.runner import run_ready_tasks app = build_app(ctx.obj["config"].get("celery")) app.set_current() logger = logging.getLogger(__name__ + ".runner") scheduler = ctx.obj["scheduler"] logger.debug("Scheduler %s" % scheduler) try: while True: logger.debug("Run ready tasks") try: ntasks = len(run_ready_tasks(scheduler, app)) if ntasks: logger.info("Scheduled %s tasks", ntasks) except Exception: logger.exception("Unexpected error in run_ready_tasks()") if not period: break time.sleep(period) except KeyboardInterrupt: ctx.exit(0) @cli.command("start-listener") @click.pass_context def listener(ctx): """Starts a swh-scheduler listener service. This service is responsible for listening at task lifecycle events and handle their workflow status in the database.""" scheduler_backend = ctx.obj["scheduler"] if not scheduler_backend: raise ValueError("Scheduler class (local/remote) must be instantiated") broker = ( ctx.obj["config"] .get("celery", {}) .get("task_broker", "amqp://guest@localhost/%2f") ) from swh.scheduler.celery_backend.pika_listener import get_listener listener = get_listener(broker, "celeryev.listener", scheduler_backend) try: listener.start_consuming() finally: listener.stop_consuming() @cli.command("rpc-serve") @click.option("--host", default="0.0.0.0", help="Host to run the scheduler server api") @click.option("--port", default=5008, type=click.INT, help="Binding port of the server") @click.option( "--debug/--nodebug", default=None, help=( "Indicates if the server should run in debug mode. " "Defaults to True if log-level is DEBUG, False otherwise." ), ) @click.pass_context def rpc_server(ctx, host, port, debug): """Starts a swh-scheduler API HTTP server. """ if ctx.obj["config"]["scheduler"]["cls"] == "remote": click.echo( "The API server can only be started with a 'local' " "configuration", err=True, ) ctx.exit(1) from swh.scheduler.api import server server.app.config.update(ctx.obj["config"]) if debug is None: debug = ctx.obj["log_level"] <= logging.DEBUG server.app.run(host, port=port, debug=bool(debug)) diff --git a/swh/scheduler/cli/task.py b/swh/scheduler/cli/task.py index e415a64..c304163 100644 --- a/swh/scheduler/cli/task.py +++ b/swh/scheduler/cli/task.py @@ -1,763 +1,768 @@ # Copyright (C) 2016-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information # WARNING: do not import unnecessary things here to keep cli startup time under # control import locale import logging +from typing import TYPE_CHECKING, Any, Dict, Iterator, Optional import click -from typing import Any, Dict, Optional, Iterator, TYPE_CHECKING - from . import cli if TYPE_CHECKING: # importing swh.storage.interface triggers the load of 300+ modules, so... from swh.model.model import Origin from swh.storage.interface import StorageInterface locale.setlocale(locale.LC_ALL, "") ARROW_LOCALE = locale.getlocale(locale.LC_TIME)[0] class DateTimeType(click.ParamType): name = "time and date" def convert(self, value, param, ctx): import arrow if not isinstance(value, arrow.Arrow): value = arrow.get(value) return value DATETIME = DateTimeType() CONTEXT_SETTINGS = dict(help_option_names=["-h", "--help"]) def format_dict(d): import datetime + import arrow ret = {} for k, v in d.items(): if isinstance(v, (arrow.Arrow, datetime.date, datetime.datetime)): v = arrow.get(v).format() elif isinstance(v, dict): v = format_dict(v) ret[k] = v return ret def pretty_print_list(list, indent=0): """Pretty-print a list""" return "".join("%s%r\n" % (" " * indent, item) for item in list) def pretty_print_dict(dict, indent=0): """Pretty-print a list""" return "".join( "%s%s: %r\n" % (" " * indent, click.style(key, bold=True), value) for key, value in sorted(dict.items()) ) def pretty_print_run(run, indent=4): fmt = ( "{indent}{backend_id} [{status}]\n" "{indent} scheduled: {scheduled} [{started}:{ended}]" ) return fmt.format(indent=" " * indent, **format_dict(run)) def pretty_print_task(task, full=False): """Pretty-print a task If 'full' is True, also print the status and priority fields. >>> import datetime >>> task = { ... 'id': 1234, ... 'arguments': { ... 'args': ['foo', 'bar', True], ... 'kwargs': {'key': 'value', 'key2': 42}, ... }, ... 'current_interval': datetime.timedelta(hours=1), ... 'next_run': datetime.datetime(2019, 2, 21, 13, 52, 35, 407818), ... 'policy': 'oneshot', ... 'priority': None, ... 'status': 'next_run_not_scheduled', ... 'type': 'test_task', ... } >>> print(click.unstyle(pretty_print_task(task))) Task 1234 Next run: ... (2019-02-21 13:52:35+00:00) Interval: 1:00:00 Type: test_task Policy: oneshot Args: 'foo' 'bar' True Keyword args: key: 'value' key2: 42 >>> print(click.unstyle(pretty_print_task(task, full=True))) Task 1234 Next run: ... (2019-02-21 13:52:35+00:00) Interval: 1:00:00 Type: test_task Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: 'foo' 'bar' True Keyword args: key: 'value' key2: 42 """ import arrow next_run = arrow.get(task["next_run"]) lines = [ "%s %s\n" % (click.style("Task", bold=True), task["id"]), click.style(" Next run: ", bold=True), "%s (%s)" % (next_run.humanize(locale=ARROW_LOCALE), next_run.format()), "\n", click.style(" Interval: ", bold=True), str(task["current_interval"]), "\n", click.style(" Type: ", bold=True), task["type"] or "", "\n", click.style(" Policy: ", bold=True), task["policy"] or "", "\n", ] if full: lines += [ click.style(" Status: ", bold=True), task["status"] or "", "\n", click.style(" Priority: ", bold=True), task["priority"] or "", "\n", ] lines += [ click.style(" Args:\n", bold=True), pretty_print_list(task["arguments"]["args"], indent=4), click.style(" Keyword args:\n", bold=True), pretty_print_dict(task["arguments"]["kwargs"], indent=4), ] return "".join(lines) @cli.group("task") @click.pass_context def task(ctx): """Manipulate tasks.""" pass @task.command("schedule") @click.option( "--columns", "-c", multiple=True, default=["type", "args", "kwargs", "next_run"], type=click.Choice(["type", "args", "kwargs", "policy", "next_run"]), help="columns present in the CSV file", ) @click.option("--delimiter", "-d", default=",") @click.argument("file", type=click.File(encoding="utf-8")) @click.pass_context def schedule_tasks(ctx, columns, delimiter, file): """Schedule tasks from a CSV input file. The following columns are expected, and can be set through the -c option: - type: the type of the task to be scheduled (mandatory) - args: the arguments passed to the task (JSON list, defaults to an empty list) - kwargs: the keyword arguments passed to the task (JSON object, defaults to an empty dict) - next_run: the date at which the task should run (datetime, defaults to now) The CSV can be read either from a named file, or from stdin (use - as filename). Use sample: cat scheduling-task.txt | \ python3 -m swh.scheduler.cli \ --database 'service=swh-scheduler-dev' \ task schedule \ --columns type --columns kwargs --columns policy \ --delimiter ';' - """ import csv import json + import arrow tasks = [] now = arrow.utcnow() scheduler = ctx.obj["scheduler"] if not scheduler: raise ValueError("Scheduler class (local/remote) must be instantiated") reader = csv.reader(file, delimiter=delimiter) for line in reader: task = dict(zip(columns, line)) args = json.loads(task.pop("args", "[]")) kwargs = json.loads(task.pop("kwargs", "{}")) task["arguments"] = { "args": args, "kwargs": kwargs, } task["next_run"] = DATETIME.convert(task.get("next_run", now), None, None) tasks.append(task) created = scheduler.create_tasks(tasks) output = [ "Created %d tasks\n" % len(created), ] for task in created: output.append(pretty_print_task(task)) click.echo_via_pager("\n".join(output)) @task.command("add") @click.argument("type", nargs=1, required=True) @click.argument("options", nargs=-1) @click.option( "--policy", "-p", default="recurring", type=click.Choice(["recurring", "oneshot"]) ) @click.option( "--priority", "-P", default=None, type=click.Choice(["low", "normal", "high"]) ) @click.option("--next-run", "-n", default=None) @click.pass_context def schedule_task(ctx, type, options, policy, priority, next_run): """Schedule one task from arguments. The first argument is the name of the task type, further ones are positional and keyword argument(s) of the task, in YAML format. Keyword args are of the form key=value. Usage sample: swh-scheduler --database 'service=swh-scheduler' \ task add list-pypi swh-scheduler --database 'service=swh-scheduler' \ task add list-debian-distribution --policy=oneshot distribution=stretch Note: if the priority is not given, the task won't have the priority set, which is considered as the lowest priority level. """ import arrow from .utils import parse_options scheduler = ctx.obj["scheduler"] if not scheduler: raise ValueError("Scheduler class (local/remote) must be instantiated") now = arrow.utcnow() (args, kw) = parse_options(options) task = { "type": type, "policy": policy, "priority": priority, "arguments": {"args": args, "kwargs": kw,}, "next_run": DATETIME.convert(next_run or now, None, None), } created = scheduler.create_tasks([task]) output = [ "Created %d tasks\n" % len(created), ] for task in created: output.append(pretty_print_task(task)) click.echo("\n".join(output)) def iter_origins( # use string annotations to prevent some pkg loading storage: "StorageInterface", page_token: "Optional[str]" = None, ) -> "Iterator[Origin]": """Iterate over origins in the storage. Optionally starting from page_token. This logs regularly an info message during pagination with the page_token. This, in order to feed it back to the cli if the process interrupted. Yields origin model objects from the storage """ while True: page_result = storage.origin_list(page_token=page_token) page_token = page_result.next_page_token yield from page_result.results if not page_token: break click.echo(f"page_token: {page_token}\n") @task.command("schedule_origins") @click.argument("type", nargs=1, required=True) @click.argument("options", nargs=-1) @click.option( "--batch-size", "-b", "origin_batch_size", default=10, show_default=True, type=int, help="Number of origins per task", ) @click.option( "--page-token", default=0, show_default=True, type=str, help="Only schedule tasks for origins whose ID is greater", ) @click.option( "--limit", default=None, type=int, help="Limit the tasks scheduling up to this number of tasks", ) @click.option("--storage-url", "-g", help="URL of the (graph) storage API") @click.option( "--dry-run/--no-dry-run", is_flag=True, default=False, help="List only what would be scheduled.", ) @click.pass_context def schedule_origin_metadata_index( ctx, type, options, storage_url, origin_batch_size, page_token, limit, dry_run ): """Schedules tasks for origins that are already known. The first argument is the name of the task type, further ones are keyword argument(s) of the task in the form key=value, where value is in YAML format. Usage sample: swh-scheduler --database 'service=swh-scheduler' \ task schedule_origins index-origin-metadata """ from itertools import islice + from swh.storage import get_storage + from .utils import parse_options, schedule_origin_batches scheduler = ctx.obj["scheduler"] storage = get_storage("remote", url=storage_url) if dry_run: scheduler = None (args, kw) = parse_options(options) if args: raise click.ClickException("Only keywords arguments are allowed.") origins = iter_origins(storage, page_token=page_token) if limit: origins = islice(origins, limit) origin_urls = (origin.url for origin in origins) schedule_origin_batches(scheduler, type, origin_urls, origin_batch_size, kw) @task.command("list-pending") @click.argument("task-types", required=True, nargs=-1) @click.option( "--limit", "-l", required=False, type=click.INT, help="The maximum number of tasks to fetch", ) @click.option( "--before", "-b", required=False, type=DATETIME, help="List all jobs supposed to run before the given date", ) @click.pass_context def list_pending_tasks(ctx, task_types, limit, before): """List the tasks that are going to be run. You can override the number of tasks to fetch """ from swh.scheduler import compute_nb_tasks_from scheduler = ctx.obj["scheduler"] if not scheduler: raise ValueError("Scheduler class (local/remote) must be instantiated") num_tasks, num_tasks_priority = compute_nb_tasks_from(limit) output = [] for task_type in task_types: pending = scheduler.peek_ready_tasks( task_type, timestamp=before, num_tasks=num_tasks, num_tasks_priority=num_tasks_priority, ) output.append("Found %d %s tasks\n" % (len(pending), task_type)) for task in pending: output.append(pretty_print_task(task)) click.echo("\n".join(output)) @task.command("list") @click.option( "--task-id", "-i", default=None, multiple=True, metavar="ID", help="List only tasks whose id is ID.", ) @click.option( "--task-type", "-t", default=None, multiple=True, metavar="TYPE", help="List only tasks of type TYPE", ) @click.option( "--limit", "-l", required=False, type=click.INT, help="The maximum number of tasks to fetch.", ) @click.option( "--status", "-s", multiple=True, metavar="STATUS", type=click.Choice( ("next_run_not_scheduled", "next_run_scheduled", "completed", "disabled") ), default=None, help="List tasks whose status is STATUS.", ) @click.option( "--policy", "-p", default=None, type=click.Choice(["recurring", "oneshot"]), help="List tasks whose policy is POLICY.", ) @click.option( "--priority", "-P", default=None, multiple=True, type=click.Choice(["all", "low", "normal", "high"]), help="List tasks whose priority is PRIORITY.", ) @click.option( "--before", "-b", required=False, type=DATETIME, metavar="DATETIME", help="Limit to tasks supposed to run before the given date.", ) @click.option( "--after", "-a", required=False, type=DATETIME, metavar="DATETIME", help="Limit to tasks supposed to run after the given date.", ) @click.option( "--list-runs", "-r", is_flag=True, default=False, help="Also list past executions of each task.", ) @click.pass_context def list_tasks( ctx, task_id, task_type, limit, status, policy, priority, before, after, list_runs ): """List tasks. """ scheduler = ctx.obj["scheduler"] if not scheduler: raise ValueError("Scheduler class (local/remote) must be instantiated") if not task_type: task_type = [x["type"] for x in scheduler.get_task_types()] # if task_id is not given, default value for status is # 'next_run_not_scheduled' # if task_id is given, default status is 'all' if task_id is None and status is None: status = ["next_run_not_scheduled"] if status and "all" in status: status = None if priority and "all" in priority: priority = None output = [] tasks = scheduler.search_tasks( task_id=task_id, task_type=task_type, status=status, priority=priority, policy=policy, before=before, after=after, limit=limit, ) if list_runs: runs = {t["id"]: [] for t in tasks} for r in scheduler.get_task_runs([task["id"] for task in