When: During a pre-defined DBA maintenance windows
Prerequisites: https://wikitech.wikimedia.org/wiki/MariaDB/Primary_switchover
- Team calendar invite
Affected wikis:: https://noc.wikimedia.org/conf/highlight.php?file=dblists/s1.dblist
Checklist:
NEW primary: db1163
OLD primary: db1184
- Check configuration differences between new and old primary:
sudo pt-config-diff --defaults-file /root/.my.cnf h=db1184.eqiad.wmnet h=db1163.eqiad.wmnet
Failover prep:
- Silence alerts on all hosts:
sudo cookbook sre.hosts.downtime --hours 1 -r "Primary switchover s1 T372524" 'A:db-section-s1'
- Set NEW primary with weight 0
sudo dbctl instance db1163 set-weight 0 sudo dbctl config commit -m "Set db1163 with weight 0 T372524"
- Depool NEW from any specific group (API, vslow, dump) if present.
sudo dbctl instance db1163 edit # If some changes were made: sudo dbctl config commit -m "Remove db1163 from API/vslow/dump T372524"
- Topology changes, move all replicas under NEW primary
sudo db-switchover --timeout=25 --only-slave-move db1184 db1163
- Disable puppet on both nodes
sudo cumin 'db1184* or db1163*' 'disable-puppet "primary switchover T372524"'
- Merge gerrit puppet change to promote NEW primary: https://gerrit.wikimedia.org/r/c/operations/puppet/+/1062782
Failover:
- Log the failover:
!log Starting s1 eqiad failover from db1184 to db1163 - T372524
- Set section read-only:
sudo dbctl --scope eqiad section s1 ro "Maintenance until 06:15 UTC - T372524" sudo dbctl --scope codfw section s1 ro "Maintenance until 06:15 UTC - T372524" sudo dbctl config commit -m "Set s1 eqiad as read-only for maintenance - T372524"
- Check s1 is indeed read-only
- Switch primaries:
sudo db-switchover --skip-slave-move db1184 db1163 echo "===== db1184 (OLD)"; sudo db-mysql db1184 -e 'show slave status\G' echo "===== db1163 (NEW)"; sudo db-mysql db1163 -e 'show slave status\G'
- Promote NEW primary in dbctl, and remove read-only
sudo dbctl --scope eqiad section s1 set-master db1163 sudo dbctl --scope eqiad section s1 rw sudo dbctl --scope codfw section s1 rw sudo dbctl config commit -m "Promote db1163 to s1 primary and set section read-write T372524"
- Restart puppet on both hosts:
sudo cumin 'db1184* or db1163*' 'run-puppet-agent -e "primary switchover T372524"'
Clean up tasks:
- Clean up heartbeat table(s).
sudo db-mysql db1163 heartbeat -e "delete from heartbeat where file like 'db1184%';"
- change events for query killer:
curl -sS 'https://gerrit.wikimedia.org/r/plugins/gitiles/operations/software/+/refs/heads/master/dbtools/events_coredb_master.sql?format=TEXT' | base64 -d | sudo db-mysql db1163 curl -sS 'https://gerrit.wikimedia.org/r/plugins/gitiles/operations/software/+/refs/heads/master/dbtools/events_coredb_slave.sql?format=TEXT' | base64 -d | sudo db-mysql db1184
- Update DNS: https://gerrit.wikimedia.org/r/c/operations/dns/+/1062783
- Update candidate primary dbctl and orchestrator notes
sudo dbctl instance db1184 set-candidate-master --section s1 true sudo dbctl instance db1163 set-candidate-master --section s1 false sudo cumin 'dborch*' 'orchestrator-client -c untag -i db1163 --tag name=candidate' sudo cumin 'dborch*' 'orchestrator-client -c tag -i db1184 --tag name=candidate'
- Check zarcillo was updated
- db-switchover should do this. If it fails, do it manually: https://phabricator.wikimedia.org/P13956
sudo db-mysql db1215 zarcillo -e "select * from masters where section = 's1';"
- (If needed): Depool db1184 for maintenance.
sudo dbctl instance db1184 depool sudo dbctl config commit -m "Depool db1184 T372524"
- Change db1184 weight to mimic the previous weight db1163:
sudo dbctl instance db1184 edit
- Update/resolve this ticket.