We recently upgraded from PostgreSQL13.5 to 13.6 and since then have seen a pattern of spiky write IOPS activity. The database receives batch updates once daily and has very little query traffic. We are experiencing these spikes even when there is no user activity on the database. The data updates that occur are almost entirely additive -- that is, there are little to no updates or deletions. The cluster has a single instance with no replicas.
Given the timing of this behavior, it seems likely it is connected to the upgrade as nothing else has changed. An image of the CloudWatch Write IOPS stats can be found here: https://postimg.cc/fVWjbTdn
AWS OFICIALAtualizada há 2 anos
I'm seeing the same behavior on a freshly created aurora postgres 10.20 instance. Were you able to find anything else out about what is going on?