tasks]): runs[r["task"]].append(r) else: runs = {} output.append("Found %d tasks\n" % (len(tasks))) for task in tasks: output.append(pretty_print_task(task, full=True)) if runs.get(task["id"]): output.append(click.style(" Executions:", bold=True)) for run in runs[task["id"]]: output.append(pretty_print_run(run, indent=4)) click.echo("\n".join(output)) @task.command("respawn") @click.argument("task-ids", required=True, nargs=-1) @click.option( "--next-run", "-n", required=False, type=DATETIME, metavar="DATETIME", default=None, help="Re spawn the selected tasks at this date", ) @click.pass_context def respawn_tasks(ctx, task_ids, next_run): """Respawn tasks. Respawn tasks given by their ids (see the 'task list' command to find task ids) at the given date (immediately by default). Eg. swh-scheduler task respawn 1 3 12 """ import arrow scheduler = ctx.obj["scheduler"] if not scheduler: raise ValueError("Scheduler class (local/remote) must be instantiated") if next_run is None: next_run = arrow.utcnow() output = [] scheduler.set_status_tasks( task_ids, status="next_run_not_scheduled", next_run=next_run ) output.append("Respawn tasks %s\n" % (task_ids,)) click.echo("\n".join(output)) @task.command("archive") @click.option( "--before", "-b", default=None, help="""Task whose ended date is anterior will be archived. Default to current month's first day.""", ) @click.option( "--after", "-a", default=None, help="""Task whose ended date is after the specified date will be archived. Default to prior month's first day.""", ) @click.option( "--batch-index", default=1000, type=click.INT, help="Batch size of tasks to read from db to archive", ) @click.option( "--bulk-index", default=200, type=click.INT, help="Batch size of tasks to bulk index", ) @click.option( "--batch-clean", default=1000, type=click.INT, help="Batch size of task to clean after archival", ) @click.option( "--dry-run/--no-dry-run", is_flag=True, default=False, help="Default to list only what would be archived.", ) @click.option("--verbose", is_flag=True, default=False, help="Verbose mode") @click.option( "--cleanup/--no-cleanup", is_flag=True, default=True, help="Clean up archived tasks (default)", ) @click.option( "--start-from", type=click.STRING, default=None, help="(Optional) default page to start from.", ) @click.pass_context def archive_tasks( ctx, before, after, batch_index, bulk_index, batch_clean, dry_run, verbose, cleanup, start_from, ): """Archive task/task_run whose (task_type is 'oneshot' and task_status is 'completed') or (task_type is 'recurring' and task_status is 'disabled'). With --dry-run flag set (default), only list those. """ - import arrow from itertools import groupby + + import arrow + from swh.core.utils import grouper from swh.scheduler.backend_es import ElasticSearchBackend config = ctx.obj["config"] scheduler = ctx.obj["scheduler"] if not scheduler: raise ValueError("Scheduler class (local/remote) must be instantiated") logging.basicConfig(level=logging.DEBUG if verbose else logging.INFO) logger = logging.getLogger(__name__) logging.getLogger("urllib3").setLevel(logging.WARN) logging.getLogger("elasticsearch").setLevel(logging.ERROR) if dry_run: logger.info("**DRY-RUN** (only reading db)") if not cleanup: logger.info("**NO CLEANUP**") es_storage = ElasticSearchBackend(**config) now = arrow.utcnow() # Default to archive tasks from a rolling month starting the week # prior to the current one if not before: before = now.shift(weeks=-1).format("YYYY-MM-DD") if not after: after = now.shift(weeks=-1).shift(months=-1).format("YYYY-MM-DD") logger.debug( "index: %s; cleanup: %s; period: [%s ; %s]" % (not dry_run, not dry_run and cleanup, after, before) ) def get_index_name( data: Dict[str, Any], es_storage: ElasticSearchBackend = es_storage ) -> str: """Given a data record, determine the index's name through its ending date. This varies greatly depending on the task_run's status. """ date = data.get("started") if not date: date = data["scheduled"] return es_storage.compute_index_name(date.year, date.month) def index_data(before, page_token, batch_index): while True: result = scheduler.filter_task_to_archive( after, before, page_token=page_token, limit=batch_index ) tasks_sorted = sorted(result["tasks"], key=get_index_name) groups = groupby(tasks_sorted, key=get_index_name) for index_name, tasks_group in groups: logger.debug("Index tasks to %s" % index_name) if dry_run: for task in tasks_group: yield task continue yield from es_storage.streaming_bulk( index_name, tasks_group, source=["task_id", "task_run_id"], chunk_size=bulk_index, ) page_token = result.get("next_page_token") if page_token is None: break gen = index_data(before, page_token=start_from, batch_index=batch_index) if cleanup: for task_ids in grouper(gen, n=batch_clean): task_ids = list(task_ids) logger.info("Clean up %s tasks: [%s, ...]" % (len(task_ids), task_ids[0])) if dry_run: # no clean up continue ctx.obj["scheduler"].delete_archived_tasks(task_ids) else: for task_ids in grouper(gen, n=batch_index): task_ids = list(task_ids) logger.info("Indexed %s tasks: [%s, ...]" % (len(task_ids), task_ids[0])) logger.debug("Done!") diff --git a/swh/scheduler/cli/task_type.py b/swh/scheduler/cli/task_type.py index 4320d44..c6368eb 100644 --- a/swh/scheduler/cli/task_type.py +++ b/swh/scheduler/cli/task_type.py @@ -1,233 +1,232 @@ # Copyright (C) 2016-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -# WARNING: do not import unnecessary things here to keep cli startup time under -# control -import click -import logging - from importlib import import_module +import logging from typing import Mapping +# WARNING: do not import unnecessary things here to keep cli startup time under +# control +import click from pkg_resources import iter_entry_points -from . import cli +from . import cli logger = logging.getLogger(__name__) DEFAULT_TASK_TYPE = { "full": { # for tasks like 'list_xxx_full()' "default_interval": "90 days", "min_interval": "90 days", "max_interval": "90 days", "backoff_factor": 1, }, "*": { # value if not suffix matches "default_interval": "1 day", "min_interval": "1 day", "max_interval": "1 day", "backoff_factor": 1, }, } PLUGIN_WORKER_DESCRIPTIONS = { entry_point.name: entry_point for entry_point in iter_entry_points("swh.workers") } @cli.group("task-type") @click.pass_context def task_type(ctx): """Manipulate task types.""" pass @task_type.command("list") @click.option("--verbose", "-v", is_flag=True, default=False, help="Verbose mode") @click.option( "--task_type", "-t", multiple=True, default=None, help="List task types of given type", ) @click.option( "--task_name", "-n", multiple=True, default=None, help="List task types of given backend task name", ) @click.pass_context def list_task_types(ctx, verbose, task_type, task_name): click.echo("Known task types:") if verbose: tmpl = ( click.style("{type}: ", bold=True) + """{backend_name} {description} interval: {default_interval} [{min_interval}, {max_interval}] backoff_factor: {backoff_factor} max_queue_length: {max_queue_length} num_retries: {num_retries} retry_delay: {retry_delay} """ ) else: tmpl = "{type}:\n {description}" for tasktype in sorted( ctx.obj["scheduler"].get_task_types(), key=lambda x: x["type"] ): if task_type and tasktype["type"] not in task_type: continue if task_name and tasktype["backend_name"] not in task_name: continue click.echo(tmpl.format(**tasktype)) @task_type.command("register") @click.option( "--plugins", "-p", "plugins", multiple=True, default=("all",), type=click.Choice(["all"] + list(PLUGIN_WORKER_DESCRIPTIONS)), help="Registers task-types for provided plugins. " "Defaults to all", ) @click.pass_context def register_task_types(ctx, plugins): """Register missing task-type entries in the scheduler. According to declared tasks in each loaded worker (e.g. lister, loader, ...) plugins. """ import celery.app.task scheduler = ctx.obj["scheduler"] if plugins == ("all",): plugins = list(PLUGIN_WORKER_DESCRIPTIONS) for plugin in plugins: entrypoint = PLUGIN_WORKER_DESCRIPTIONS[plugin] logger.info("Loading entrypoint for plugin %s", plugin) registry_entry = entrypoint.load()() for task_module in registry_entry["task_modules"]: mod = import_module(task_module) for task_name in (x for x in dir(mod) if not x.startswith("_")): logger.debug("Loading task name %s", task_name) taskobj = getattr(mod, task_name) if isinstance(taskobj, celery.app.task.Task): tt_name = task_name.replace("_", "-") task_cfg = registry_entry.get("task_types", {}).get(tt_name, {}) ensure_task_type(task_module, tt_name, taskobj, task_cfg, scheduler) def ensure_task_type( task_module: str, task_type: str, swhtask, task_config: Mapping, scheduler ): """Ensure a given task-type (for the task_module) exists in the scheduler. Args: task_module: task module we are currently checking for task type consistency task_type: the type of the task to check/insert (correspond to the 'type' field in the db) swhtask (SWHTask): the SWHTask instance the task-type correspond to task_config: a dict with specific/overloaded values for the task-type to be created scheduler: the scheduler object used to access the scheduler db """ for suffix, defaults in DEFAULT_TASK_TYPE.items(): if task_type.endswith("-" + suffix): task_type_dict = defaults.copy() break else: task_type_dict = DEFAULT_TASK_TYPE["*"].copy() task_type_dict["type"] = task_type task_type_dict["backend_name"] = swhtask.name if swhtask.__doc__: task_type_dict["description"] = swhtask.__doc__.splitlines()[0] task_type_dict.update(task_config) current_task_type = scheduler.get_task_type(task_type) if current_task_type: # Ensure the existing task_type is consistent in the scheduler if current_task_type["backend_name"] != task_type_dict["backend_name"]: logger.warning( "Existing task type %s for module %s has a " "different backend name than current " "code version provides (%s vs. %s)", task_type, task_module, current_task_type["backend_name"], task_type_dict["backend_name"], ) else: logger.info("Create task type %s in scheduler", task_type) logger.debug(" %s", task_type_dict) scheduler.create_task_type(task_type_dict) @task_type.command("add") @click.argument("type", required=True) @click.argument("task-name", required=True) @click.argument("description", required=True) @click.option( "--default-interval", "-i", default="90 days", help='Default interval ("90 days" by default)', ) @click.option( "--min-interval", default=None, help="Minimum interval (default interval if not set)", ) @click.option( "--max-interval", "-i", default=None, help="Maximal interval (default interval if not set)", ) @click.option("--backoff-factor", "-f", type=float, default=1, help="Backoff factor") @click.pass_context def add_task_type( ctx, type, task_name, description, default_interval, min_interval, max_interval, backoff_factor, ): """Create a new task type """ scheduler = ctx.obj["scheduler"] if not scheduler: raise ValueError("Scheduler class (local/remote) must be instantiated") task_type = dict( type=type, backend_name=task_name, description=description, default_interval=default_interval, min_interval=min_interval, max_interval=max_interval, backoff_factor=backoff_factor, max_queue_length=None, num_retries=None, retry_delay=None, ) scheduler.create_task_type(task_type) click.echo("OK") diff --git a/swh/scheduler/cli/utils.py b/swh/scheduler/cli/utils.py index 2fb02a5..688965a 100644 --- a/swh/scheduler/cli/utils.py +++ b/swh/scheduler/cli/utils.py @@ -1,90 +1,90 @@ # Copyright (C) 2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information # WARNING: do not import unnecessary things here to keep cli startup time under # control import click - TASK_BATCH_SIZE = 1000 # Number of tasks per query to the scheduler def schedule_origin_batches(scheduler, task_type, origins, origin_batch_size, kwargs): from itertools import islice + from swh.scheduler.utils import create_task_dict nb_origins = 0 nb_tasks = 0 while True: task_batch = [] for _ in range(TASK_BATCH_SIZE): # Group origins origin_batch = [] for origin in islice(origins, origin_batch_size): origin_batch.append(origin) nb_origins += len(origin_batch) if not origin_batch: break # Create a task for these origins args = [origin_batch] task_dict = create_task_dict(task_type, "oneshot", *args, **kwargs) task_batch.append(task_dict) # Schedule a batch of tasks if not task_batch: break nb_tasks += len(task_batch) if scheduler: scheduler.create_tasks(task_batch) click.echo("Scheduled %d tasks (%d origins)." % (nb_tasks, nb_origins)) # Print final status. if nb_tasks: click.echo("Done.") else: click.echo("Nothing to do (no origin metadata matched the criteria).") def parse_argument(option): import yaml try: return yaml.safe_load(option) except Exception: raise click.ClickException("Invalid argument: {}".format(option)) def parse_options(options): """Parses options from a CLI as YAML and turns it into Python args and kwargs. >>> parse_options([]) ([], {}) >>> parse_options(['foo', 'bar']) (['foo', 'bar'], {}) >>> parse_options(['[foo, bar]']) ([['foo', 'bar']], {}) >>> parse_options(['"foo"', '"bar"']) (['foo', 'bar'], {}) >>> parse_options(['foo="bar"']) ([], {'foo': 'bar'}) >>> parse_options(['"foo"', 'bar="baz"']) (['foo'], {'bar': 'baz'}) >>> parse_options(['42', 'bar=False']) ([42], {'bar': False}) >>> parse_options(['42', 'bar=false']) ([42], {'bar': False}) >>> parse_options(['42', '"foo']) Traceback (most recent call last): ... click.exceptions.ClickException: Invalid argument: "foo """ kw_pairs = [x.split("=", 1) for x in options if "=" in x] args = [parse_argument(x) for x in options if "=" not in x] kw = {k: parse_argument(v) for (k, v) in kw_pairs} return (args, kw) diff --git a/swh/scheduler/elasticsearch_memory.py b/swh/scheduler/elasticsearch_memory.py index da59852..2dd958c 100644 --- a/swh/scheduler/elasticsearch_memory.py +++ b/swh/scheduler/elasticsearch_memory.py @@ -1,146 +1,144 @@ # Copyright (C) 2018-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """Memory Elastic Search backend """ +from ast import literal_eval import datetime # noqa serialization purposes import hashlib import logging - -from ast import literal_eval from typing import Optional import psycopg2 # noqa serialization purposes - logger = logging.getLogger(__name__) class BasicSerializer: """For memory elastic search implementation (not for production) """ def __init__(self, *args, **kwargs): pass def dumps(self, *args, **kwargs): return str(*args) class BasicTransport: """For memory elastic search implementation, (not for production) """ def __init__(self, *args, **kwargs): self.serializer = BasicSerializer() class MemoryElasticsearch: """Memory Elasticsearch instance (for test purposes) Partial implementation oriented towards index storage (and not search) For now, its sole client is the scheduler for task