When: Anytime, writes will be disabled
Prerequisites: https://wikitech.wikimedia.org/wiki/MariaDB/Primary_switchover
Checklist:
NEW primary: es1039
OLD primary: es1035
- Check configuration differences between new and old primary:
sudo pt-config-diff --defaults-file /root/.my.cnf h=es1035.eqiad.wmnet h=es1039.eqiad.wmnet
- Set section read-only:
sudo dbctl --scope eqiad section es7 ro "Maintenance - T400028" sudo dbctl --scope codfw section es7 ro "Maintenance - T400028" sudo dbctl config commit -m "Set es7 eqiad as read-only for maintenance - T400028"
- Check es7 is indeed read-only
Failover prep:
- Silence alerts on all hosts:
sudo cookbook sre.hosts.downtime --hours 1 -r "Primary switchover es7 T400028" 'A:db-section-es7'
- Set NEW primary with weight 0 (and depool it from API or vslow/dump groups if it is present).
sudo dbctl instance es1039 set-weight 0 sudo dbctl config commit -m "Set es1039 with weight 0 T400028"
- Topology changes, move all replicas under NEW primary
sudo db-switchover --timeout=25 --only-slave-move es1035 es1039
- Disable puppet on both nodes
sudo cumin 'es1035* or es1039*' 'disable-puppet "primary switchover T400028"'
- Merge gerrit puppet change to promote NEW primary: FIXME
Failover:
- Log the failover:
!log Starting es7 eqiad failover from es1035 to es1039 - T400028
- Switch primaries:
sudo db-switchover --skip-slave-move es1035 es1039 echo "===== es1035 (OLD)"; sudo db-mysql es1035 -e 'show slave status\G' echo "===== es1039 (NEW)"; sudo db-mysql es1039 -e 'show slave status\G'
- Promote NEW primary in dbctl, and remove read-only
sudo dbctl --scope eqiad section es7 set-master es1039 sudo dbctl config commit -m "Promote es1039 to es7 primary T400028"
- Clean up heartbeat table(s).
sudo db-mysql es1039 heartbeat -e "delete from heartbeat where file like 'es1035%';"
- Restart puppet on both hosts:
sudo cumin 'es1035* or es1039*' 'run-puppet-agent -e "primary switchover T400028"'
Clean up tasks:
- 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 es1039 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 es1035
- Merge DNS change: FIX ME
sudo authdns-update
- Update candidate primary dbctl and orchestrator notes
sudo dbctl instance es1035 set-candidate-master --section es7 true sudo dbctl instance es1039 set-candidate-master --section es7 false sudo cumin 'dborch*' 'orchestrator-client -c untag -i es1039 --tag name=candidate' sudo cumin 'dborch*' 'orchestrator-client -c tag -i es1035 --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 = 'es7';"
- (If needed): Depool es1035 for maintenance.
sudo dbctl instance es1035 depool sudo dbctl config commit -m "Depool es1035 T400028"
- Change es1035 weight to mimic the previous weight es1039:
sudo dbctl instance es1035 edit
- Set section back to read-write:
sudo dbctl --scope eqiad section es7 rw sudo dbctl --scope codfw section es7 rw sudo dbctl config commit -m "Set es7 eqiad back to read-write - T400028"
- Update/resolve this ticket.