Alert for generating XLOG too fast

I received an alert indicating that Postgres was generating XLOG too quickly.

XLOG is being generated at a rate of 1.4016089244444443e+07 bytes/s on db3.cluster.gitlab.com:9187

The alert links to the postgresql runbook but there is no section of that runbook that corresponds to the alert.

In trying to figure out what was happening, I found there was a jump in the db servers' I/O utilization, iowait, and performed checkpoints. However there didn't appear to be a correlating change in the other parts of the application. GitLab.com was up and running quite fast at the time.

The alert cleared, but we should investigate this further and at the very least create a runbook of some sort.

The monitoring data from the time of the rise in IO can be found at the performance site.

cc/ @gl-infra