archival purposes. """ def __init__(self, *args, **kwargs): self.index = {} self.mapping = {} self.settings = {} self.indices = self # HACK self.main_mapping_key: Optional[str] = None self.main_settings_key: Optional[str] = None self.transport = BasicTransport() def create(self, index, **kwargs): logger.debug(f"create index {index}") logger.debug(f"indices: {self.index}") logger.debug(f"mapping: {self.mapping}") logger.debug(f"settings: {self.settings}") self.index[index] = { "status": "opened", "data": {}, "mapping": self.get_mapping(self.main_mapping_key), "settings": self.get_settings(self.main_settings_key), } logger.debug(f"index {index} created") def close(self, index, **kwargs): """Close index""" idx = self.index.get(index) if idx: idx["status"] = "closed" def open(self, index, **kwargs): """Open index""" idx = self.index.get(index) if idx: idx["status"] = "opened" def bulk(self, body, **kwargs): """Bulk insert document in index""" assert isinstance(body, str) all_data = body.split("\n") if all_data[-1] == "": all_data = all_data[:-1] # drop the empty line if any ids = [] # data is sent as tuple (index, data-to-index) for i in range(0, len(all_data), 2): # The first entry is about the index to use # not about a data to index # find the index index_data = literal_eval(all_data[i]) idx_name = index_data["index"]["_index"] # associated data to index data = all_data[i + 1] _id = hashlib.sha1(data.encode("utf-8")).hexdigest() parsed_data = eval(data) # for datetime self.index[idx_name]["data"][_id] = parsed_data ids.append(_id) # everything is indexed fine return {"items": [{"index": {"status": 200, "_id": _id,}} for _id in ids]} def mget(self, *args, body, index, **kwargs): """Bulk indexed documents retrieval""" idx = self.index[index] docs = [] idx_docs = idx["data"] for _id in body["ids"]: doc = idx_docs.get(_id) if doc: d = { "found": True, "_source": doc, } docs.append(d) return {"docs": docs} def stats(self, index, **kwargs): idx = self.index[index] # will raise if it does not exist if not idx or idx["status"] == "closed": raise ValueError("Closed index") # simulate issue if index closed def exists(self, index, **kwargs): return self.index.get(index) is not None def put_mapping(self, index, body, **kwargs): self.mapping[index] = body self.main_mapping_key = index def get_mapping(self, index, **kwargs): return self.mapping.get(index) or self.index.get(index, {}).get("mapping", {}) def put_settings(self, index, body, **kwargs): self.settings[index] = body self.main_settings_key = index def get_settings(self, index, **kwargs): return self.settings.get(index) or self.index.get(index, {}).get("settings", {}) diff --git a/swh/scheduler/interface.py b/swh/scheduler/interface.py index 69787ea..0496093 100644 --- a/swh/scheduler/interface.py +++ b/swh/scheduler/interface.py @@ -1,312 +1,311 @@ # Copyright (C) 2015-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from typing import Any, Dict, Iterable, List, Optional from uuid import UUID from swh.core.api import remote_api_endpoint - from swh.scheduler.model import ( ListedOrigin, ListedOriginPageToken, Lister, PaginatedListedOriginList, ) class SchedulerInterface: @remote_api_endpoint("task_type/create") def create_task_type(self, task_type): """Create a new task type ready for scheduling. Args: task_type (dict): a dictionary with the following keys: - type (str): an identifier for the task type - description (str): a human-readable description of what the task does - backend_name (str): the name of the task in the job-scheduling backend - default_interval (datetime.timedelta): the default interval between two task runs - min_interval (datetime.timedelta): the minimum interval between two task runs - max_interval (datetime.timedelta): the maximum interval between two task runs - backoff_factor (float): the factor by which the interval changes at each run - max_queue_length (int): the maximum length of the task queue for this task type """ ... @remote_api_endpoint("task_type/get") def get_task_type(self, task_type_name): """Retrieve the task type with id task_type_name""" ... @remote_api_endpoint("task_type/get_all") def get_task_types(self): """Retrieve all registered task types""" ... @remote_api_endpoint("task/create") def create_tasks(self, tasks, policy="recurring"): """Create new tasks. Args: tasks (list): each task is a dictionary with the following keys: - type (str): the task type - arguments (dict): the arguments for the task runner, keys: - args (list of str): arguments - kwargs (dict str -> str): keyword arguments - next_run (datetime.datetime): the next scheduled run for the task Returns: a list of created tasks. """ ... @remote_api_endpoint("task/set_status") def set_status_tasks(self, task_ids, status="disabled", next_run=None): """Set the tasks' status whose ids are listed. If given, also set the next_run date. """ ... @remote_api_endpoint("task/disable") def disable_tasks(self, task_ids): """Disable the tasks whose ids are listed.""" ... @remote_api_endpoint("task/search") def search_tasks( self, task_id=None, task_type=None, status=None, priority=None, policy=None, before=None, after=None, limit=None, ): """Search tasks from selected criterions""" ... @remote_api_endpoint("task/get") def get_tasks(self, task_ids): """Retrieve the info of tasks whose ids are listed.""" ... @remote_api_endpoint("task/peek_ready") def peek_ready_tasks( self, task_type, timestamp=None, num_tasks=None, num_tasks_priority=None, ): """Fetch the list of ready tasks Args: task_type (str): filtering task per their type timestamp (datetime.datetime): peek tasks that need to be executed before that timestamp num_tasks (int): only peek at num_tasks tasks (with no priority) num_tasks_priority (int): only peek at num_tasks_priority tasks (with priority) Returns: a list of tasks """ ... @remote_api_endpoint("task/grab_ready") def grab_ready_tasks( self, task_type, timestamp=None, num_tasks=None, num_tasks_priority=None, ): """Fetch the list of ready tasks, and mark them as scheduled Args: task_type (str): filtering task per their type timestamp (datetime.datetime): grab tasks that need to be executed before that timestamp num_tasks (int): only grab num_tasks tasks (with no priority) num_tasks_priority (int): only grab oneshot num_tasks tasks (with priorities) Returns: a list of tasks """ ... @remote_api_endpoint("task_run/schedule_one") def schedule_task_run(self, task_id, backend_id, metadata=None, timestamp=None): """Mark a given task as scheduled, adding a task_run entry in the database. Args: task_id (int): the identifier for the task being scheduled backend_id (str): the identifier of the job in the backend metadata (dict): metadata to add to the task_run entry timestamp (datetime.datetime): the instant the event occurred Returns: a fresh task_run entry """ ... @remote_api_endpoint("task_run/schedule") def mass_schedule_task_runs(self, task_runs): """Schedule a bunch of task runs. Args: task_runs (list): a list of dicts with keys: - task (int): the identifier for the task being scheduled - backend_id (str): the identifier of the job in the backend - metadata (dict): metadata to add to the task_run entry - scheduled (datetime.datetime): the instant the event occurred Returns: None """ ... @remote_api_endpoint("task_run/start") def start_task_run(self, backend_id, metadata=None, timestamp=None): """Mark a given task as started, updating the corresponding task_run entry in the database. Args: backend_id (str): the identifier of the job in the backend metadata (dict): metadata to add to the task_run entry timestamp (datetime.datetime): the instant the event occurred Returns: the updated task_run entry """ ... @remote_api_endpoint("task_run/end") def end_task_run( self, backend_id, status, metadata=None, timestamp=None, result=None, ): """Mark a given task as ended, updating the corresponding task_run entry in the database. Args: backend_id (str): the identifier of the job in the backend status (str): how the task ended; one of: 'eventful', 'uneventful', 'failed' metadata (dict): metadata to add to the task_run entry timestamp (datetime.datetime): the instant the event occurred Returns: the updated task_run entry """ ... @remote_api_endpoint("task/filter_for_archive") def filter_task_to_archive( self, after_ts: str, before_ts: str, limit: int = 10, page_token: Optional[str] = None, ) -> Dict[str, Any]: """Compute the tasks to archive within the datetime interval [after_ts, before_ts[. The method returns a paginated result. Returns: dict with the following keys: - **next_page_token**: opaque token to be used as `page_token` to retrieve the next page of result. If absent, there is no more pages to gather. - **tasks**: list of task dictionaries with the following keys: **id** (str): origin task id **started** (Optional[datetime]): started date **scheduled** (datetime): scheduled date **arguments** (json dict): task's arguments ... """ ... @remote_api_endpoint("task/delete_archived") def delete_archived_tasks(self, task_ids): """Delete archived tasks as much as possible. Only the task_ids whose complete associated task_run have been cleaned up will be. """ ... @remote_api_endpoint("task_run/get") def get_task_runs(self, task_ids, limit=None): """Search task run for a task id""" ... @remote_api_endpoint("lister/get_or_create") def get_or_create_lister( self, name: str, instance_name: Optional[str] = None ) -> Lister: """Retrieve information about the given instance of the lister from the database, or create the entry if it did not exist. """ ... @remote_api_endpoint("lister/update") def update_lister(self, lister: Lister) -> Lister: """Update the state for the given lister instance in the database. Returns: a new Lister object, with all fields updated from the database Raises: StaleData if the `updated` timestamp for the lister instance in database doesn't match the one passed by the user. """ ... @remote_api_endpoint("origins/record") def record_listed_origins( self, listed_origins: Iterable[ListedOrigin] ) -> List[ListedOrigin]: """Record a set of origins that a lister has listed. This performs an "upsert": origins with the same (lister_id, url, visit_type) values are updated with new values for extra_loader_arguments, last_update and last_seen. """ ... @remote_api_endpoint("origins/get") def get_listed_origins( self, lister_id: Optional[UUID] = None, url: Optional[str] = None, limit: int = 1000, page_token: Optional[ListedOriginPageToken] = None, ) -> PaginatedListedOriginList: """Get information on the listed origins matching either the `url` or `lister_id`, or both arguments. Use the `limit` and `page_token` arguments for continuation. The next page token, if any, is returned in the PaginatedListedOriginList object. """ ... @remote_api_endpoint("priority_ratios/get") def get_priority_ratios(self): ... diff --git a/swh/scheduler/model.py b/swh/scheduler/model.py index 904aaf9..e3827ff 100644 --- a/swh/scheduler/model.py +++ b/swh/scheduler/model.py @@ -1,193 +1,193 @@ # Copyright (C) 2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime -from uuid import UUID from typing import Any, Dict, List, Optional, Tuple, Union +from uuid import UUID import attr import attr.converters from attrs_strict import type_validator @attr.s class BaseSchedulerModel: """Base class for database-backed objects. These database-backed objects are defined through attrs-based attributes that match the columns of the database 1:1. This is a (very) lightweight ORM. These attrs-based attributes have metadata specific to the functionality expected from these fields in the database: - `primary_key`: the column is a primary key; it should be filtered out when doing an `update` of the object - `auto_primary_key`: the column is a primary key, which is automatically handled by the database. It will not be inserted to. This must be matched with a database-side default value. - `auto_now_add`: the column is a timestamp that is set to the current time when the object is inserted, and never updated afterwards. This must be matched with a database-side default value. - `auto_now`: the column is a timestamp that is set to the current time when the object is inserted or updated. """ _pk_cols: Optional[Tuple[str, ...]] = None _select_cols: Optional[Tuple[str, ...]] = None _insert_cols_and_metavars: Optional[Tuple[Tuple[str, ...], Tuple[str, ...]]] = None @classmethod def primary_key_columns(cls) -> Tuple[str, ...]: """Get the primary key columns for this object type""" if cls._pk_cols is None: columns: List[str] = [] for field in attr.fields(cls): if any( field.metadata.get(flag) for flag in ("auto_primary_key", "primary_key") ): columns.append(field.name) cls._pk_cols = tuple(sorted(columns)) return cls._pk_cols @classmethod def select_columns(cls) -> Tuple[str, ...]: """Get all the database columns needed for a `select` on this object type""" if cls._select_cols is None: columns: List[str] = [] for field in attr.fields(cls): columns.append(field.name) cls._select_cols = tuple(sorted(columns)) return cls._select_cols @classmethod def insert_columns_and_metavars(cls) -> Tuple[Tuple[str, ...], Tuple[str, ...]]: """Get the database columns and metavars needed for an `insert` or `update` on this object type. This implements support for the `auto_*` field metadata attributes. """ if cls._insert_cols_and_metavars is None: zipped_cols_and_metavars: List[Tuple[str, str]] = [] for field in attr.fields(cls): if any( field.metadata.get(flag) for flag in ("auto_now_add", "auto_primary_key") ): continue elif field.metadata.get("auto_now"): zipped_cols_and_metavars.append((field.name, "now()")) else: zipped_cols_and_metavars.append((field.name, f"%({field.name})s")) zipped_cols_and_metavars.sort() cols, metavars = zip(*zipped_cols_and_metavars) cls._insert_cols_and_metavars = cols, metavars return cls._insert_cols_and_metavars @attr.s class Lister(BaseSchedulerModel): name = attr.ib(type=str, validator=[type_validator()]) instance_name = attr.ib(type=str, validator=[type_validator()]) # Populated by database id = attr.ib( type=Optional[UUID], validator=type_validator(), default=None, metadata={"auto_primary_key": True}, ) current_state = attr.ib( type=Dict[str, Any], validator=[type_validator()], factory=dict ) created = attr.ib( type=Optional[datetime.datetime], validator=[type_validator()], default=None, metadata={"auto_now_add": True}, ) updated = attr.ib( type=Optional[datetime.datetime], validator=[type_validator()], default=None, metadata={"auto_now": True}, ) @attr.s class ListedOrigin(BaseSchedulerModel): """Basic information about a listed origin, output by a lister""" lister_id = attr.ib( type=UUID, validator=[type_validator()], metadata={"primary_key": True} ) url = attr.ib( type=str, validator=[type_validator()], metadata={"primary_key": True} ) visit_type = attr.ib( type=str, validator=[type_validator()], metadata={"primary_key": True} ) extra_loader_arguments = attr.ib( type=Dict[str, str], validator=[type_validator()], factory=dict ) last_update = attr.ib( type=Optional[datetime.datetime], validator=[type_validator()], default=None, ) enabled = attr.ib(type=bool, validator=[type_validator()], default=True) first_seen = attr.ib( type=Optional[datetime.datetime], validator=[type_validator()], default=None, metadata={"auto_now_add": True}, ) last_seen = attr.ib( type=Optional[datetime.datetime], validator=[type_validator()], default=None, metadata={"auto_now": True}, ) ListedOriginPageToken = Tuple[UUID, str] def convert_listed_origin_page_token( input: Union[None, ListedOriginPageToken, List[Union[UUID, str]]] ) -> Optional[ListedOriginPageToken]: if input is None: return None if isinstance(input, tuple): return input x, y = input assert isinstance(x, UUID) assert isinstance(y, str) return (x, y) @attr.s class PaginatedListedOriginList(BaseSchedulerModel): """A list of listed origins, with a continuation token""" origins = attr.ib(type=List[ListedOrigin], validator=[type_validator()]) next_page_token = attr.ib( type=Optional[ListedOriginPageToken], validator=[type_validator()], converter=convert_listed_origin_page_token, default=None, ) diff --git a/swh/scheduler/pytest_plugin.py b/swh/scheduler/pytest_plugin.py index 723bc3a..a7ae36c 100644 --- a/swh/scheduler/pytest_plugin.py +++ b/swh/scheduler/pytest_plugin.py @@ -1,111 +1,108 @@ # Copyright (C) 2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from datetime import timedelta import glob import os from celery.contrib.testing import worker from celery.contrib.testing.app import TestApp, setup_default_app - import pkg_resources import pytest from swh.core.utils import numfile_sortkey as sortkey - import swh.scheduler from swh.scheduler import get_scheduler - SQL_DIR = os.path.join(os.path.dirname(swh.scheduler.