On Mon, Nov 20, 2017 at 11:12 AM, David Pacheco wrote:
> I understand if the community isn't interested in fixing this case if
> other users aren't seeing it much, but surely it's still a bug that this
> unusual case can result in a deadlock?
>
I've filed bug 14945 to cover this issue:
https://www.postgresql.org/message-id/20171204201055.27108.18283%40wrigleys.postgresql.org
While the underlying cause in this case was likely operator error (i.e.,
too little memory provisioned), the deadlock seems clearly like an issue
that could be hit by others and requires operator intervention to restore
service. I understand it might still be sufficiently rare, and the fix
sufficiently complex, that the community doesn't end up prioritizing it.
Thanks,
Dave
