Page MenuHomeSoftware Heritage

Meta-task: upgrade infrastructure to Debian Bullseye
Closed, MigratedEdits Locked

Description

Plan:

  • Upgrade the node to bullseye
  • Update the inventory [1] once the migration is done to mention the new distribution
  • Update Vagrantfile to use the corresponding debian 11 box

[1] https://inventory.internal.softwareheritage.org/

Servers to upgrade

  • T3771 Proxmox hypervisors
    • beaubourg.softwareheritage.org
    • branly.internal.softwareheritage.org
    • hypervisor3.internal.softwareheritage.org
    • pompidou.internal.softwareheritage.org
    • uffizi.softwareheritage.org

(!) Proxmox/Ceph upgrade ?

  • T3761: riverside.internal.softwareheritage.org (sentry)
  • T3762: kafka[1-4].internal.staging.swh.network
  • T3751: cluster log
    • esnode[1-3].internal.softwareheritage.org
    • kibana0.internal.softwareheritage.org
    • logstash0.internal.softwareheritage.org (ELK stack upgrade to 7.15.2)
  • T3751: search elasticsearch
    • search-esnode[4-6].internal.softwareheritage.org
    • esnode0.internal.staging.swh.network
  • T3801: Postgresql databases
    • belvedere.internal.softwareheritage.org
    • db1.internal.staging.swh.network
    • somerset.internal.softwareheritage.org
  • T3749: hedgedoc.softwareheritage.org (bardo.internal.admin.swh.network)
  • T3802: bojimans.internal.softwareheritage.org (netbox)
  • T3807: kelvingrove.internal.softwareheritage.org (keycloak)
  • Admin / DNS
    • T3800: ns0.euwest.azure.internal.softwareheritage.org
    • T3808: pergamon.softwareheritage.org ((!): check the plan in the task)
  • T3770: CI servers
    • jenkins-debian1.internal.softwareheritage.org
    • thyssen.internal.softwareheritage.org

Delayed migrations

Need some actions to separate swh services from the raw storage or
a bullseye version of the swh packages

  • counters
    • counters0.internal.staging.swh.network
    • counters1.internal.softwareheritage.org

redis needs storage, we could separate the search services / journal client from redis

  • Object storage / Storage
    • saam.internal.softwareheritage.org
    • storage1.internal.staging.swh.network
    • banco.softwareheritage.org

need to generate a swh-storage package for bullseye / don't use the debian package deployment
need to generate an obj-storage package for bullseye / don't use the debian package deployment
checkup backup compatibility

  • graph
    • granet.internal.softwareheritage.org

need to check the dependencies / installed packages (-> one is python3-swh.graph.client)
possibility to deploy the graph service dynamically ?

  • getty.internal.softwareheritage.org
    • used as a sandbox
    • CMAK compatibility ?
  • scheduler
    • scheduler0.internal.staging.swh.network
    • saatchi.internal.softwareheritage.org

Need to separate rabbitmq from the swh services and journal clients

(!) rabbitmq version has changed, 3.7.8 in debian10, 3.8.9 in debian 11.
The puppet module seems broken due to wrong dependencies when installing rabbitmqadmin

  • tate.softwareheritage.org

phabricator should be migrated to gitlab
mediawiki should be migrated
jenkins and keycloak RP should be kept / migrated

Servers left in busters

services will be migrated to a new deployment system

  • worker[0-3].internal.staging.swh.network
  • worker[01-13].euwest.azure.internal.softwareheritage.org
  • worker[01-17].softwareheritage.org
  • search0.internal.staging.swh.network
  • search1.internal.softwareheritage.org
  • deposit.internal.staging.swh.network
  • vault.internal.staging.swh.network
  • vangogh.euwest.azure.internal.softwareheritage.org
  • webapp.internal.staging.swh.network
  • webapp1.internal.softwareheritage.org
  • moma.softwareheritage.org
  • T3799: rp0.internal.staging.swh.network
  • rp1.internal.admin.swh.network
  • objstorage0.internal.staging.swh.network expose the objstorage service to the mirrors
  • storage01.euwest.azure.internal.softwareheritage.org r/o storage
    1. Services to upgrade
  • build
    • ci images
    • base swh image
  • debian packages
    • librdkafka ?
    • monitoring-plugins-systemd
    • prometheus-statsd-exporter
    • python3-elasticsearch
    • python3-amqp
    • python3-confluent-kafka
    • python3-importlib-metadata
    • python3-kombu
    • <to be continued>

Desktop

We need to double check with their owner for the upgrade whether they want to do it or not (questions asked on irc, awaiting response)

  • T3716: grouss: giverny.softwareheritage.org (currently from jessie to buster)
  • seirl: grand-palais.softwareheritage.org (migrated to bullseye)

Firewalls

  • T3706, T3760 minor then major opnsense upgrade (not bullseye migration)
    • glyptotek.internal.softwareheritage.org (bsd)
    • pushkin.internal.softwareheritage.org (bsd)

Decommission

  • T3792: louvre.internal.softwareheritage.org
  • T3805: boatbucket.internal.softwareheritage.org
  • journal0.internal.staging.swh.network
  • dbreplica0.euwest.azure.internal.softwareheritage.org (failing main db replica, cf. T2581)
  • dbreplica1.euwest.azure.internal.softwareheritage.org (indexer replica db, pg 11)

Unneeded migration

  • mirror-test.internal.staging.swh.network

node used by douardda, no swh packages are installed on it
Reason:

Tue 14 Dec 2021 06:09:57 PM CET
16:59 <+ardumont> douardda: btw, do you still need mirror-test node, can we migrate it or shall we drop it out of the migration path (T3579)?
17:00 <+douardda> I won't use it any time soon, so you can drop it

Related Objects

Event Timeline

There are a very large number of changes, so older changes are hidden. Show Older Changes
ardumont updated the task description. (Show Details)
vsellier updated the task description. (Show Details)
ardumont changed the status of subtask T3761: Migrate riverside to bullseye from Open to Work in Progress.
ardumont changed the task status from Open to Work in Progress.Dec 3 2021, 10:42 AM
ardumont updated the task description. (Show Details)
ardumont updated the task description. (Show Details)
vsellier changed the status of subtask T3800: migrate ns0 to bullseye from Open to Work in Progress.Dec 13 2021, 2:37 PM
vsellier updated the task description. (Show Details)

Closing this task as all the possible upgrade are done.
The delayed upgrades will be followed in dedicated task as it will be integradated in a more global task relative to the elastic infrastructure or the pergamon splitting task

vsellier claimed this task.
gitlab-migration changed the status of subtask T3716: Migrate giverny from stretch to buster from Resolved to Migrated.
gitlab-migration changed the status of subtask T3749: Migrate hedgedoc stack to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3750: Upgrade ELK stack to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3751: Upgrade swh-search elasticsearch servers to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3760: (Major) opnsense upgrade to 21.1.6 from Resolved to Migrated.
gitlab-migration changed the status of subtask T3761: Migrate riverside to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3762: Upgrade kafka nodes to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3770: Migrate jenkins nodes to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3799: Migrate rp0.internal.staging.swh.network to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3800: migrate ns0 to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3801: Migrate production database servers to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3802: Migrate bojimans (netbox) to bullseye from Resolved to Migrated.
gitlab-migration changed the status of subtask T3807: Migrate kelvingrove (keycloak) to bullseye from Resolved to Migrated.