origin_save.py
No OneTemporary
Actions

Size

24 KB

Subscribers

None

origin_save.py
View Options

	# Copyright (C) 2018-2021 The Software Heritage developers
	# See the AUTHORS file at the top-level directory of this distribution
	# License: GNU Affero General Public License version 3, or any later version
	# See top-level LICENSE file for more information

	from bisect import bisect_right
	from datetime import datetime, timedelta, timezone
	from itertools import product
	import json
	import logging
	from typing import Any, Dict, List, Optional, Tuple

	from prometheus_client import Gauge
	import requests
	import sentry_sdk

	from django.core.exceptions import ObjectDoesNotExist, ValidationError
	from django.core.validators import URLValidator
	from django.db.models import QuerySet
	from django.utils.html import escape

	from swh.scheduler.utils import create_oneshot_task_dict
	from swh.web import config
	from swh.web.common import archive
	from swh.web.common.exc import BadInputExc, ForbiddenExc, NotFoundExc
	from swh.web.common.models import (
	SAVE_REQUEST_ACCEPTED,
	SAVE_REQUEST_PENDING,
	SAVE_REQUEST_REJECTED,
	SAVE_TASK_FAILED,
	SAVE_TASK_NOT_CREATED,
	SAVE_TASK_NOT_YET_SCHEDULED,
	SAVE_TASK_RUNNING,
	SAVE_TASK_SCHEDULED,
	SAVE_TASK_SUCCEEDED,
	SaveAuthorizedOrigin,
	SaveOriginRequest,
	SaveUnauthorizedOrigin,
	)
	from swh.web.common.origin_visits import get_origin_visits
	from swh.web.common.typing import OriginInfo
	from swh.web.common.utils import SWH_WEB_METRICS_REGISTRY, parse_iso8601_date_to_utc

	scheduler = config.scheduler()

	logger = logging.getLogger(__name__)


	def get_origin_save_authorized_urls() -> List[str]:
	"""
	Get the list of origin url prefixes authorized to be
	immediately loaded into the archive (whitelist).

	Returns:
	list: The list of authorized origin url prefix
	"""
	return [origin.url for origin in SaveAuthorizedOrigin.objects.all()]


	def get_origin_save_unauthorized_urls() -> List[str]:
	"""
	Get the list of origin url prefixes forbidden to be
	loaded into the archive (blacklist).

	Returns:
	list: the list of unauthorized origin url prefix
	"""
	return [origin.url for origin in SaveUnauthorizedOrigin.objects.all()]


	def can_save_origin(origin_url: str) -> str:
	"""
	Check if a software origin can be saved into the archive.

	Based on the origin url, the save request will be either:

	* immediately accepted if the url is whitelisted
	* rejected if the url is blacklisted
	* put in pending state for manual review otherwise

	Args:
	origin_url (str): the software origin url to check

	Returns:
	str: the origin save request status, either accepted,
	rejected or pending
	"""
	# origin url may be blacklisted
	for url_prefix in get_origin_save_unauthorized_urls():
	if origin_url.startswith(url_prefix):
	return SAVE_REQUEST_REJECTED

	# if the origin url is in the white list, it can be immediately saved
	for url_prefix in get_origin_save_authorized_urls():
	if origin_url.startswith(url_prefix):
	return SAVE_REQUEST_ACCEPTED

	# otherwise, the origin url needs to be manually verified
	return SAVE_REQUEST_PENDING


	# map visit type to scheduler task
	# TODO: do not hardcode the task name here (T1157)
	_visit_type_task = {"git": "load-git", "hg": "load-hg", "svn": "load-svn"}


	# map scheduler task status to origin save status
	_save_task_status = {
	"next_run_not_scheduled": SAVE_TASK_NOT_YET_SCHEDULED,
	"next_run_scheduled": SAVE_TASK_SCHEDULED,
	"completed": SAVE_TASK_SUCCEEDED,
	"disabled": SAVE_TASK_FAILED,
	}

	# map scheduler task_run status to origin save status
	_save_task_run_status = {
	"scheduled": SAVE_TASK_SCHEDULED,
	"started": SAVE_TASK_RUNNING,
	"eventful": SAVE_TASK_SUCCEEDED,
	"uneventful": SAVE_TASK_SUCCEEDED,
	"failed": SAVE_TASK_FAILED,
	"permfailed": SAVE_TASK_FAILED,
	"lost": SAVE_TASK_FAILED,
	}


	def get_savable_visit_types() -> List[str]:
	"""
	Get the list of visit types that can be performed
	through a save request.

