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/s1.dblist
Checklist:
NEW primary: db2103
OLD primary: db2112
- Check configuration differences between new and old primary:
sudo pt-config-diff --defaults-file /root/.my.cnf h=db2112.codfw.wmnet h=db2103.codfw.wmnet
$ sudo pt-config-diff --defaults-file /root/.my.cnf h=db2112.codfw.wmnet h=db2103.codfw.wmnet 20 config differences Variable db2112 db2103 ========================= ========================= ========================= general_log_file db2112.log db2103.log gtid_binlog_pos 171970572-171970572-39... 171970572-171970572-39... gtid_binlog_state 171970572-171970572-39... 171970572-171970572-39... gtid_current_pos 0-171970637-5484646134... 0-171970637-5484646134... gtid_domain_id 180363268 180355190 gtid_slave_pos 0-171970637-5484646134... 0-171970637-5484646134... hostname db2112 db2103 innodb_version 10.4.25 10.4.26 log_bin_basename /srv/sqldata/db2112-bin /srv/sqldata/db2103-bin log_bin_index /srv/sqldata/db2112-bi... /srv/sqldata/db2103-bi... pid_file /srv/sqldata/db2112.pid /srv/sqldata/db2103.pid read_only OFF ON report_host db2112.codfw.wmnet db2103.codfw.wmnet rpl_semi_sync_master_e... ON OFF rpl_semi_sync_slave_en... OFF ON server_id 180363268 180355190 slow_query_log_file db2112-slow.log db2103-slow.log version 10.4.25-MariaDB-log 10.4.26-MariaDB-log version_source_revision 23ddc3518f999e003d54f7... 65e8506ca9d03967191b6e... wsrep_node_name db2112 db2103
Failover prep:
- Silence alerts on all hosts:
sudo cookbook sre.hosts.downtime --hours 2 -r "Primary switchover s1 T350142" 'A:db-section-s1'
- Set NEW primary with weight 0 (and depool it from API or vslow/dump groups if it is present).
⚠️ Take note of the weight to restore it on the old primary later.
sudo dbctl instance db2103 set-weight 0 sudo dbctl config commit -m "Set db2103 with weight 0 T350142"
- Topology changes, move all replicas under NEW primary
sudo db-switchover --timeout=25 --only-slave-move db2112 db2103
- Disable puppet on both nodes
sudo cumin 'db2112* or db2103*' 'disable-puppet "primary switchover T350142"'
- Merge gerrit puppet change to promote NEW primary:
sudo puppet-merge
Failover:
- Log the failover:
!log Starting s1 codfw failover from db2112 to db2103 - T350142
- Set section read-only:
sudo dbctl --scope codfw section s1 ro "Maintenance until 06:15 UTC - T350142" sudo dbctl config commit -m "Set s1 codfw as read-only for maintenance - T350142"
- Check s1 is indeed read-only
select @@read_only;
- Switch primaries:
sudo db-switchover --skip-slave-move db2112 db2103
⚠️ Retrieve the 2 curl commands that the script returns
echo "===== db2112 (OLD)"; sudo db-mysql db2112 -e 'show slave status\G' echo "===== db2103 (NEW)"; sudo db-mysql db2103 -e 'show slave status\G'
- Promote NEW primary in dbctl, and remove read-only
sudo dbctl --scope codfw section s1 set-master db2103 sudo dbctl --scope codfw section s1 rw sudo dbctl config commit -m "Promote db2103 to s1 primary and set section read-write T350142"
- Restart puppet on both hosts:
sudo cumin 'db2112* or db2103*' 'run-puppet-agent -e "primary switchover T350142"'
Clean up tasks:
- Clean up heartbeat table(s).
sudo db-mysql db2103 heartbeat -e "delete from heartbeat where file like 'db2112%';"
- change events for query killer:
events_coredb_master.sql on the new primary db2103 events_coredb_slave.sql on the new slave db2112
- Update DNS:
ssh ns0.wikimedia.org sudo authdns-update
- Update candidate primary dbctl and orchestrator notes
sudo dbctl instance db2112 set-candidate-master --section s1 true sudo dbctl instance db2103 set-candidate-master --section s1 false (dborch1001): sudo orchestrator-client -c untag -i db2103 --tag name=candidate (dborch1001): sudo orchestrator-client -c tag -i db2112 --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 = 's1';"
- (If needed): Depool db2112 for maintenance.
sudo dbctl instance db2112 depool sudo dbctl config commit -m "Depool db2112 T350142"
- Change db2112 weight to mimic the previous weight db2103:
sudo dbctl instance db2112 edit
- Apply outstanding schema changes to db2112 (if any)
- Update/resolve this ticket.