__file__), "sql") DUMP_FILES = os.path.join(SQL_DIR, "*.sql") # celery tasks for testing purpose; tasks themselves should be # in swh/scheduler/tests/tasks.py TASK_NAMES = ["ping", "multiping", "add", "error", "echo"] @pytest.fixture def swh_scheduler_config(request, postgresql): scheduler_config = { "db": postgresql.dsn, } all_dump_files = sorted(glob.glob(DUMP_FILES), key=sortkey) cursor = postgresql.cursor() for fname in all_dump_files: with open(fname) as fobj: cursor.execute(fobj.read()) postgresql.commit() return scheduler_config @pytest.fixture def swh_scheduler(swh_scheduler_config): scheduler = get_scheduler("local", swh_scheduler_config) for taskname in TASK_NAMES: scheduler.create_task_type( { "type": "swh-test-{}".format(taskname), "description": "The {} testing task".format(taskname), "backend_name": "swh.scheduler.tests.tasks.{}".format(taskname), "default_interval": timedelta(days=1), "min_interval": timedelta(hours=6), "max_interval": timedelta(days=12), } ) return scheduler # this alias is used to be able to easily instantiate a db-backed Scheduler # eg. for the RPC client/server test suite. swh_db_scheduler = swh_scheduler @pytest.fixture(scope="session") def swh_scheduler_celery_app(): """Set up a Celery app as swh.scheduler and swh worker tests would expect it""" test_app = TestApp( set_as_current=True, enable_logging=True, task_cls="swh.scheduler.task:SWHTask", config={ "accept_content": ["application/x-msgpack", "application/json"], "task_serializer": "msgpack", "result_serializer": "json", }, ) with setup_default_app(test_app, use_trap=False): from swh.scheduler.celery_backend import config config.app = test_app test_app.set_default() test_app.set_current() yield test_app @pytest.fixture(scope="session") def swh_scheduler_celery_includes(): """List of task modules that should be loaded by the swh_scheduler_celery_worker on startup.""" task_modules = ["swh.scheduler.tests.tasks"] for entrypoint in pkg_resources.iter_entry_points("swh.workers"): task_modules.extend(entrypoint.load()().get("task_modules", [])) return task_modules @pytest.fixture(scope="session") def swh_scheduler_celery_worker( swh_scheduler_celery_app, swh_scheduler_celery_includes, ): """Spawn a worker""" for module in swh_scheduler_celery_includes: swh_scheduler_celery_app.loader.import_task_module(module) with worker.start_worker(swh_scheduler_celery_app, pool="solo") as w: yield w diff --git a/swh/scheduler/tests/common.py b/swh/scheduler/tests/common.py index 9c2c463..13c38d3 100644 --- a/swh/scheduler/tests/common.py +++ b/swh/scheduler/tests/common.py @@ -1,104 +1,103 @@ # Copyright (C) 2017-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import copy import datetime - TEMPLATES = { "git": { "type": "update-git", "arguments": {"args": [], "kwargs": {},}, "next_run": None, }, "hg": { "type": "update-hg", "arguments": {"args": [], "kwargs": {},}, "next_run": None, "policy": "oneshot", }, } TASK_TYPES = { "git": { "type": "update-git", "description": "Update a git repository", "backend_name": "swh.loader.git.tasks.UpdateGitRepository", "default_interval": datetime.timedelta(days=64), "min_interval": datetime.timedelta(hours=12), "max_interval": datetime.timedelta(days=64), "backoff_factor": 2, "max_queue_length": None, "num_retries": 7, "retry_delay": datetime.timedelta(hours=2), }, "hg": { "type": "update-hg", "description": "Update a mercurial repository", "backend_name": "swh.loader.mercurial.tasks.UpdateHgRepository", "default_interval": datetime.timedelta(days=64), "min_interval": datetime.timedelta(hours=12), "max_interval": datetime.timedelta(days=64), "backoff_factor": 2, "max_queue_length": None, "num_retries": 7, "retry_delay": datetime.timedelta(hours=2), }, } def tasks_from_template(template, max_timestamp, num, num_priority=0, priorities=None): """Build tasks from template """ def _task_from_template(template, next_run, priority, *args, **kwargs): ret = copy.deepcopy(template) ret["next_run"] = next_run if priority: ret["priority"] = priority if args: ret["arguments"]["args"] = list(args) if kwargs: ret["arguments"]["kwargs"] = kwargs return ret def _pop_priority(priorities): if not priorities: return None for priority, remains in priorities.items(): if remains > 0: priorities[priority] = remains - 1 return priority return None if num_priority and priorities: priorities = { priority: ratio * num_priority for priority, ratio in priorities.items() } tasks = [] for i in range(num + num_priority): priority = _pop_priority(priorities) tasks.append( _task_from_template( template, max_timestamp - datetime.timedelta(microseconds=i), priority, "argument-%03d" % i, **{"kwarg%03d" % i: "bogus-kwarg"}, ) ) return tasks LISTERS = ( {"name": "github"}, {"name": "gitlab", "instance_name": "gitlab"}, {"name": "gitlab", "instance_name": "freedesktop"}, {"name": "npm"}, {"name": "pypi"}, ) diff --git a/swh/scheduler/tests/conftest.py b/swh/scheduler/tests/conftest.py index d187c42..91e92bc 100644 --- a/swh/scheduler/tests/conftest.py +++ b/swh/scheduler/tests/conftest.py @@ -1,42 +1,41 @@ # Copyright (C) 2016-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -import os from datetime import datetime, timezone +import os from typing import List import pytest from swh.scheduler.model import ListedOrigin, Lister from swh.scheduler.tests.common import LISTERS - # make sure we are not fooled by CELERY_ config environment vars for var in [x for x in os.environ.keys() if x.startswith("CELERY")]: os.environ.pop(var) # test_cli tests depends on a en/C locale, so ensure it os.environ["LC_ALL"] = "C.UTF-8" @pytest.fixture def stored_lister(swh_scheduler) -> Lister: """Store a lister in the scheduler and return its information""" return swh_scheduler.get_or_create_lister(**LISTERS[0]) @pytest.fixture def listed_origins(stored_lister) -> List[ListedOrigin]: """Return a (fixed) set of 1000 listed origins""" return [ ListedOrigin( lister_id=stored_lister.id, url=f"https://example.com/{i:04d}.git", visit_type="git", last_update=datetime(2020, 6, 15, 16, 0, 0, i, tzinfo=timezone.utc), ) for i in range(1000) ] diff --git a/swh/scheduler/tests/es/conftest.py b/swh/scheduler/tests/es/conftest.py index 8d86522..6b3028d 100644 --- a/swh/scheduler/tests/es/conftest.py +++ b/swh/scheduler/tests/es/conftest.py @@ -1,49 +1,48 @@ # Copyright (C) 2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -import yaml - import pytest +import yaml from swh.scheduler import get_scheduler @pytest.fixture def swh_sched_config(swh_scheduler_config): return { "scheduler": {"cls": "local", "args": swh_scheduler_config,}, "elasticsearch": { "cls": "memory", "args": {"index_name_prefix": "swh-tasks",}, }, } @pytest.fixture def swh_sched_config_file(swh_sched_config, monkeypatch, tmp_path): conffile = str(tmp_path / "elastic.yml") with open(conffile, "w") as f: f.write(yaml.dump(swh_sched_config)) monkeypatch.setenv("SWH_CONFIG_FILENAME", conffile) return conffile @pytest.fixture def swh_sched(swh_sched_config): return get_scheduler(**swh_sched_config["scheduler"]) @pytest.fixture def swh_elasticsearch_backend(swh_sched_config): from swh.scheduler.backend_es import ElasticSearchBackend backend = ElasticSearchBackend(**swh_sched_config) backend.initialize() return backend @pytest.fixture def swh_elasticsearch_memory(swh_elasticsearch_backend): return swh_elasticsearch_backend.storage diff --git a/swh/scheduler/tests/es/test_backend_es.py b/swh/scheduler/tests/es/test_backend_es.py index 6b332ad..1c7e2c0 100644 --- a/swh/scheduler/tests/es/test_backend_es.py +++ b/swh/scheduler/tests/es/test_backend_es.py @@ -1,80 +1,79 @@ # Copyright (C) 2019-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime -import pytest - import elasticsearch +import pytest from swh.scheduler.backend_es import get_elasticsearch -from ..common import tasks_from_template, TEMPLATES +from ..common import TEMPLATES, tasks_from_template def test_get_elasticsearch(): with pytest.raises(ValueError, match="Unknown elasticsearch class"): get_elasticsearch("unknown") es = get_elasticsearch("memory") assert es from swh.scheduler.elasticsearch_memory import MemoryElasticsearch assert isinstance(es, MemoryElasticsearch) es = get_elasticsearch("local") assert es assert isinstance(es, elasticsearch.Elasticsearch) def test_backend_setup_basic(swh_elasticsearch_backend): """Elastic search instance should allow to create/close/check index """ index_name = "swh-tasks-2010-01" try: swh_elasticsearch_backend.storage.indices.get_mapping(index_name) except (elasticsearch.exceptions.NotFoundError, KeyError): pass assert not swh_elasticsearch_backend.storage.indices.exists(index_name) swh_elasticsearch_backend.create(index_name) assert swh_elasticsearch_backend.storage.indices.exists(index_name) assert swh_elasticsearch_backend.is_index_opened(index_name) # index exists with a mapping mapping = swh_elasticsearch_backend.storage.indices.get_mapping(index_name) assert mapping != {} def test_backend_setup_index(swh_elasticsearch_backend): """Elastic search instance should allow to bulk index """ template_git = TEMPLATES["git"] next_run_date = datetime.datetime.utcnow() - datetime.timedelta(days=1) tasks = tasks_from_template(template_git, next_run_date, 1) index_name = swh_elasticsearch_backend.compute_index_name( next_run_date.year, next_run_date.month ) assert not swh_elasticsearch_backend.storage.indices.exists(index_name) tasks = list(swh_elasticsearch_backend.streaming_bulk(index_name, tasks)) assert len(tasks) > 0 for output_task in tasks: assert output_task is not None assert output_task["type"] == template_git["type"] assert output_task["arguments"] is not None next_run = output_task["next_run"] if isinstance(next_run, str): # real elasticsearch assert next_run == next_run_date.isoformat() else: # memory implem. does not really index assert next_run == next_run_date assert swh_elasticsearch_backend.storage.indices.exists(index_name) assert swh_elasticsearch_backend.is_index_opened(index_name) mapping = swh_elasticsearch_backend.storage.indices.get_mapping(index_name) assert mapping != {} diff --git a/swh/scheduler/tests/es/test_cli_task.py b/swh/scheduler/tests/es/test_cli_task.py index ff8ba46..d0182fd 100644 --- a/swh/scheduler/tests/es/test_cli_task.py +++ b/swh/scheduler/tests/es/test_cli_task.py @@ -1,114 +1,111 @@ # Copyright (C) 2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -import arrow import datetime import logging -import uuid import random +import uuid -import pytest - +import arrow from click.testing import CliRunner +import pytest from swh.scheduler.cli import cli - -from ..common import tasks_from_template, TASK_TYPES, TEMPLATES - +from ..common import TASK_TYPES, TEMPLATES, tasks_from_template logger = logging.getLogger(__name__) @pytest.mark.usefixtures("swh_elasticsearch_backend") def test_cli_archive_tasks(swh_sched, swh_sched_config_file): scheduler = swh_sched template_git = TEMPLATES["git"] template_hg = TEMPLATES["hg"] # first initialize scheduler's db (is this still needed?) for tt in TASK_TYPES.values(): scheduler.create_task_type(tt) next_run_start = arrow.utcnow().datetime - datetime.timedelta(days=1) recurring = tasks_from_template(template_git, next_run_start, 100) oneshots = tasks_from_template( template_hg, next_run_start - datetime.timedelta(days=1), 50 ) past_time = next_run_start - datetime.timedelta(days=7) all_tasks = recurring + oneshots result = scheduler.create_tasks(all_tasks) assert len(result) == len(all_tasks) # simulate task run backend_tasks = [ { "task": task["id"], "backend_id": str(uuid.uuid4()), "scheduled": next_run_start - datetime.timedelta(minutes=i % 60), } for i, task in enumerate(result) ] scheduler.mass_schedule_task_runs(backend_tasks) # Disable some tasks tasks_to_disable = set() for task in result: status = random.choice(["disabled", "completed"]) if status == "disabled": tasks_to_disable.add(task["id"]) scheduler.disable_tasks(tasks_to_disable) git_tasks = scheduler.search_tasks(task_type=template_git["type"]) hg_tasks = scheduler.search_tasks(task_type=template_hg["type"]) assert len(git_tasks) + len(hg_tasks) == len(all_tasks) # Ensure the task_run are in expected state task_runs = scheduler.get_task_runs([t["id"] for t in git_tasks + hg_tasks]) # Same for the tasks for t in git_tasks + hg_tasks: if t["id"] in tasks_to_disable: assert t["status"] == "disabled" future_time = next_run_start + datetime.timedelta(days=1) for tr in task_runs: assert past_time <= tr["scheduled"] assert tr["scheduled"] < future_time runner = CliRunner() result = runner.invoke( cli, [ "--config-file", swh_sched_config_file, "task", "archive", "--after", past_time.isoformat(), "--before", future_time.isoformat(), "--cleanup", ], obj={"log_level": logging.DEBUG,}, ) assert result.exit_code == 0, result.output # disabled tasks should no longer be in the scheduler git_tasks = scheduler.search_tasks(task_type=template_git["type"]) hg_tasks = scheduler.search_tasks(task_type=template_hg["type"]) remaining_tasks = git_tasks + hg_tasks count_disabled = 0 for task in remaining_tasks: logger.debug(f"task status: {task['status']}") if task["status"] == "disabled": count_disabled += 1 assert count_disabled == 0 assert len(remaining_tasks) == len(all_tasks) - len(tasks_to_disable) diff --git a/swh/scheduler/tests/es/test_elasticsearch_memory.py b/swh/scheduler/tests/es/test_elasticsearch_memory.py index ada5ef0..5c79622 100644 --- a/swh/scheduler/tests/es/test_elasticsearch_memory.py +++ b/swh/scheduler/tests/es/test_elasticsearch_memory.py @@ -1,151 +1,150 @@ # Copyright (C) 2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime import hashlib import logging import random +from typing import Any, Dict import pytest from swh.scheduler.elasticsearch_memory import BasicSerializer, BasicTransport -from ..common import tasks_from_template, TEMPLATES -from typing import Any, Dict - +from ..common import TEMPLATES, tasks_from_template logger = logging.getLogger(__name__) def test_serializer(): s = BasicSerializer() assert s data = {"something": [1, 2, 3], "cool": {"1": "2"}} actual_data = s.dumps(data) assert isinstance(actual_data, str) assert actual_data == str(data) def test_basic_transport(): b = BasicTransport() assert b assert isinstance(b.serializer, BasicSerializer) def test_index_manipulation(swh_elasticsearch_memory): index_name = "swh-tasks-xxxx" indices = swh_elasticsearch_memory.index assert not swh_elasticsearch_memory.exists(index_name) assert index_name not in indices # so stat raises with pytest.raises(Exception): swh_elasticsearch_memory.stats(index_name) # we create the index swh_elasticsearch_memory.create(index_name) # now the index exists assert swh_elasticsearch_memory.exists(index_name) assert index_name in indices # it's opened assert indices[index_name]["status"] == "opened" # so stats is happy swh_elasticsearch_memory.stats(index_name) # open the index, nothing changes swh_elasticsearch_memory.open(index_name) assert indices[index_name]["status"] == "opened" # close the index swh_elasticsearch_memory.close(index_name) assert indices[index_name]["status"] == "closed" # reopen the index (fun times) swh_elasticsearch_memory.open(index_name) assert indices[index_name]["status"] == "opened" def test_bulk_and_mget(swh_elasticsearch_memory): # initialize tasks template_git = TEMPLATES["git"] next_run_start = datetime.datetime.utcnow() - datetime.timedelta(days=1) tasks = tasks_from_template(template_git, next_run_start, 100) def compute_id(stask): return hashlib.sha1(stask.encode("utf-8")).hexdigest() body = [] ids_to_task = {} for task in tasks: date = task["next_run"] index_name = f"swh-tasks-{date.year}-{date.month}" idx = {"index": {"_index": index_name}} sidx = swh_elasticsearch_memory.transport.serializer.dumps(idx) body.append(sidx) stask = swh_elasticsearch_memory.transport.serializer.dumps(task) body.append(stask) _id = compute_id(stask) ids_to_task[_id] = task logger.debug(f"_id: {_id}, task: {task}") # store # create the index first swh_elasticsearch_memory.create(index_name) # then bulk insert new data result = swh_elasticsearch_memory.bulk("\n".join(body)) # no guarantee in the order assert result actual_items = result["items"] assert len(actual_items) == len(ids_to_task) def get_id(data: Dict[str, Any]) -> str: return data["index"]["_id"] actual_items = sorted(actual_items, key=get_id) expected_items = { "items": [{"index": {"status": 200, "_id": _id}} for _id in list(ids_to_task)] } expected_items = sorted(expected_items["items"], key=get_id) assert actual_items == expected_items # retrieve nb_docs = 10 ids = list(ids_to_task) random_ids = [] # add some inexistent ids for i in range(16): noisy_id = f"{i}" * 40 random_ids.append(noisy_id) random_ids.extend(random.sample(ids, nb_docs)) # add relevant ids for i in range(16, 32): noisy_id = f"{i}" * 40 random_ids.append(noisy_id) result = swh_elasticsearch_memory.mget(index=index_name, body={"ids": random_ids}) assert result["docs"] assert len(result["docs"]) == nb_docs, "no random and inexistent id found" for doc in result["docs"]: assert doc["found"] actual_task = doc["_source"] _id = compute_id(str(actual_task)) expected_task = ids_to_task[_id] assert actual_task == expected_task diff --git a/swh/scheduler/tests/test_api_client.py b/swh/scheduler/tests/test_api_client.py index d931ac0..eae6b0c 100644 --- a/swh/scheduler/tests/test_api_client.py +++ b/swh/scheduler/tests/test_api_client.py @@ -1,75 +1,75 @@ # Copyright (C) 2018 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -import pytest from flask import url_for +import pytest -import swh.scheduler.api.server as server from swh.scheduler.api.client import RemoteScheduler +import swh.scheduler.api.server as server from swh.scheduler.tests.test_scheduler import TestScheduler # noqa # tests are executed using imported class (TestScheduler) using overloaded # swh_scheduler fixture below # the Flask app used as server in these tests @pytest.fixture def app(swh_db_scheduler): assert hasattr(server, "scheduler") server.scheduler = swh_db_scheduler yield server.app # the RPCClient class used as client used in these tests @pytest.fixture def swh_rpc_client_class(): return RemoteScheduler @pytest.fixture def swh_scheduler(swh_rpc_client, app): yield swh_rpc_client def test_site_map(flask_app_client): sitemap = flask_app_client.get(url_for("site_map")) assert sitemap.headers["Content-Type"] == "application/json" rules = set(x["rule"] for x in sitemap.json) # we expect at least these rules expected_rules = set( "/" + rule for rule in ( "lister/get_or_create", "lister/update", "origins/get", "origins/record", "priority_ratios/get", "task/create", "task/delete_archived", "task/disable", "task/filter_for_archive", "task/get", "task/grab_ready", "task/peek_ready", "task/search", "task/set_status", "task_run/end", "task_run/get", "task_run/schedule", "task_run/schedule_one", "task_run/start", "task_type/create", "task_type/get", "task_type/get_all", ) ) assert rules == expected_rules def test_root(flask_app_client): root = flask_app_client.get("/") assert root.status_code == 200 assert b"Software Heritage scheduler RPC server" in root.data diff --git a/swh/scheduler/tests/test_celery_tasks.py b/swh/scheduler/tests/test_celery_tasks.py index b61cea1..ab8be43 100644 --- a/swh/scheduler/tests/test_celery_tasks.py +++ b/swh/scheduler/tests/test_celery_tasks.py @@ -1,176 +1,174 @@ -from time import sleep from itertools import count +from time import sleep -from celery.result import GroupResult -from celery.result import AsyncResult - +from celery.result import AsyncResult, GroupResult import pytest -from swh.scheduler.utils import create_task_dict from swh.scheduler.celery_backend.runner import run_ready_tasks +from swh.scheduler.utils import create_task_dict def test_ping(swh_scheduler_celery_app, swh_scheduler_celery_worker): res = swh_scheduler_celery_app.send_task("swh.scheduler.tests.tasks.ping") assert res res.wait() assert res.successful() assert res.result == "OK" def test_ping_with_kw(swh_scheduler_celery_app, swh_scheduler_celery_worker): res = swh_scheduler_celery_app.send_task( "swh.scheduler.tests.tasks.ping", kwargs={"a": 1} ) assert res res.wait() assert res.successful() assert res.result == "OK (kw={'a': 1})" def test_multiping(swh_scheduler_celery_app, swh_scheduler_celery_worker): "Test that a task that spawns subtasks (group) works" res = swh_scheduler_celery_app.send_task( "swh.scheduler.tests.tasks.multiping", kwargs={"n": 5} ) assert res res.wait() assert res.successful() # retrieve the GroupResult for this task and wait for all the subtasks # to complete promise_id = res.result assert promise_id promise = GroupResult.restore(promise_id, app=swh_scheduler_celery_app) for i in range(5): if promise.ready(): break sleep(1) results = [x.get() for x in promise.results] assert len(results) == 5 for i in range(5): assert ("OK (kw={'i': %s})" % i) in results def test_scheduler_fixture( swh_scheduler_celery_app, swh_scheduler_celery_worker, swh_scheduler ): "Test that the scheduler fixture works properly" task_type = swh_scheduler.get_task_type("swh-test-ping") assert task_type assert task_type["backend_name"] == "swh.scheduler.tests.tasks.ping" swh_scheduler.create_tasks([create_task_dict("swh-test-ping", "oneshot")]) backend_tasks = run_ready_tasks(swh_scheduler, swh_scheduler_celery_app) assert backend_tasks for task in backend_tasks: # Make sure the task completed AsyncResult(id=task["backend_id"]).get() def test_task_return_value( swh_scheduler_celery_app, swh_scheduler_celery_worker, swh_scheduler ): task_type = swh_scheduler.get_task_type("swh-test-add") assert task_type assert task_type["backend_name"] == "swh.scheduler.tests.tasks.add" swh_scheduler.create_tasks([create_task_dict("swh-test-add", "oneshot", 12, 30)]) backend_tasks = run_ready_tasks(swh_scheduler, swh_scheduler_celery_app) assert len(backend_tasks) == 1 task = backend_tasks[0] value = AsyncResult(id=task["backend_id"]).get() assert value == 42 def test_task_exception( swh_scheduler_celery_app, swh_scheduler_celery_worker, swh_scheduler ): task_type = swh_scheduler.get_task_type("swh-test-error") assert task_type assert task_type["backend_name"] == "swh.scheduler.tests.tasks.error" swh_scheduler.create_tasks([create_task_dict("swh-test-error", "oneshot")]) backend_tasks = run_ready_tasks(swh_scheduler, swh_scheduler_celery_app) assert len(backend_tasks) == 1 task = backend_tasks[0] result = AsyncResult(id=task["backend_id"]) with pytest.raises(NotImplementedError): result.get() def test_statsd(swh_scheduler_celery_app, swh_scheduler_celery_worker, mocker): m = mocker.patch("swh.scheduler.task.Statsd._send_to_server") mocker.patch("swh.scheduler.task.ts", side_effect=count()) mocker.patch("swh.core.statsd.monotonic", side_effect=count()) res = swh_scheduler_celery_app.send_task("swh.scheduler.tests.tasks.echo") assert res res.wait() assert res.successful() assert res.result == {} m.assert_any_call( "swh_task_called_count:1|c|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) m.assert_any_call( "swh_task_start_ts:0|g|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) m.assert_any_call( "swh_task_end_ts:1|g|" "#status:uneventful,task:swh.scheduler.tests.tasks.echo," "worker:unknown worker" ) m.assert_any_call( "swh_task_duration_seconds:1000|ms|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) m.assert_any_call( "swh_task_success_count:1|c|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) def test_statsd_with_status( swh_scheduler_celery_app, swh_scheduler_celery_worker, mocker ): m = mocker.patch("swh.scheduler.task.Statsd._send_to_server") mocker.patch("swh.scheduler.task.ts", side_effect=count()) mocker.patch("swh.core.statsd.monotonic", side_effect=count()) res = swh_scheduler_celery_app.send_task( "swh.scheduler.tests.tasks.echo", kwargs={"status": "eventful"} ) assert res res.wait() assert res.successful() assert res.result == {"status": "eventful"} m.assert_any_call( "swh_task_called_count:1|c|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) m.assert_any_call( "swh_task_start_ts:0|g|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) m.assert_any_call( "swh_task_end_ts:1|g|" "#status:eventful,task:swh.scheduler.tests.tasks.echo," "worker:unknown worker" ) m.assert_any_call( "swh_task_duration_seconds:1000|ms|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) m.assert_any_call( "swh_task_success_count:1|c|" "#task:swh.scheduler.tests.tasks.echo,worker:unknown worker" ) diff --git a/swh/scheduler/tests/test_cli.py b/swh/scheduler/tests/test_cli.py index d124dfa..4619119 100644 --- a/swh/scheduler/tests/test_cli.py +++ b/swh/scheduler/tests/test_cli.py @@ -1,797 +1,795 @@ # Copyright (C) 2019-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime +from itertools import islice +import logging import re import tempfile -import logging - -from itertools import islice - from unittest.mock import patch -from click.testing import CliRunner +from click.testing import CliRunner import pytest +from swh.core.api.classes import stream_results from swh.model.model import Origin -from swh.storage import get_storage from swh.scheduler.cli import cli from swh.scheduler.utils import create_task_dict -from swh.core.api.classes import stream_results +from swh.storage import get_storage CLI_CONFIG = """ scheduler: cls: foo args: {} """ def invoke(scheduler, catch_exceptions, args): runner = CliRunner() with patch( "swh.scheduler.get_scheduler" ) as get_scheduler_mock, tempfile.NamedTemporaryFile( "a", suffix=".yml" ) as config_fd: config_fd.write(CLI_CONFIG) config_fd.seek(0) get_scheduler_mock.return_value = scheduler args = ["-C" + config_fd.name,] + args result = runner.invoke(cli, args, obj={"log_level": logging.WARNING}) if not catch_exceptions and result.exception: print(result.output) raise result.exception return result def test_schedule_tasks(swh_scheduler): csv_data = ( b'swh-test-ping;[["arg1", "arg2"]];{"key": "value"};' + datetime.datetime.utcnow().isoformat().encode() + b"\n" + b'swh-test-ping;[["arg3", "arg4"]];{"key": "value"};' + datetime.datetime.utcnow().isoformat().encode() + b"\n" ) with tempfile.NamedTemporaryFile(suffix=".csv") as csv_fd: csv_fd.write(csv_data) csv_fd.seek(0) result = invoke( swh_scheduler, False, ["task", "schedule", "-d", ";", csv_fd.name] ) expected = r""" Created 2 tasks Task 1 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: recurring Args: \['arg1', 'arg2'\] Keyword args: key: 'value' Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: recurring Args: \['arg3', 'arg4'\] Keyword args: key: 'value' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_schedule_tasks_columns(swh_scheduler): with tempfile.NamedTemporaryFile(suffix=".csv") as csv_fd: csv_fd.write(b'swh-test-ping;oneshot;["arg1", "arg2"];{"key": "value"}\n') csv_fd.seek(0) result = invoke( swh_scheduler, False, [ "task", "schedule", "-c", "type", "-c", "policy", "-c", "args", "-c", "kwargs", "-d", ";", csv_fd.name, ], ) expected = r""" Created 1 tasks Task 1 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Args: 'arg1' 'arg2' Keyword args: key: 'value' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_schedule_task(swh_scheduler): result = invoke( swh_scheduler, False, ["task", "add", "swh-test-ping", "arg1", "arg2", "key=value",], ) expected = r""" Created 1 tasks Task 1 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: recurring Args: 'arg1' 'arg2' Keyword args: key: 'value' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_pending_tasks_none(swh_scheduler): result = invoke(swh_scheduler, False, ["task", "list-pending", "swh-test-ping",]) expected = r""" Found 0 swh-test-ping tasks """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_pending_tasks(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task2["next_run"] += datetime.timedelta(days=1) swh_scheduler.create_tasks([task1, task2]) result = invoke(swh_scheduler, False, ["task", "list-pending", "swh-test-ping",]) expected = r""" Found 1 swh-test-ping tasks Task 1 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Args: Keyword args: key: 'value1' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output swh_scheduler.grab_ready_tasks("swh-test-ping") result = invoke(swh_scheduler, False, ["task", "list-pending", "swh-test-ping",]) expected = r""" Found 0 swh-test-ping tasks """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_pending_tasks_filter(swh_scheduler): task = create_task_dict("swh-test-multiping", "oneshot", key="value") swh_scheduler.create_tasks([task]) result = invoke(swh_scheduler, False, ["task", "list-pending", "swh-test-ping",]) expected = r""" Found 0 swh-test-ping tasks """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_pending_tasks_filter_2(swh_scheduler): swh_scheduler.create_tasks( [ create_task_dict("swh-test-multiping", "oneshot", key="value"), create_task_dict("swh-test-ping", "oneshot", key="value2"), ] ) result = invoke(swh_scheduler, False, ["task", "list-pending", "swh-test-ping",]) expected = r""" Found 1 swh-test-ping tasks Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Args: Keyword args: key: 'value2' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output # Fails because "task list-pending --limit 3" only returns 2 tasks, because # of how compute_nb_tasks_from works. @pytest.mark.xfail def test_list_pending_tasks_limit(swh_scheduler): swh_scheduler.create_tasks( [ create_task_dict("swh-test-ping", "oneshot", key="value%d" % i) for i in range(10) ] ) result = invoke( swh_scheduler, False, ["task", "list-pending", "swh-test-ping", "--limit", "3",] ) expected = r""" Found 2 swh-test-ping tasks Task 1 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Args: Keyword args: key: 'value0' Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Args: Keyword args: key: 'value1' Task 3 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Args: Keyword args: key: 'value2' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_pending_tasks_before(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task1["next_run"] += datetime.timedelta(days=3) task2["next_run"] += datetime.timedelta(days=1) swh_scheduler.create_tasks([task1, task2]) result = invoke( swh_scheduler, False, [ "task", "list-pending", "swh-test-ping", "--before", (datetime.date.today() + datetime.timedelta(days=2)).isoformat(), ], ) expected = r""" Found 1 swh-test-ping tasks Task 2 Next run: in a day \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Args: Keyword args: key: 'value2' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_tasks(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task1["next_run"] += datetime.timedelta(days=3, hours=2) swh_scheduler.create_tasks([task1, task2]) swh_scheduler.grab_ready_tasks("swh-test-ping") result = invoke(swh_scheduler, False, ["task", "list",]) expected = r""" Found 2 tasks Task 1 Next run: in 3 days \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value1' Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_scheduled Priority:\x20 Args: Keyword args: key: 'value2' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_tasks_id(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task3 = create_task_dict("swh-test-ping", "oneshot", key="value3") swh_scheduler.create_tasks([task1, task2, task3]) result = invoke(swh_scheduler, False, ["task", "list", "--task-id", "2",]) expected = r""" Found 1 tasks Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value2' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_tasks_id_2(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task3 = create_task_dict("swh-test-ping", "oneshot", key="value3") swh_scheduler.create_tasks([task1, task2, task3]) result = invoke( swh_scheduler, False, ["task", "list", "--task-id", "2", "--task-id", "3"] ) expected = r""" Found 2 tasks Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value2' Task 3 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value3' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_tasks_type(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-multiping", "oneshot", key="value2") task3 = create_task_dict("swh-test-ping", "oneshot", key="value3") swh_scheduler.create_tasks([task1, task2, task3]) result = invoke( swh_scheduler, False, ["task", "list", "--task-type", "swh-test-ping"] ) expected = r""" Found 2 tasks Task 1 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value1' Task 3 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value3' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_tasks_limit(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task3 = create_task_dict("swh-test-ping", "oneshot", key="value3") swh_scheduler.create_tasks([task1, task2, task3]) result = invoke(swh_scheduler, False, ["task", "list", "--limit", "2",]) expected = r""" Found 2 tasks Task 1 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value1' Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value2' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_tasks_before(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task1["next_run"] += datetime.timedelta(days=3, hours=2) swh_scheduler.create_tasks([task1, task2]) swh_scheduler.grab_ready_tasks("swh-test-ping") result = invoke( swh_scheduler, False, [ "task", "list", "--before", (datetime.date.today() + datetime.timedelta(days=2)).isoformat(), ], ) expected = r""" Found 1 tasks Task 2 Next run: just now \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_scheduled Priority:\x20 Args: Keyword args: key: 'value2' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def test_list_tasks_after(swh_scheduler): task1 = create_task_dict("swh-test-ping", "oneshot", key="value1") task2 = create_task_dict("swh-test-ping", "oneshot", key="value2") task1["next_run"] += datetime.timedelta(days=3, hours=2) swh_scheduler.create_tasks([task1, task2]) swh_scheduler.grab_ready_tasks("swh-test-ping") result = invoke( swh_scheduler, False, [ "task", "list", "--after", (datetime.date.today() + datetime.timedelta(days=2)).isoformat(), ], ) expected = r""" Found 1 tasks Task 1 Next run: in 3 days \(.*\) Interval: 1 day, 0:00:00 Type: swh-test-ping Policy: oneshot Status: next_run_not_scheduled Priority:\x20 Args: Keyword args: key: 'value1' """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), result.output def _fill_storage_with_origins(storage, nb_origins): origins = [Origin(url=f"http://example.com/{i}") for i in range(nb_origins)] storage.origin_add(origins) return origins @pytest.fixture def storage(): """An instance of in-memory storage that gets injected into the CLI functions.""" storage = get_storage(cls="memory") with patch("swh.storage.get_storage") as get_storage_mock: get_storage_mock.return_value = storage yield storage @patch("swh.scheduler.cli.utils.TASK_BATCH_SIZE", 3) def test_task_schedule_origins_dry_run(swh_scheduler, storage): """Tests the scheduling when origin_batch_size*task_batch_size is a divisor of nb_origins.""" _fill_storage_with_origins(storage, 90) result = invoke( swh_scheduler, False, ["task", "schedule_origins", "--dry-run", "swh-test-ping",], ) # Check the output expected = r""" Scheduled 3 tasks \(30 origins\). Scheduled 6 tasks \(60 origins\). Scheduled 9 tasks \(90 origins\). Done. """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), repr(result.output) # Check scheduled tasks tasks = swh_scheduler.search_tasks() assert len(tasks) == 0 def _assert_origin_tasks_contraints(tasks, max_tasks, max_task_size, expected_origins): # check there are not too many tasks assert len(tasks) <= max_tasks # check tasks are not too large assert all(len(task["arguments"]["args"][0]) <= max_task_size for task in tasks) # check the tasks are exhaustive assert sum([len(task["arguments"]["args"][0]) for task in tasks]) == len( expected_origins ) assert set.union(*(set(task["arguments"]["args"][0]) for task in tasks)) == { origin.url for origin in expected_origins } @patch("swh.scheduler.cli.utils.TASK_BATCH_SIZE", 3) def test_task_schedule_origins(swh_scheduler, storage): """Tests the scheduling when neither origin_batch_size or task_batch_size is a divisor of nb_origins.""" origins = _fill_storage_with_origins(storage, 70) result = invoke( swh_scheduler, False, ["task", "schedule_origins", "swh-test-ping", "--batch-size", "20",], ) # Check the output expected = r""" Scheduled 3 tasks \(60 origins\). Scheduled 4 tasks \(70 origins\). Done. """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), repr(result.output) # Check tasks tasks = swh_scheduler.search_tasks() _assert_origin_tasks_contraints(tasks, 4, 20, origins) assert all(task["arguments"]["kwargs"] == {} for task in tasks) def test_task_schedule_origins_kwargs(swh_scheduler, storage): """Tests support of extra keyword-arguments.""" origins = _fill_storage_with_origins(storage, 30) result = invoke( swh_scheduler, False, [ "task", "schedule_origins", "swh-test-ping", "--batch-size", "20", 'key1="value1"', 'key2="value2"', ], ) # Check the output expected = r""" Scheduled 2 tasks \(30 origins\). Done. """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), repr(result.output) # Check tasks tasks = swh_scheduler.search_tasks() _assert_origin_tasks_contraints(tasks, 2, 20, origins) assert all( task["arguments"]["kwargs"] == {"key1": "value1", "key2": "value2"} for task in tasks ) def test_task_schedule_origins_with_limit(swh_scheduler, storage): """Tests support of extra keyword-arguments.""" _fill_storage_with_origins(storage, 50) limit = 20 expected_origins = list(islice(stream_results(storage.origin_list), limit)) nb_origins = len(expected_origins) assert nb_origins == limit max_task_size = 5 nb_tasks, remainder = divmod(nb_origins, max_task_size) assert remainder == 0 # made the numbers go round result = invoke( swh_scheduler, False, [ "task", "schedule_origins", "swh-test-ping", "--batch-size", max_task_size, "--limit", limit, ], ) # Check the output expected = rf""" Scheduled {nb_tasks} tasks \({nb_origins} origins\). Done. """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), repr(result.output) tasks = swh_scheduler.search_tasks() _assert_origin_tasks_contraints(tasks, max_task_size, nb_origins, expected_origins) def test_task_schedule_origins_with_page_token(swh_scheduler, storage): """Tests support of extra keyword-arguments.""" nb_total_origins = 50 origins = _fill_storage_with_origins(storage, nb_total_origins) # prepare page_token and origins result expectancy page_result = storage.origin_list(limit=10) assert len(page_result.results) == 10 page_token = page_result.next_page_token assert page_token is not None # remove the first 10 origins listed as we won't see those in tasks expected_origins = [o for o in origins if o not in page_result.results] nb_origins = len(expected_origins) assert nb_origins == nb_total_origins - len(page_result.results) max_task_size = 10 nb_tasks, remainder = divmod(nb_origins, max_task_size) assert remainder == 0 result = invoke( swh_scheduler, False, [ "task", "schedule_origins", "swh-test-ping", "--batch-size", max_task_size, "--page-token", page_token, ], ) # Check the output expected = rf""" Scheduled {nb_tasks} tasks \({nb_origins} origins\). Done. """.lstrip() assert result.exit_code == 0, result.output assert re.fullmatch(expected, result.output, re.MULTILINE), repr(result.output) # Check tasks tasks = swh_scheduler.search_tasks() _assert_origin_tasks_contraints(tasks, max_task_size, nb_origins, expected_origins) diff --git a/swh/scheduler/tests/test_cli_task_type.py b/swh/scheduler/tests/test_cli_task_type.py index c158760..adf2ebe 100644 --- a/swh/scheduler/tests/test_cli_task_type.py +++ b/swh/scheduler/tests/test_cli_task_type.py @@ -1,128 +1,127 @@ # Copyright (C) 2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -import pytest import traceback -import yaml -import pkg_resources from click.testing import CliRunner +import pkg_resources +import pytest +import yaml from swh.scheduler import get_scheduler from swh.scheduler.cli import cli - FAKE_MODULE_ENTRY_POINTS = { "lister.gnu=swh.lister.gnu:register", "lister.pypi=swh.lister.pypi:register", } @pytest.fixture def mock_pkg_resources(monkeypatch): """Monkey patch swh.scheduler's mock_pkg_resources.iter_entry_point call """ def fake_iter_entry_points(*args, **kwargs): """Substitute fake function to return a fixed set of entrypoints """ - from pkg_resources import EntryPoint, Distribution + from pkg_resources import Distribution, EntryPoint d = Distribution() return [EntryPoint.parse(entry, dist=d) for entry in FAKE_MODULE_ENTRY_POINTS] original_method = pkg_resources.iter_entry_points monkeypatch.setattr(pkg_resources, "iter_entry_points", fake_iter_entry_points) yield # reset monkeypatch: is that needed? monkeypatch.setattr(pkg_resources, "iter_entry_points", original_method) @pytest.fixture def local_sched_config(swh_scheduler_config): """Expose the local scheduler configuration """ return {"scheduler": {"cls": "local", "args": swh_scheduler_config}} @pytest.fixture def local_sched_configfile(local_sched_config, tmp_path): """Write in temporary location the local scheduler configuration """ configfile = tmp_path / "config.yml" configfile.write_text(yaml.dump(local_sched_config)) return configfile.as_posix() def test_register_ttypes_all( mock_pkg_resources, local_sched_config, local_sched_configfile ): """Registering all task types""" for command in [ ["--config-file", local_sched_configfile, "task-type", "register"], ["--config-file", local_sched_configfile, "task-type", "register", "-p", "all"], [ "--config-file", local_sched_configfile, "task-type", "register", "-p", "lister.gnu", "-p", "lister.pypi", ], ]: result = CliRunner().invoke(cli, command) assert result.exit_code == 0, traceback.print_exception(*result.exc_info) scheduler = get_scheduler(**local_sched_config["scheduler"]) all_tasks = [ "list-gnu-full", "list-pypi", ] for task in all_tasks: task_type_desc = scheduler.get_task_type(task) assert task_type_desc assert task_type_desc["type"] == task assert task_type_desc["backoff_factor"] == 1 def test_register_ttypes_filter( mock_pkg_resources, local_sched_config, local_sched_configfile ): """Filtering on one worker should only register its associated task type """ result = CliRunner().invoke( cli, [ "--config-file", local_sched_configfile, "task-type", "register", "--plugins", "lister.gnu", ], ) assert result.exit_code == 0, traceback.print_exception(*result.exc_info) scheduler = get_scheduler(**local_sched_config["scheduler"]) all_tasks = [ "list-gnu-full", ] for