On 9/14/18 8:43 AM, Jeremy Schneider wrote:
Hi Chris - this is an interesting one that we do see from time to time; seems worth responding here as actually our best understanding right now is that this is something in community code, not AWS-specific.
On 9/13/18 16:10, Adrian Klaver wrote:The thing is, what you are doing ("(e.g. reboot, changing instance size, etc.)") are instance operations not database operations. That comes under AWS's purview.Correct, managing reboots and hardware reconfigurations would be the responsibility of AWS. However Chris' issue here is just that PostgreSQL itself took a long time to shut down. I'm not aware of anything RDS-specific with this.
The thing is I do not remember any posts to this list mentioning the same problem on a platform outside RDS. A quick search seems to confirm that.
Would it be possible to see the RDS shutdown script?
I don't know about this specific incident, but I do know that the RDS team has seen cases where a backend gets into a state (like a system call) where it's not checking signals and thus doesn't receive or process the postmaster's request to quit. We've seen these processes delay shutdowns and also block recovery on streaming replicas.
The particulars of that state?
FYI, yes there is a timeout on the RDS side. The basic workflow is to try to shutdown postgres the normal way, and if it hasn't cleanly shut down after a period of time then forcefully kill it.
Hope this helps, Jeremy
-- Adrian Klaver adrian.klaver@xxxxxxxxxxx