	Returns:
	list: the list of saveable visit types
	"""
	return sorted(list(_visit_type_task.keys()))


	def _check_visit_type_savable(visit_type: str) -> None:
	allowed_visit_types = ", ".join(get_savable_visit_types())
	if visit_type not in _visit_type_task:
	raise BadInputExc(
	"Visit of type %s can not be saved! "
	"Allowed types are the following: %s" % (visit_type, allowed_visit_types)
	)


	_validate_url = URLValidator(schemes=["http", "https", "svn", "git"])


	def _check_origin_url_valid(origin_url: str) -> None:
	try:
	_validate_url(origin_url)
	except ValidationError:
	raise BadInputExc(
	"The provided origin url (%s) is not valid!" % escape(origin_url)
	)


	def _get_visit_info_for_save_request(
	save_request: SaveOriginRequest,
	) -> Tuple[Optional[datetime], Optional[str]]:
	visit_date = None
	visit_status = None
	time_now = datetime.now(tz=timezone.utc)
	time_delta = time_now - save_request.request_date
	# stop trying to find a visit date one month after save request submission
	# as those requests to storage are expensive and associated loading task
	# surely ended up with errors
	if time_delta.days <= 30:
	try:
	origin_info = archive.lookup_origin(OriginInfo(url=save_request.origin_url))
	origin_visits = get_origin_visits(origin_info)
	visit_dates = [parse_iso8601_date_to_utc(v["date"]) for v in origin_visits]
	i = bisect_right(visit_dates, save_request.request_date)
	if i != len(visit_dates):
	visit_date = visit_dates[i]
	visit_status = origin_visits[i]["status"]
	if origin_visits[i]["status"] not in ("full", "partial", "not_found"):
	visit_date = None
	except Exception as exc:
	sentry_sdk.capture_exception(exc)
	return visit_date, visit_status


	def _check_visit_update_status(
	save_request: SaveOriginRequest, save_task_status: str
	) -> Tuple[Optional[datetime], str]:
	visit_date, visit_status = _get_visit_info_for_save_request(save_request)
	save_request.visit_date = visit_date
	# visit has been performed, mark the saving task as succeed
	if visit_date and visit_status is not None:
	save_task_status = SAVE_TASK_SUCCEEDED
	elif visit_status in ("created", "ongoing"):
	save_task_status = SAVE_TASK_RUNNING
	elif visit_status in ("not_found", "failed"):
	save_task_status = SAVE_TASK_FAILED
	else:
	time_now = datetime.now(tz=timezone.utc)
	time_delta = time_now - save_request.request_date
	# consider the task as failed if it is still in scheduled state
	# 30 days after its submission
	if time_delta.days > 30:
	save_task_status = SAVE_TASK_FAILED
	return visit_date, save_task_status


	def _save_request_dict(
	save_request: SaveOriginRequest,
	task: Optional[Dict[str, Any]] = None,
	task_run: Optional[Dict[str, Any]] = None,
	) -> Dict[str, Any]:
	must_save = False
	visit_date = save_request.visit_date
	# save task still in scheduler db
	if task:
	save_task_status = _save_task_status[task["status"]]
	if task_run:
	save_task_status = _save_task_run_status[task_run["status"]]
	# Consider request from which a visit date has already been found
	# as succeeded to avoid retrieving it again
	if save_task_status == SAVE_TASK_SCHEDULED and visit_date:
	save_task_status = SAVE_TASK_SUCCEEDED
	if (
	save_task_status in (SAVE_TASK_FAILED, SAVE_TASK_SUCCEEDED)
	and not visit_date
	):
	visit_date, visit_status = _get_visit_info_for_save_request(save_request)
	save_request.visit_date = visit_date
	if visit_status in ("failed", "not_found"):
	save_task_status = SAVE_TASK_FAILED
	must_save = True
	# Check tasks still marked as scheduled / not yet scheduled
	if save_task_status in (SAVE_TASK_SCHEDULED, SAVE_TASK_NOT_YET_SCHEDULED):
	visit_date, save_task_status = _check_visit_update_status(
	save_request, save_task_status
	)

