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/s2.dblist
Checklist:
NEW primary: db1222
OLD primary: db1162
- Check configuration differences between new and old primary:
sudo pt-config-diff --defaults-file /root/.my.cnf h=db1162.eqiad.wmnet h=db1222.eqiad.wmnet
Failover prep:
- Silence alerts on all hosts:
sudo cookbook sre.hosts.downtime --hours 1 -r "Primary switchover s2 T362036" 'A:db-section-s2'
- Set NEW primary with weight 0 (and depool it from API or vslow/dump groups if it is present).
sudo dbctl instance db1222 set-weight 0 sudo dbctl config commit -m "Set db1222 with weight 0 T362036"
- Topology changes, move all replicas under NEW primary
sudo db-switchover --timeout=25 --only-slave-move db1162 db1222
- Disable puppet on both nodes
sudo cumin 'db1162* or db1222*' 'disable-puppet "primary switchover T362036"'
- Merge gerrit puppet change to promote NEW primary: https://gerrit.wikimedia.org/r/1017451
Failover:
- Log the failover:
!log Starting s2 eqiad failover from db1162 to db1222 - T362036
- Set section read-only:
sudo dbctl --scope eqiad section s2 ro "Maintenance until 06:15 UTC - T362036" sudo dbctl --scope codfw section s2 ro "Maintenance until 06:15 UTC - T362036" sudo dbctl config commit -m "Set s2 eqiad as read-only for maintenance - T362036"
- Check s2 is indeed read-only
- Switch primaries:
sudo db-switchover --skip-slave-move db1162 db1222 echo "===== db1162 (OLD)"; sudo db-mysql db1162 -e 'show slave status\G' echo "===== db1222 (NEW)"; sudo db-mysql db1222 -e 'show slave status\G'
- Promote NEW primary in dbctl, and remove read-only
sudo dbctl --scope eqiad section s2 set-master db1222 sudo dbctl --scope eqiad section s2 rw sudo dbctl --scope codfw section s2 rw sudo dbctl config commit -m "Promote db1222 to s2 primary and set section read-write T362036"
- Restart puppet on both hosts:
sudo cumin 'db1162* or db1222*' 'run-puppet-agent -e "primary switchover T362036"'
Clean up tasks:
- Clean up heartbeat table(s).
sudo db-mysql db1222 heartbeat -e "delete from heartbeat where file like 'db1162%';"
- change events for query killer:
events_coredb_master.sql on the new primary db1222 events_coredb_slave.sql on the new slave db1162
- Update DNS: https://gerrit.wikimedia.org/r/1017452
- Update candidate primary dbctl and orchestrator notes
sudo dbctl instance db1162 set-candidate-master --section s2 true sudo dbctl instance db1222 set-candidate-master --section s2 false (dborch1001): sudo orchestrator-client -c untag -i db1222 --tag name=candidate (dborch1001): sudo orchestrator-client -c tag -i db1162 --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 = 's2';"
- (If needed): Depool db1162 for maintenance.
sudo dbctl instance db1162 depool sudo dbctl config commit -m "Depool db1162 T362036"
- Change db1162 weight to mimic the previous weight db1222:
sudo dbctl instance db1162 edit
- Apply outstanding schema changes to db1162 (if any)
- Update/resolve this ticket.