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/s6.dblist
Checklist:
NEW primary: db1201
OLD primary: db1173
- Check configuration differences between new and old primary:
sudo pt-config-diff --defaults-file /root/.my.cnf h=db1173.eqiad.wmnet h=db1201.eqiad.wmnet
Failover prep:
- Silence alerts on all hosts:
sudo cookbook sre.hosts.downtime --hours 1 -r "Primary switchover s6 T371365" 'A:db-section-s6'
- Set NEW primary with weight 0
sudo dbctl instance db1201 set-weight 0 sudo dbctl config commit -m "Set db1201 with weight 0 T371365"
- Depool NEW from any specific group (API, vslow, dump) if present.
sudo dbctl instance db1201 edit # If some changes were made: sudo dbctl config commit -m "Remove db1201 from API/vslow/dump T371365"
- Topology changes, move all replicas under NEW primary
sudo db-switchover --timeout=25 --only-slave-move db1173 db1201
- Disable puppet on both nodes
sudo cumin 'db1173* or db1201*' 'disable-puppet "primary switchover T371365"'
- Merge gerrit puppet change to promote NEW primary: https://gerrit.wikimedia.org/r/c/operations/puppet/+/1058141
Failover:
- Log the failover:
!log Starting s6 eqiad failover from db1173 to db1201 - T371365
- Set section read-only:
sudo dbctl --scope eqiad section s6 ro "Maintenance until 06:15 UTC - T371365" sudo dbctl --scope codfw section s6 ro "Maintenance until 06:15 UTC - T371365" sudo dbctl config commit -m "Set s6 eqiad as read-only for maintenance - T371365"
- Check s6 is indeed read-only
- Switch primaries:
sudo db-switchover --skip-slave-move db1173 db1201 echo "===== db1173 (OLD)"; sudo db-mysql db1173 -e 'show slave status\G' echo "===== db1201 (NEW)"; sudo db-mysql db1201 -e 'show slave status\G'
- Promote NEW primary in dbctl, and remove read-only
sudo dbctl --scope eqiad section s6 set-master db1201 sudo dbctl --scope eqiad section s6 rw sudo dbctl --scope codfw section s6 rw sudo dbctl config commit -m "Promote db1201 to s6 primary and set section read-write T371365"
- Restart puppet on both hosts:
sudo cumin 'db1173* or db1201*' 'run-puppet-agent -e "primary switchover T371365"'
Clean up tasks:
- Clean up heartbeat table(s).
sudo db-mysql db1201 heartbeat -e "delete from heartbeat where file like 'db1173%';"
- 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 db1201 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 db1173
- Update DNS: https://gerrit.wikimedia.org/r/c/operations/dns/+/1058142
- Update candidate primary dbctl and orchestrator notes
sudo dbctl instance db1173 set-candidate-master --section s6 true sudo dbctl instance db1201 set-candidate-master --section s6 false sudo cumin 'dborch*' 'orchestrator-client -c untag -i db1201 --tag name=candidate' sudo cumin 'dborch*' 'orchestrator-client -c tag -i db1173 --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 = 's6';"
- (If needed): Depool db1173 for maintenance.
sudo dbctl instance db1173 depool sudo dbctl config commit -m "Depool db1173 T371365"
- Change db1173 weight to mimic the previous weight db1201:
sudo dbctl instance db1173 edit
- Update/resolve this ticket.