	# save task may have been archived
	else:
	save_task_status = save_request.loading_task_status
	if save_task_status in (SAVE_TASK_SCHEDULED, SAVE_TASK_NOT_YET_SCHEDULED):
	visit_date, save_task_status = _check_visit_update_status(
	save_request, save_task_status
	)

	else:
	save_task_status = save_request.loading_task_status

	if (
	# avoid to override final loading task status when already found
	# as visit status is no longer checked once a visit date has been found
	save_request.loading_task_status not in (SAVE_TASK_FAILED, SAVE_TASK_SUCCEEDED)
	and save_request.loading_task_status != save_task_status
	):
	save_request.loading_task_status = save_task_status
	must_save = True

	if must_save:
	save_request.save()

	return {
	"id": save_request.id,
	"visit_type": save_request.visit_type,
	"origin_url": save_request.origin_url,
	"save_request_date": save_request.request_date.isoformat(),
	"save_request_status": save_request.status,
	"save_task_status": save_request.loading_task_status,
	"visit_date": visit_date.isoformat() if visit_date else None,
	}


	def create_save_origin_request(visit_type: str, origin_url: str) -> Dict[str, Any]:
	"""
	Create a loading task to save a software origin into the archive.

	This function aims to create a software origin loading task
	trough the use of the swh-scheduler component.

	First, some checks are performed to see if the visit type and origin
	url are valid but also if the the save request can be accepted.
	If those checks passed, the loading task is then created.
	Otherwise, the save request is put in pending or rejected state.

	All the submitted save requests are logged into the swh-web
	database to keep track of them.

	Args:
	visit_type (str): the type of visit to perform (currently only
	``git`` but ``svn`` and ``hg`` will soon be available)
	origin_url (str): the url of the origin to save

	Raises:
	BadInputExc: the visit type or origin url is invalid
	ForbiddenExc: the provided origin url is blacklisted

	Returns:
	dict: A dict describing the save request with the following keys:

	* visit_type: the type of visit to perform
	* origin_url: the url of the origin
	* save_request_date: the date the request was submitted
	* save_request_status: the request status, either accepted,
	rejected or pending
	* save_task_status: the origin loading task status, either
	not created, not yet scheduled, scheduled,
	succeed or failed


	"""
	_check_visit_type_savable(visit_type)
	_check_origin_url_valid(origin_url)
	save_request_status = can_save_origin(origin_url)
	task = None

	# if the origin save request is accepted, create a scheduler
	# task to load it into the archive
	if save_request_status == SAVE_REQUEST_ACCEPTED:
	# create a task with high priority
	kwargs = {
	"priority": "high",
	"url": origin_url,
	}
	sor = None
	# get list of previously sumitted save requests
	current_sors = list(
	SaveOriginRequest.objects.filter(
	visit_type=visit_type, origin_url=origin_url
	)
	)

	can_create_task = False
	# if no save requests previously submitted, create the scheduler task
	if not current_sors:
	can_create_task = True
	else:
	# get the latest submitted save request
	sor = current_sors[0]
	# if it was in pending state, we need to create the scheduler task
	# and update the save request info in the database
	if sor.status == SAVE_REQUEST_PENDING:
	can_create_task = True
	# a task has already been created to load the origin
	elif sor.loading_task_id != -1:
	# get the scheduler task and its status
	tasks = scheduler.get_tasks([sor.loading_task_id])
	task = tasks[0] if tasks else None
	task_runs = scheduler.get_task_runs([sor.loading_task_id])
	task_run = task_runs[0] if task_runs else None
	save_request = _save_request_dict(sor, task, task_run)
	task_status = save_request["save_task_status"]
	# create a new scheduler task only if the previous one has been
	# already executed
	if (
	task_status == SAVE_TASK_FAILED
	or task_status == SAVE_TASK_SUCCEEDED
	):
	can_create_task = True
	sor = None
	else:
	can_create_task = False

