Page MenuHomePhabricator

Document best-practice for hinted-handoff
Open, LowPublic

Description

Update and improve our Cassandra documentation to reflect lessons learned from https://wikitech.wikimedia.org/wiki/Incidents/2022-08-10_cassandra_disk_space.

  • When & how to disable hinted-handoff (maintenance windows, exceptional outage scenarios)
  • Why, when & how to truncate hints volume
  • Guidance on sizing a hints volume when provisioning new clusters
  • Using a dedicated hints volume when provisioning
NOTE: Where applicable, update Icinga notification description with link to relevant documentation

See also: T314941: RESTBase Cassandra high utilization alarms (instance-data)