task in all_tasks: task_type_desc = scheduler.get_task_type(task) assert task_type_desc assert task_type_desc["type"] == task assert task_type_desc["backoff_factor"] == 1 diff --git a/swh/scheduler/tests/test_common.py b/swh/scheduler/tests/test_common.py index b2bc8d4..7a0de09 100644 --- a/swh/scheduler/tests/test_common.py +++ b/swh/scheduler/tests/test_common.py @@ -1,65 +1,65 @@ # Copyright (C) 2017-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime -from .common import tasks_from_template, TEMPLATES +from .common import TEMPLATES, tasks_from_template def test_tasks_from_template_no_priority(): nb_tasks = 3 template = TEMPLATES["git"] next_run = datetime.datetime.utcnow() tasks = tasks_from_template(template, next_run, nb_tasks) assert len(tasks) == nb_tasks for i, t in enumerate(tasks): assert t["type"] == template["type"] assert t["arguments"] is not None assert t.get("policy") is None # not defined in template assert len(t["arguments"]["args"]) == 1 assert len(t["arguments"]["kwargs"].keys()) == 1 assert t["next_run"] == next_run - datetime.timedelta(microseconds=i) assert t.get("priority") is None def test_tasks_from_template_priority(): nb_tasks_no_priority = 3 nb_tasks_priority = 10 template = TEMPLATES["hg"] priorities = { "high": 0.5, "normal": 0.3, "low": 0.2, } next_run = datetime.datetime.utcnow() tasks = tasks_from_template( template, next_run, nb_tasks_no_priority, num_priority=nb_tasks_priority, priorities=priorities, ) assert len(tasks) == nb_tasks_no_priority + nb_tasks_priority repartition_priority = {k: 0 for k in priorities.keys()} for i, t in enumerate(tasks): assert t["type"] == template["type"] assert t["arguments"] is not None assert t["policy"] == template["policy"] assert len(t["arguments"]["args"]) == 1 assert len(t["arguments"]["kwargs"].keys()) == 1 assert t["next_run"] == next_run - datetime.timedelta(microseconds=i) priority = t.get("priority") if priority: assert priority in priorities repartition_priority[priority] += 1 assert repartition_priority == { k: v * nb_tasks_priority for k, v in priorities.items() } diff --git a/swh/scheduler/tests/test_scheduler.py b/swh/scheduler/tests/test_scheduler.py index 8ffa85e..bdc8e57 100644 --- a/swh/scheduler/tests/test_scheduler.py +++ b/swh/scheduler/tests/test_scheduler.py @@ -1,730 +1,729 @@ # Copyright (C) 2017-2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information +from collections import defaultdict import copy import datetime -import random -import uuid - -from collections import defaultdict import inspect +import random from typing import Any, Dict, List, Optional +import uuid from arrow import utcnow import attr import pytest from swh.scheduler.exc import StaleData from swh.scheduler.interface import SchedulerInterface from swh.scheduler.model import ListedOrigin, ListedOriginPageToken -from .common import tasks_from_template, TEMPLATES, TASK_TYPES, LISTERS +from .common import LISTERS, TASK_TYPES, TEMPLATES, tasks_from_template def subdict(d, keys=None, excl=()): if keys is None: keys = [k for k in d.keys()] return {k: d[k] for k in keys if k not in excl} class TestScheduler: def test_interface(self, swh_scheduler): """Checks all methods of SchedulerInterface are implemented by this backend, and that they have the same signature.""" # Create an instance of the protocol (which cannot be instantiated # directly, so this creates a subclass, then instantiates it) interface = type("_", (SchedulerInterface,), {})() assert "create_task_type" in dir(interface) missing_methods = [] for meth_name in dir(interface): if meth_name.startswith("_"): continue interface_meth = getattr(interface, meth_name) try: concrete_meth = getattr(swh_scheduler, meth_name) except AttributeError: if not getattr(interface_meth, "deprecated_endpoint", False): # The backend is missing a (non-deprecated) endpoint missing_methods.append(meth_name) continue expected_signature = inspect.signature(interface_meth) actual_signature = inspect.signature(concrete_meth) assert expected_signature == actual_signature, meth_name assert missing_methods == [] def test_get_priority_ratios(self, swh_scheduler): assert swh_scheduler.get_priority_ratios() == { "high": 0.5, "normal": 0.3, "low": 0.2, } def test_add_task_type(self, swh_scheduler): tt = TASK_TYPES["git"] swh_scheduler.create_task_type(tt) assert tt == swh_scheduler.get_task_type(tt["type"]) tt2 = TASK_TYPES["hg"] swh_scheduler.create_task_type(tt2) assert tt == swh_scheduler.get_task_type(tt["type"]) assert tt2 == swh_scheduler.get_task_type(tt2["type"]) def test_create_task_type_idempotence(self, swh_scheduler): tt = TASK_TYPES["git"] swh_scheduler.create_task_type(tt) swh_scheduler.create_task_type(tt) assert tt == swh_scheduler.get_task_type(tt["type"]) def test_get_task_types(self, swh_scheduler): tt, tt2 = TASK_TYPES["git"], TASK_TYPES["hg"] swh_scheduler.create_task_type(tt) swh_scheduler.create_task_type(tt2) actual_task_types = swh_scheduler.get_task_types() assert tt in actual_task_types assert tt2 in actual_task_types def test_create_tasks(self, swh_scheduler): priority_ratio = self._priority_ratio(swh_scheduler) self._create_task_types(swh_scheduler) num_tasks_priority = 100 tasks_1 = tasks_from_template(TEMPLATES["git"], utcnow(), 100) tasks_2 = tasks_from_template( TEMPLATES["hg"], utcnow(), 100, num_tasks_priority, priorities=priority_ratio, ) tasks = tasks_1 + tasks_2 # tasks are returned only once with their ids ret1 = swh_scheduler.create_tasks(tasks + tasks_1 + tasks_2) set_ret1 = set([t["id"] for t in ret1]) # creating the same set result in the same ids ret = swh_scheduler.create_tasks(tasks) set_ret = set([t["id"] for t in ret]) # Idempotence results assert set_ret == set_ret1 assert len(ret) == len(ret1) ids = set() actual_priorities = defaultdict(int) for task, orig_task in zip(ret, tasks): task = copy.deepcopy(task) task_type = TASK_TYPES[orig_task["type"].split("-")[-1]] assert task["id"] not in ids assert task["status"] == "next_run_not_scheduled" assert task["current_interval"] == task_type["default_interval"] assert task["policy"] == orig_task.get("policy", "recurring") priority = task.get("priority") if priority: actual_priorities[priority] += 1 assert task["retries_left"] == (task_type["num_retries"] or 0) ids.add(task["id"]) del task["id"] del task["status"] del task["current_interval"] del task["retries_left"] if "policy" not in orig_task: del task["policy"] if "priority" not in orig_task: del task["priority"] assert task == orig_task assert dict(actual_priorities) == { priority: int(ratio * num_tasks_priority) for priority, ratio in priority_ratio.items() } def test_peek_ready_tasks_no_priority(self, swh_scheduler): self._create_task_types(swh_scheduler) t = utcnow() task_type = TEMPLATES["git"]["type"] tasks = tasks_from_template(TEMPLATES["git"], t, 100) random.shuffle(tasks) swh_scheduler.create_tasks(tasks) ready_tasks = swh_scheduler.peek_ready_tasks(task_type) assert len(ready_tasks) == len(tasks) for i in range(len(ready_tasks) - 1): assert ready_tasks[i]["next_run"] <= ready_tasks[i + 1]["next_run"] # Only get the first few ready tasks limit = random.randrange(5, 5 + len(tasks) // 2) ready_tasks_limited = swh_scheduler.peek_ready_tasks(task_type, num_tasks=limit) assert len(ready_tasks_limited) == limit assert ready_tasks_limited == ready_tasks[:limit] # Limit by timestamp max_ts = tasks[limit - 1]["next_run"] ready_tasks_timestamped = swh_scheduler.peek_ready_tasks( task_type, timestamp=max_ts ) for ready_task in ready_tasks_timestamped: assert ready_task["next_run"] <= max_ts # Make sure we get proper behavior for the first ready tasks assert ready_tasks[: len(ready_tasks_timestamped)] == ready_tasks_timestamped # Limit by both ready_tasks_both = swh_scheduler.peek_ready_tasks( task_type, timestamp=max_ts, num_tasks=limit // 3 ) assert len(ready_tasks_both) <= limit // 3 for ready_task in ready_tasks_both: assert ready_task["next_run"] <= max_ts assert ready_task in ready_tasks[: limit // 3] def _priority_ratio(self, swh_scheduler): return swh_scheduler.get_priority_ratios() def test_peek_ready_tasks_mixed_priorities(self, swh_scheduler): priority_ratio = self._priority_ratio(swh_scheduler) self._create_task_types(swh_scheduler) t = utcnow() task_type = TEMPLATES["git"]["type"] num_tasks_priority = 100 num_tasks_no_priority = 100 # Create tasks with and without priorities tasks = tasks_from_template( TEMPLATES["git"], t, num=num_tasks_no_priority, num_priority=num_tasks_priority, priorities=priority_ratio, ) random.shuffle(tasks) swh_scheduler.create_tasks(tasks) # take all available tasks ready_tasks = swh_scheduler.peek_ready_tasks(task_type) assert len(ready_tasks) == len(tasks) assert num_tasks_priority + num_tasks_no_priority == len(ready_tasks) count_tasks_per_priority = defaultdict(int) for task in ready_tasks: priority = task.get("priority") if priority: count_tasks_per_priority[priority] += 1 assert dict(count_tasks_per_priority) == { priority: int(ratio * num_tasks_priority) for priority, ratio in priority_ratio.items() } # Only get some ready tasks num_tasks = random.randrange(5, 5 + num_tasks_no_priority // 2) num_tasks_priority = random.randrange(5, num_tasks_priority // 2) ready_tasks_limited = swh_scheduler.peek_ready_tasks( task_type, num_tasks=num_tasks, num_tasks_priority=num_tasks_priority ) count_tasks_per_priority = defaultdict(int) for task in ready_tasks_limited: priority = task.get("priority") count_tasks_per_priority[priority] += 1 import math for priority, ratio in priority_ratio.items(): expected_count = math.ceil(ratio * num_tasks_priority) actual_prio = count_tasks_per_priority[priority] assert actual_prio == expected_count or actual_prio == expected_count + 1 assert count_tasks_per_priority[None] == num_tasks def test_grab_ready_tasks(self, swh_scheduler): priority_ratio = self._priority_ratio(swh_scheduler) self._create_task_types(swh_scheduler) t = utcnow() task_type = TEMPLATES["git"]["type"] num_tasks_priority = 100 num_tasks_no_priority = 100 # Create tasks with and without priorities tasks = tasks_from_template( TEMPLATES["git"], t, num=num_tasks_no_priority, num_priority=num_tasks_priority, priorities=priority_ratio, ) random.shuffle(tasks) swh_scheduler.create_tasks(tasks) first_ready_tasks = swh_scheduler.peek_ready_tasks( task_type, num_tasks=10, num_tasks_priority=10 ) grabbed_tasks = swh_scheduler.grab_ready_tasks( task_type, num_tasks=10, num_tasks_priority=10 ) for peeked, grabbed in zip(first_ready_tasks, grabbed_tasks): assert peeked["status"] == "next_run_not_scheduled" del peeked["status"] assert grabbed["status"] == "next_run_scheduled" del grabbed["status"] assert peeked == grabbed assert peeked["priority"] == grabbed["priority"] def test_get_tasks(self, swh_scheduler): self._create_task_types(swh_scheduler) t = utcnow() tasks = tasks_from_template(TEMPLATES["git"], t, 100) tasks = swh_scheduler.create_tasks(tasks) random.shuffle(tasks) while len(tasks) > 1: length = random.randrange(1, len(tasks)) cur_tasks = sorted(tasks[:length], key=lambda x: x["id"]) tasks[:length] = [] ret = swh_scheduler.get_tasks(task["id"] for task in cur_tasks) # result is not guaranteed to be sorted ret.sort(key=lambda x: x["id"]) assert ret == cur_tasks def test_search_tasks(self, swh_scheduler): def make_real_dicts(lst): """RealDictRow is not a real dict.""" return [dict(d.items()) for d in lst] self._create_task_types(swh_scheduler) t = utcnow() tasks = tasks_from_template(TEMPLATES["git"], t, 100) tasks = swh_scheduler.create_tasks(tasks) assert make_real_dicts(swh_scheduler.search_tasks()) == make_real_dicts(tasks) def assert_filtered_task_ok( self, task: Dict[str, Any], after: datetime.datetime, before: datetime.datetime ) -> None: """Ensure filtered tasks have the right expected properties (within the range, recurring disabled, etc..) """ started = task["started"] date = started if started is not None else task["scheduled"] assert after <= date and date <= before if task["task_policy"] == "oneshot": assert task["task_status"] in ["completed", "disabled"] if task["task_policy"] == "recurring": assert task["task_status"] in ["disabled"] def test_filter_task_to_archive(self, swh_scheduler): """Filtering only list disabled recurring or completed oneshot tasks """ self._create_task_types(swh_scheduler) _time = utcnow() recurring = tasks_from_template(TEMPLATES["git"], _time, 12) oneshots = tasks_from_template(TEMPLATES["hg"], _time, 12) total_tasks = len(recurring) + len(oneshots) # simulate scheduling tasks pending_tasks = swh_scheduler.create_tasks(recurring + oneshots) backend_tasks = [ { "task": task["id"], "backend_id": str(uuid.uuid4()), "scheduled": utcnow(), } for task in pending_tasks ] swh_scheduler.mass_schedule_task_runs(backend_tasks) # we simulate the task are being done _tasks = [] for task in backend_tasks: t = swh_scheduler.end_task_run(task["backend_id"], status="eventful") _tasks.append(t) # Randomly update task's status per policy status_per_policy = {"recurring": 0, "oneshot": 0} status_choice = { # policy: [tuple (1-for-filtering, 'associated-status')] "recurring": [ (1, "disabled"), (0, "completed"), (0, "next_run_not_scheduled"), ], "oneshot": [ (0, "next_run_not_scheduled"), (1, "disabled"), (1, "completed"), ], } tasks_to_update = defaultdict(list) _task_ids = defaultdict(list) # randomize 'disabling' recurring task or 'complete' oneshot task for task in pending_tasks: policy = task["policy"] _task_ids[policy].append(task["id"]) status = random.choice(status_choice[policy]) if status[0] != 1: continue # elected for filtering status_per_policy[policy] += status[0] tasks_to_update[policy].append(task["id"]) swh_scheduler.disable_tasks(tasks_to_update["recurring"]) # hack: change the status to something else than completed/disabled swh_scheduler.set_status_tasks( _task_ids["oneshot"], status="next_run_not_scheduled" ) # complete the tasks to update swh_scheduler.set_status_tasks(tasks_to_update["oneshot"], status="completed") total_tasks_filtered = ( status_per_policy["recurring"] + status_per_policy["oneshot"] ) # no pagination scenario # retrieve tasks to archive after = _time.shift(days=-1) after_ts = after.format("YYYY-MM-DD") before = utcnow().shift(days=1) before_ts = before.format("YYYY-MM-DD") tasks_result = swh_scheduler.filter_task_to_archive( after_ts=after_ts, before_ts=before_ts, limit=total_tasks ) tasks_to_archive = tasks_result["tasks"] assert len(tasks_to_archive) == total_tasks_filtered assert tasks_result.get("next_page_token") is None actual_filtered_per_status = {"recurring": 0, "oneshot": 0} for task in tasks_to_archive: self.assert_filtered_task_ok(task, after, before) actual_filtered_per_status[task["task_policy"]] += 1 assert actual_filtered_per_status == status_per_policy # pagination scenario nb_tasks = 3 