	if can_create_task:
	# effectively create the scheduler task
	task_dict = create_oneshot_task_dict(_visit_type_task[visit_type], **kwargs)
	task = scheduler.create_tasks([task_dict])[0]

	# pending save request has been accepted
	if sor:
	sor.status = SAVE_REQUEST_ACCEPTED
	sor.loading_task_id = task["id"]
	sor.save()
	else:
	sor = SaveOriginRequest.objects.create(
	visit_type=visit_type,
	origin_url=origin_url,
	status=save_request_status,
	loading_task_id=task["id"],
	)
	# save request must be manually reviewed for acceptation
	elif save_request_status == SAVE_REQUEST_PENDING:
	# check if there is already such a save request already submitted,
	# no need to add it to the database in that case
	try:
	sor = SaveOriginRequest.objects.get(
	visit_type=visit_type, origin_url=origin_url, status=save_request_status
	)
	# if not add it to the database
	except ObjectDoesNotExist:
	sor = SaveOriginRequest.objects.create(
	visit_type=visit_type, origin_url=origin_url, status=save_request_status
	)
	# origin can not be saved as its url is blacklisted,
	# log the request to the database anyway
	else:
	sor = SaveOriginRequest.objects.create(
	visit_type=visit_type, origin_url=origin_url, status=save_request_status
	)

	if save_request_status == SAVE_REQUEST_REJECTED:
	raise ForbiddenExc(
	(
	'The "save code now" request has been rejected '
	"because the provided origin url is blacklisted."
	)
	)

	assert sor is not None
	return _save_request_dict(sor, task)


	def get_save_origin_requests_from_queryset(
	requests_queryset: QuerySet,
	) -> List[Dict[str, Any]]:
	"""
	Get all save requests from a SaveOriginRequest queryset.

	Args:
	requests_queryset (django.db.models.QuerySet): input
	SaveOriginRequest queryset

	Returns:
	list: A list of save origin requests dict as described in
	:func:`swh.web.common.origin_save.create_save_origin_request`
	"""
	task_ids = []
	for sor in requests_queryset:
	task_ids.append(sor.loading_task_id)
	save_requests = []
	if task_ids:
	tasks = scheduler.get_tasks(task_ids)
	tasks = {task["id"]: task for task in tasks}
	task_runs = scheduler.get_task_runs(tasks)
	task_runs = {task_run["task"]: task_run for task_run in task_runs}
	for sor in requests_queryset:
	sr_dict = _save_request_dict(
	sor, tasks.get(sor.loading_task_id), task_runs.get(sor.loading_task_id)
	)
	save_requests.append(sr_dict)
	return save_requests


	def get_save_origin_requests(visit_type: str, origin_url: str) -> List[Dict[str, Any]]:
	"""
	Get all save requests for a given software origin.

	Args:
	visit_type (str): the type of visit
	origin_url (str): the url of the origin

	Raises:
	BadInputExc: the visit type or origin url is invalid
	swh.web.common.exc.NotFoundExc: no save requests can be found for the
	given origin

	Returns:
	list: A list of save origin requests dict as described in
	:func:`swh.web.common.origin_save.create_save_origin_request`
	"""
	_check_visit_type_savable(visit_type)
	_check_origin_url_valid(origin_url)
	sors = SaveOriginRequest.objects.filter(
	visit_type=visit_type, origin_url=origin_url
	)
	if sors.count() == 0:
	raise NotFoundExc(
	("No save requests found for visit of type " "%s on origin with url %s.")
	% (visit_type, origin_url)
	)
	return get_save_origin_requests_from_queryset(sors)


	def get_save_origin_task_info(
	save_request_id: int, full_info: bool = True
	) -> Dict[str, Any]:
	"""
	Get detailed information about an accepted save origin request
	and its associated loading task.

