As part of upgrading s3 to debian buster/mariadb 10.4, we need to switch the master to be db1157 (which is already running buster/10.4).
When: Mon 21th June at 05:00 AM UTC.
Checklist:
- Double check db1157 has report_host enabled T271106
- Create a task to communicate the chosen date and send an announcement to the community: T284650
NEW master: db1157
OLD master: db1123
- Check configuration differences between new and old master:
sudo pt-config-diff h=db1123.eqiad.wmnet,F=/root/.my.cnf h=db1157.eqiad.wmnet,F=/root/.my.cnf
Failover prep:
- Silence alerts on all hosts:
sudo cookbook sre.hosts.downtime --hours 1 -r "Master switchover s3 T284648" 'A:db-section-s3'
- Set NEW master with weight 0
sudo dbctl instance db1157 set-weight 0 sudo dbctl config commit -m "Set db1157 with weight 0 T284648"
- Topology changes, move all replicas under NEW master
sudo db-switchover --timeout=15 --only-slave-move db1123.eqiad.wmnet db1157.eqiad.wmnet
- Disable puppet on both nodes
sudo cumin 'db1123* or db1157*' 'disable-puppet "master switchover T284648"'
- Merge gerrit puppet change to promote NEW master: https://gerrit.wikimedia.org/r/c/operations/puppet/+/698981
Failover:
- Log the failover:
!log Starting s3 eqiad failover from db1123 to db1157 - T284648
- Set section read-only:
sudo dbctl --scope eqiad section s3 ro "Maintenance until 05:15 UTC - T284648" sudo dbctl config commit -m "Set s3 eqiad as read-only for maintenance - T284648"
- Check s3 is indeed read-only
- Switch masters:
sudo db-switchover --skip-slave-move db1123 db1157 echo "===== db1123 (OLD)"; sudo mysql.py -h db1123 -e 'show slave status\G' echo "===== db1157 (NEW)"; sudo mysql.py -h db1157 -e 'show slave status\G'
- Promote NEW master in dbctl, and remove read-only
sudo dbctl --scope eqiad section s3 set-master db1157 sudo dbctl --scope eqiad section s3 rw sudo dbctl config commit -m "Promote db1157 to s3 master and set section read-write T284648"
- Restart puppet on both hosts:
sudo cumin 'db1123* or db1157*' 'run-puppet-agent -e "master switchover T284648"'
Clean up tasks:
- change events for query killer:
events_coredb_master.sql on the new master db1157 events_coredb_slave.sql on the new slave db1123
- Update DNS: https://gerrit.wikimedia.org/r/c/operations/dns/+/698982
- Update candidate master dbctl notes
sudo dbctl instance db1123 set-candidate-master --section s3 true sudo dbctl instance db1157 set-candidate-master --section s3 false
- Check tendril was updated
- Check zarcillo was updated
- db-switchover should do this. If it fails, do it manually: https://phabricator.wikimedia.org/P13956
- Depool OLD master, as it's running 10.1, replicating from a 10.4 master
sudo dbctl instance db1123 depool sudo dbctl config commit -m "Depool db1123 until it's reimaged to buster T284648"
- Apply outstanding schema changes to db1123 (if any)
- Update/resolve this ticket.