tasks_result = swh_scheduler.filter_task_to_archive( after_ts=after_ts, before_ts=before_ts, limit=nb_tasks ) tasks_to_archive2 = tasks_result["tasks"] assert len(tasks_to_archive2) == nb_tasks next_page_token = tasks_result["next_page_token"] assert next_page_token is not None all_tasks = tasks_to_archive2 while next_page_token is not None: # Retrieve paginated results tasks_result = swh_scheduler.filter_task_to_archive( after_ts=after_ts, before_ts=before_ts, limit=nb_tasks, page_token=next_page_token, ) tasks_to_archive2 = tasks_result["tasks"] assert len(tasks_to_archive2) <= nb_tasks all_tasks.extend(tasks_to_archive2) next_page_token = tasks_result.get("next_page_token") actual_filtered_per_status = {"recurring": 0, "oneshot": 0} for task in all_tasks: self.assert_filtered_task_ok(task, after, before) actual_filtered_per_status[task["task_policy"]] += 1 assert actual_filtered_per_status == status_per_policy def test_delete_archived_tasks(self, swh_scheduler): self._create_task_types(swh_scheduler) _time = utcnow() recurring = tasks_from_template(TEMPLATES["git"], _time, 12) oneshots = tasks_from_template(TEMPLATES["hg"], _time, 12) total_tasks = len(recurring) + len(oneshots) pending_tasks = swh_scheduler.create_tasks(recurring + oneshots) backend_tasks = [ { "task": task["id"], "backend_id": str(uuid.uuid4()), "scheduled": utcnow(), } for task in pending_tasks ] swh_scheduler.mass_schedule_task_runs(backend_tasks) _tasks = [] percent = random.randint(0, 100) # random election removal boundary for task in backend_tasks: t = swh_scheduler.end_task_run(task["backend_id"], status="eventful") c = random.randint(0, 100) if c <= percent: _tasks.append({"task_id": t["task"], "task_run_id": t["id"]}) swh_scheduler.delete_archived_tasks(_tasks) all_tasks = [task["id"] for task in swh_scheduler.search_tasks()] tasks_count = len(all_tasks) tasks_run_count = len(swh_scheduler.get_task_runs(all_tasks)) assert tasks_count == total_tasks - len(_tasks) assert tasks_run_count == total_tasks - len(_tasks) def test_get_task_runs_no_task(self, swh_scheduler): """No task exist in the scheduler's db, get_task_runs() should always return an empty list. """ assert not swh_scheduler.get_task_runs(task_ids=()) assert not swh_scheduler.get_task_runs(task_ids=(1, 2, 3)) assert not swh_scheduler.get_task_runs(task_ids=(1, 2, 3), limit=10) def test_get_task_runs_no_task_executed(self, swh_scheduler): """No task has been executed yet, get_task_runs() should always return an empty list. """ self._create_task_types(swh_scheduler) _time = utcnow() recurring = tasks_from_template(TEMPLATES["git"], _time, 12) oneshots = tasks_from_template(TEMPLATES["hg"], _time, 12) swh_scheduler.create_tasks(recurring + oneshots) assert not swh_scheduler.get_task_runs(task_ids=()) assert not swh_scheduler.get_task_runs(task_ids=(1, 2, 3)) assert not swh_scheduler.get_task_runs(task_ids=(1, 2, 3), limit=10) def test_get_task_runs_with_scheduled(self, swh_scheduler): """Some tasks have been scheduled but not executed yet, get_task_runs() should not return an empty list. limit should behave as expected. """ self._create_task_types(swh_scheduler) _time = utcnow() recurring = tasks_from_template(TEMPLATES["git"], _time, 12) oneshots = tasks_from_template(TEMPLATES["hg"], _time, 12) total_tasks = len(recurring) + len(oneshots) pending_tasks = swh_scheduler.create_tasks(recurring + oneshots) backend_tasks = [ { "task": task["id"], "backend_id": str(uuid.uuid4()), "scheduled": utcnow(), } for task in pending_tasks ] swh_scheduler.mass_schedule_task_runs(backend_tasks) assert not swh_scheduler.get_task_runs(task_ids=[total_tasks + 1]) btask = backend_tasks[0] runs = swh_scheduler.get_task_runs(task_ids=[btask["task"]]) assert len(runs) == 1 run = runs[0] assert subdict(run, excl=("id",)) == { "task": btask["task"], "backend_id": btask["backend_id"], "scheduled": btask["scheduled"], "started": None, "ended": None, "metadata": None, "status": "scheduled", } runs = swh_scheduler.get_task_runs( task_ids=[bt["task"] for bt in backend_tasks], limit=2 ) assert len(runs) == 2 runs = swh_scheduler.get_task_runs( task_ids=[bt["task"] for bt in backend_tasks] ) assert len(runs) == total_tasks keys = ("task", "backend_id", "scheduled") assert ( sorted([subdict(x, keys) for x in runs], key=lambda x: x["task"]) == backend_tasks ) def test_get_task_runs_with_executed(self, swh_scheduler): """Some tasks have been executed, get_task_runs() should not return an empty list. limit should behave as expected. """ self._create_task_types(swh_scheduler) _time = utcnow() recurring = tasks_from_template(TEMPLATES["git"], _time, 12) oneshots = tasks_from_template(TEMPLATES["hg"], _time, 12) pending_tasks = swh_scheduler.create_tasks(recurring + oneshots) backend_tasks = [ { "task": task["id"], "backend_id": str(uuid.uuid4()), "scheduled": utcnow(), } for task in pending_tasks ] swh_scheduler.mass_schedule_task_runs(backend_tasks) btask = backend_tasks[0] ts = utcnow() swh_scheduler.start_task_run( btask["backend_id"], metadata={"something": "stupid"}, timestamp=ts ) runs = swh_scheduler.get_task_runs(task_ids=[btask["task"]]) assert len(runs) == 1 assert subdict(runs[0], excl=("id")) == { "task": btask["task"], "backend_id": btask["backend_id"], "scheduled": btask["scheduled"], "started": ts, "ended": None, "metadata": {"something": "stupid"}, "status": "started", } ts2 = utcnow() swh_scheduler.end_task_run( btask["backend_id"], metadata={"other": "stuff"}, timestamp=ts2, status="eventful", ) runs = swh_scheduler.get_task_runs(task_ids=[btask["task"]]) assert len(runs) == 1 assert subdict(runs[0], excl=("id")) == { "task": btask["task"], "backend_id": btask["backend_id"], "scheduled": btask["scheduled"], "started": ts, "ended": ts2, "metadata": {"something": "stupid", "other": "stuff"}, "status": "eventful", } def test_get_or_create_lister(self, swh_scheduler): db_listers = [] for lister_args in LISTERS: db_listers.append(swh_scheduler.get_or_create_lister(**lister_args)) for lister, lister_args in zip(db_listers, LISTERS): assert lister.name == lister_args["name"] assert lister.instance_name == lister_args.get("instance_name", "") lister_get_again = swh_scheduler.get_or_create_lister( lister.name, lister.instance_name ) assert lister == lister_get_again def test_update_lister(self, swh_scheduler, stored_lister): lister = attr.evolve(stored_lister, current_state={"updated": "now"}) updated_lister = swh_scheduler.update_lister(lister) assert updated_lister.updated > lister.updated assert updated_lister == attr.evolve(lister, updated=updated_lister.updated) def test_update_lister_stale(self, swh_scheduler, stored_lister): swh_scheduler.update_lister(stored_lister) with pytest.raises(StaleData) as exc: swh_scheduler.update_lister(stored_lister) assert "state not updated" in exc.value.args[0] def test_record_listed_origins(self, swh_scheduler, listed_origins): ret = swh_scheduler.record_listed_origins(listed_origins) assert set(returned.url for returned in ret) == set( origin.url for origin in listed_origins ) assert all(origin.first_seen == origin.last_seen for origin in ret) def test_record_listed_origins_upsert(self, swh_scheduler, listed_origins): # First, insert `cutoff` origins cutoff = 100 assert cutoff < len(listed_origins) ret = swh_scheduler.record_listed_origins(listed_origins[:cutoff]) assert len(ret) == cutoff # Then, insert all origins, including the `cutoff` first. ret = swh_scheduler.record_listed_origins(listed_origins) assert len(ret) == len(listed_origins) # Two different "first seen" values assert len(set(origin.first_seen for origin in ret)) == 2 # But a single "last seen" value assert len(set(origin.last_seen for origin in ret)) == 1 def test_get_listed_origins_exact(self, swh_scheduler, listed_origins): swh_scheduler.record_listed_origins(listed_origins) for i, origin in enumerate(listed_origins): ret = swh_scheduler.get_listed_origins( lister_id=origin.lister_id, url=origin.url ) assert ret.next_page_token is None assert len(ret.origins) == 1 assert ret.origins[0].lister_id == origin.lister_id assert ret.origins[0].url == origin.url @pytest.mark.parametrize("num_origins,limit", [(20, 6), (5, 42), (20, 20)]) def test_get_listed_origins_limit( self, swh_scheduler, listed_origins, num_origins, limit ) -> None: added_origins = sorted( listed_origins[:num_origins], key=lambda o: (o.lister_id, o.url) ) swh_scheduler.record_listed_origins(added_origins) returned_origins: List[ListedOrigin] = [] call_count = 0 next_page_token: Optional[ListedOriginPageToken] = None while True: call_count += 1 ret = swh_scheduler.get_listed_origins( lister_id=listed_origins[0].lister_id, limit=limit, page_token=next_page_token, ) returned_origins.extend(ret.origins) next_page_token = ret.next_page_token if next_page_token is None: break assert call_count == (num_origins // limit) + 1 assert len(returned_origins) == num_origins assert [(origin.lister_id, origin.url) for origin in returned_origins] == [ (origin.lister_id, origin.url) for origin in added_origins ] def test_get_listed_origins_all(self, swh_scheduler, listed_origins) -> None: swh_scheduler.record_listed_origins(listed_origins) ret = swh_scheduler.get_listed_origins(limit=len(listed_origins) + 1) assert ret.next_page_token is None assert len(ret.origins) == len(listed_origins) def _create_task_types(self, scheduler): for tt in TASK_TYPES.values(): scheduler.create_task_type(tt) diff --git a/swh/scheduler/tests/test_server.py b/swh/scheduler/tests/test_server.py index e41c80c..c291d81 100644 --- a/swh/scheduler/tests/test_server.py +++ b/swh/scheduler/tests/test_server.py @@ -1,110 +1,111 @@ # Copyright (C) 2019 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import copy + import pytest import yaml from swh.scheduler.api.server import load_and_check_config def prepare_config_file(tmpdir, content, name="config.yml"): """Prepare configuration file in `$tmpdir/name` with content `content`. Args: tmpdir (LocalPath): root directory content (str/dict): Content of the file either as string or as a dict. If a dict, converts the dict into a yaml string. name (str): configuration filename Returns path (str) of the configuration file prepared. """ config_path = tmpdir / name if isinstance(content, dict): # convert if needed content = yaml.dump(content) config_path.write_text(content, encoding="utf-8") # pytest on python3.5 does not support LocalPath manipulation, so # convert path to string return str(config_path) def test_load_and_check_config_no_configuration(): """Inexistent configuration files raises""" with pytest.raises(EnvironmentError) as e: load_and_check_config(None) assert e.value.args[0] == "Configuration file must be defined" config_path = "/some/inexistent/config.yml" with pytest.raises(FileNotFoundError) as e: load_and_check_config(config_path) assert e.value.args[0] == "Configuration file %s does not exist" % (config_path,) def test_load_and_check_config_wrong_configuration(tmpdir): """Wrong configuration raises""" config_path = prepare_config_file(tmpdir, "something: useless") with pytest.raises(KeyError) as e: load_and_check_config(config_path) assert e.value.args[0] == "Missing '%scheduler' configuration" def test_load_and_check_config_remote_config_local_type_raise(tmpdir): """'local' configuration without 'local' storage raises""" config = {"scheduler": {"cls": "remote", "args": {}}} config_path = prepare_config_file(tmpdir, config) with pytest.raises(ValueError) as e: load_and_check_config(config_path, type="local") assert ( e.value.args[0] == "The scheduler backend can only be started with a 'local'" " configuration" ) def test_load_and_check_config_local_incomplete_configuration(tmpdir): """Incomplete 'local' configuration should raise""" config = { "scheduler": { "cls": "local", "args": {"db": "database", "something": "needed-for-test",}, } } for key in ["db", "args"]: c = copy.deepcopy(config) if key == "db": source = c["scheduler"]["args"] else: source = c["scheduler"] source.pop(key) config_path = prepare_config_file(tmpdir, c) with pytest.raises(KeyError) as e: load_and_check_config(config_path) assert ( e.value.args[0] == "Invalid configuration; missing '%s' config entry" % key ) def test_load_and_check_config_local_config_fine(tmpdir): """Local configuration is fine""" config = {"scheduler": {"cls": "local", "args": {"db": "db",}}} config_path = prepare_config_file(tmpdir, config) cfg = load_and_check_config(config_path, type="local") assert cfg == config def test_load_and_check_config_remote_config_fine(tmpdir): """'Remote configuration is fine""" config = {"scheduler": {"cls": "remote", "args": {}}} config_path = prepare_config_file(tmpdir, config) cfg = load_and_check_config(config_path, type="any") assert cfg == config diff --git a/swh/scheduler/tests/test_utils.py b/swh/scheduler/tests/test_utils.py index 655f644..85aecc6 100644 --- a/swh/scheduler/tests/test_utils.py +++ b/swh/scheduler/tests/test_utils.py @@ -1,79 +1,79 @@ # Copyright (C) 2017-2018 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information -import unittest from datetime import timezone +import unittest from unittest.mock import patch from swh.scheduler import utils class UtilsTest(unittest.TestCase): @patch("swh.scheduler.utils.datetime") def test_create_oneshot_task_dict_simple(self, mock_datetime): mock_datetime.now.return_value = "some-date" actual_task = utils.create_oneshot_task_dict("some-task-type") expected_task = { "policy": "oneshot", "type": "some-task-type", "next_run": "some-date", "arguments": {"args": [], "kwargs": {},}, } self.assertEqual(actual_task, expected_task) mock_datetime.now.assert_called_once_with(tz=timezone.utc) @patch("swh.scheduler.utils.datetime") def test_create_oneshot_task_dict_other_call(self, mock_datetime): mock_datetime.now.return_value = "some-other-date" actual_task = utils.create_oneshot_task_dict( "some-task-type", "arg0", "arg1", priority="high", other_stuff="normal" ) expected_task = { "policy": "oneshot", "type": "some-task-type", "next_run": "some-other-date", "arguments": { "args": ("arg0", "arg1"), "kwargs": {"other_stuff": "normal"}, }, "priority": "high", } self.assertEqual(actual_task, expected_task) mock_datetime.now.assert_called_once_with(tz=timezone.utc) @patch("swh.scheduler.utils.datetime") def test_create_task_dict(self, mock_datetime): mock_datetime.now.return_value = "date" actual_task = utils.create_task_dict( "task-type", "recurring", "arg0", "arg1", priority="low", other_stuff="normal", retries_left=3, ) expected_task = { "policy": "recurring", "type": "task-type", "next_run": "date", "arguments": { "args": ("arg0", "arg1"), "kwargs": {"other_stuff": "normal"}, }, "priority": "low", "retries_left": 3, } self.assertEqual(actual_task, expected_task) mock_datetime.now.assert_called_once_with(tz=timezone.utc)