	If the associated loading task info is archived and removed
	from the scheduler database, returns an empty dictionary.

	Args:
	save_request_id: identifier of a save origin request
	full_info: whether to return detailed info for staff users

	Returns:
	A dictionary with the following keys:

	- type: loading task type
	- arguments: loading task arguments
	- id: loading task database identifier
	- backend_id: loading task celery identifier
	- scheduled: loading task scheduling date
	- ended: loading task termination date
	- status: loading task execution status

	Depending on the availability of the task logs in the elasticsearch
	cluster of Software Heritage, the returned dictionary may also
	contain the following keys:

	- name: associated celery task name
	- message: relevant log message from task execution
	- duration: task execution time (only if it succeeded)
	- worker: name of the worker that executed the task
	"""
	try:
	save_request = SaveOriginRequest.objects.get(id=save_request_id)
	except ObjectDoesNotExist:
	return {}

	task = scheduler.get_tasks([save_request.loading_task_id])
	task = task[0] if task else None
	if task is None:
	return {}

	task_run = scheduler.get_task_runs([task["id"]])
	task_run = task_run[0] if task_run else None
	if task_run is None:
	return {}
	task_run["type"] = task["type"]
	task_run["arguments"] = task["arguments"]
	task_run["id"] = task_run["task"]
	del task_run["task"]
	del task_run["metadata"]

	es_workers_index_url = config.get_config()["es_workers_index_url"]
	if not es_workers_index_url:
	return task_run
	es_workers_index_url += "/_search"

	if save_request.visit_date:
	min_ts = save_request.visit_date
	max_ts = min_ts + timedelta(days=7)
	else:
	min_ts = save_request.request_date
	max_ts = min_ts + timedelta(days=30)
	min_ts_unix = int(min_ts.timestamp()) * 1000
	max_ts_unix = int(max_ts.timestamp()) * 1000

	save_task_status = _save_task_status[task["status"]]
	priority = "3" if save_task_status == SAVE_TASK_FAILED else "6"

	query = {
	"bool": {
	"must": [
	{"match_phrase": {"priority": {"query": priority}}},
	{"match_phrase": {"swh_task_id": {"query": task_run["backend_id"]}}},
	{
	"range": {
	"@timestamp": {
	"gte": min_ts_unix,
	"lte": max_ts_unix,
	"format": "epoch_millis",
	}
	}
	},
	]
	}
	}

	try:
	response = requests.post(
	es_workers_index_url,
	json={"query": query, "sort": ["@timestamp"]},
	timeout=30,
	)
	results = json.loads(response.text)
	if results["hits"]["total"]["value"] >= 1:
	task_run_info = results["hits"]["hits"][-1]["_source"]
	if "swh_logging_args_runtime" in task_run_info:
	duration = task_run_info["swh_logging_args_runtime"]
	task_run["duration"] = duration
	if "message" in task_run_info:
	task_run["message"] = task_run_info["message"]
	if "swh_logging_args_name" in task_run_info:
	task_run["name"] = task_run_info["swh_logging_args_name"]
	elif "swh_task_name" in task_run_info:
	task_run["name"] = task_run_info["swh_task_name"]
	if "hostname" in task_run_info:
	task_run["worker"] = task_run_info["hostname"]
	elif "host" in task_run_info:
	task_run["worker"] = task_run_info["host"]
	except Exception as exc:
	logger.warning("Request to Elasticsearch failed\n%s", exc)
	sentry_sdk.capture_exception(exc)

	if not full_info:
	for field in ("id", "backend_id", "worker"):
	# remove some staff only fields
	task_run.pop(field, None)
	if "message" in task_run and "Loading failure" in task_run["message"]:
	# hide traceback for non staff users, only display exception
	message_lines = task_run["message"].split("\n")
	message = ""
	for line in message_lines:
	if line.startswith("Traceback"):
	break
	message += f"{line}\n"
	message += message_lines[-1]
	task_run["message"] = message

	return task_run


	SUBMITTED_SAVE_REQUESTS_METRIC = "swh_web_submitted_save_requests"

