Christophe Pettus <xof(at)thebuild(dot)com> writes:
> The other observation is that the startup process is holding a *lot* of locks:
Hmm. Is it possible that the process is replaying the abort of a
transaction with a lot of subtransactions? It seems like maybe
you could be getting into an O(N^2) situation, though for N in
the low thousands it's hard to see that loop taking so long that
you'd think things were stuck.
> # select mode, count(*) from pg_locks where pid=5882 group by mode;
> AccessExclusiveLock | 7133
Is that number changing at all?
regards, tom lane