Document timeout option of pg-upgrade command
Summary
The timeout option of gitlab-ctl pg-upgrade
is not documented. This option is required for upgrading large databases.
Details
The timeout option is not documented in https://docs.gitlab.com/ee/administration/postgresql/replication_and_failover.html#upgrading-postgresql-major-version-in-a-patroni-cluster. The use of this would have prevented failures to upgrade for multiple customers. We need to expand documentation to include the option, so that customers are aware before attempting an upgrade of any instance likely to run into the default limit (read, larger & distributed systems)
Once gitlab-ctl pg-upgrade
fails due to the timeout, the revert behaviors also fail. Failover then further has issues, due to the instance performing the step no longer being available.
Actionable
-
Expand gitlab-ctl pg-upgrade
documentation to include reference to the timeout options, and their default values -
(possible) Expand guidance to include recommendations of reasonable timeouts for larger instances