	_submitted_save_requests_gauge = Gauge(
	name=SUBMITTED_SAVE_REQUESTS_METRIC,
	documentation="Number of submitted origin save requests",
	labelnames=["status", "visit_type"],
	registry=SWH_WEB_METRICS_REGISTRY,
	)


	ACCEPTED_SAVE_REQUESTS_METRIC = "swh_web_accepted_save_requests"

	_accepted_save_requests_gauge = Gauge(
	name=ACCEPTED_SAVE_REQUESTS_METRIC,
	documentation="Number of accepted origin save requests",
	labelnames=["load_task_status", "visit_type"],
	registry=SWH_WEB_METRICS_REGISTRY,
	)


	# Metric on the delay of save code now request per status and visit_type. This is the
	# time difference between the save code now is requested and the time it got ingested.
	ACCEPTED_SAVE_REQUESTS_DELAY_METRIC = "swh_web_save_requests_delay_seconds"

	_accepted_save_requests_delay_gauge = Gauge(
	name=ACCEPTED_SAVE_REQUESTS_DELAY_METRIC,
	documentation="Save Requests Duration",
	labelnames=["load_task_status", "visit_type"],
	registry=SWH_WEB_METRICS_REGISTRY,
	)


	def compute_save_requests_metrics() -> None:
	"""Compute Prometheus metrics related to origin save requests:

	- Number of submitted origin save requests
	- Number of accepted origin save requests
	- Save Code Now requests delay between request time and actual time of ingestion

	"""

	request_statuses = (
	SAVE_REQUEST_ACCEPTED,
	SAVE_REQUEST_REJECTED,
	SAVE_REQUEST_PENDING,
	)

	load_task_statuses = (
	SAVE_TASK_NOT_CREATED,
	SAVE_TASK_NOT_YET_SCHEDULED,
	SAVE_TASK_SCHEDULED,
	SAVE_TASK_SUCCEEDED,
	SAVE_TASK_FAILED,
	SAVE_TASK_RUNNING,
	)

	visit_types = get_savable_visit_types()

	labels_set = product(request_statuses, visit_types)

	for labels in labels_set:
	_submitted_save_requests_gauge.labels(*labels).set(0)

	labels_set = product(load_task_statuses, visit_types)

	for labels in labels_set:
	_accepted_save_requests_gauge.labels(*labels).set(0)

	duration_load_task_statuses = (
	SAVE_TASK_FAILED,
	SAVE_TASK_SUCCEEDED,
	)

	for labels in product(duration_load_task_statuses, visit_types):
	_accepted_save_requests_delay_gauge.labels(*labels).set(0)

	for sor in SaveOriginRequest.objects.all():
	if sor.status == SAVE_REQUEST_ACCEPTED:
	_accepted_save_requests_gauge.labels(
	load_task_status=sor.loading_task_status, visit_type=sor.visit_type
	).inc()

	_submitted_save_requests_gauge.labels(
	status=sor.status, visit_type=sor.visit_type
	).inc()

	if (
	sor.loading_task_status in (SAVE_TASK_SUCCEEDED, SAVE_TASK_FAILED)
	and sor.visit_date is not None
	and sor.request_date is not None
	):
	delay = sor.visit_date.timestamp() - sor.request_date.timestamp()
	_accepted_save_requests_delay_gauge.labels(
	load_task_status=sor.loading_task_status, visit_type=sor.visit_type
	).inc(delay)

File Metadata

Mime Type: text/x-python
Expires: Thu, Apr 17, 7:55 AM (4 d, 19 h ago)
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 3287693

origin_save.pyNo OneTemporaryActions

origin_save.pyView Options

File Metadata

Event Timeline

origin_save.py
No OneTemporary
Actions

origin_save.py
View Options