XLOG generation peak
All times in UTC.
- 5:39pm: PagerDuty alert: "Postgres is generating XLOG too fast, expect this to cause replication lag" https://gitlab.pagerduty.com/incidents/PGXBWQH
- 5:44pm: Peak corroborated. Already starting to trend down

Some other anormalities at the time:
- Postgres IO Utilization

- NFS timeout peak for nfs-file-12

- 5:49pm: Alert resolved. XLOG generation levels back to normal:

/cc @gl-infra
issue