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/s5.dblist
Checklist:
NEW primary: db1230
OLD primary: db1183
- Check configuration differences between new and old primary:
sudo pt-config-diff --defaults-file /root/.my.cnf h=db1183.eqiad.wmnet h=db1230.eqiad.wmnet
Failover prep:
- Silence alerts on all hosts:
sudo cookbook sre.hosts.downtime --hours 1 -r "Primary switchover s5 T369616" 'A:db-section-s5'
- Set NEW primary with weight 0
sudo dbctl instance db1230 set-weight 0 sudo dbctl config commit -m "Set db1230 with weight 0 T369616"
- Depool NEW from any specific group (API, vslow, dump) if present.
sudo dbctl instance db1230 edit # If some changes were made: sudo dbctl config commit -m "Remove db1230 from API/vslow/dump T369616"
- Topology changes, move all replicas under NEW primary
sudo db-switchover --timeout=25 --only-slave-move db1183 db1230
- Disable puppet on both nodes
sudo cumin 'db1183* or db1230*' 'disable-puppet "primary switchover T369616"'
- Merge gerrit puppet change to promote NEW primary: https://gerrit.wikimedia.org/r/c/operations/puppet/+/1052948
Failover:
- Log the failover:
!log Starting s5 eqiad failover from db1183 to db1230 - T369616
- Set section read-only:
sudo dbctl --scope eqiad section s5 ro "Maintenance until 06:15 UTC - T369616" sudo dbctl --scope codfw section s5 ro "Maintenance until 06:15 UTC - T369616" sudo dbctl config commit -m "Set s5 eqiad as read-only for maintenance - T369616"
- Check s5 is indeed read-only
- Switch primaries:
sudo db-switchover --skip-slave-move db1183 db1230 echo "===== db1183 (OLD)"; sudo db-mysql db1183 -e 'show slave status\G' echo "===== db1230 (NEW)"; sudo db-mysql db1230 -e 'show slave status\G'
- Promote NEW primary in dbctl, and remove read-only
sudo dbctl --scope eqiad section s5 set-master db1230 sudo dbctl --scope eqiad section s5 rw sudo dbctl --scope codfw section s5 rw sudo dbctl config commit -m "Promote db1230 to s5 primary and set section read-write T369616"
- Restart puppet on both hosts:
sudo cumin 'db1183* or db1230*' 'run-puppet-agent -e "primary switchover T369616"'
Clean up tasks:
- Clean up heartbeat table(s).
sudo db-mysql db1230 heartbeat -e "delete from heartbeat where file like 'db1183%';"
- 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 db1230 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 db1183
- Update DNS: https://gerrit.wikimedia.org/r/c/operations/dns/+/1052949
- Update candidate primary dbctl and orchestrator notes
sudo dbctl instance db1183 set-candidate-master --section s5 true sudo dbctl instance db1230 set-candidate-master --section s5 false (dborch1001): sudo orchestrator-client -c untag -i db1230 --tag name=candidate (dborch1001): sudo orchestrator-client -c tag -i db1183 --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 = 's5';"
- (If needed): Depool db1183 for maintenance.
sudo dbctl instance db1183 depool sudo dbctl config commit -m "Depool db1183 T369616"
- Change db1183 weight to mimic the previous weight db1230:
sudo dbctl instance db1183 edit
- Update/resolve this ticket.