Steps:
- Stop replication on pc1014.
- Merge CR to change pc3 primary to be pc1014: https://gerrit.wikimedia.org/r/c/operations/mediawiki-config/+/787497
- Deploy: scap sync-file wmf-config/ProductionServices.php "Set pc1014 as pc3 primary T307101"
- Downtime pc3: sudo cookbook sre.hosts.downtime --hours 1 -r "Rebooting pc1013 T307101" 'A:db-section-pc3'
- Reboot pc1013: SKIP_DBCTL=1 SKIP_START_SLAVE=1 ~kormat/bin/reboot-host T303174 pc1013.eqiad.wmnet
- Revert previous CR
- Deploy: scap sync-file wmf-config/ProductionServices.php "Set pc1013 as pc3 primary T307101"
- Start replication on pc1014.
Note that pc2013 will stay trying to replicate from pc1013 during this time.
Afterwards:
- Move pc1014 back to pc1
- Truncate all tables on pc1014