Andres Freund wrote:
> I think this isn't particularly pretty, but it seems to be working well
> enough, and changing it would be pretty invasive. So keeping in line
> with all that code seems to be easier.
OK, I'm convinced with this part to remove the call of
LockSharedObjectForSession that uses dontWait and replace it by a loop
in ResolveRecoveryConflictWithDatabase.
> Note that InRecovery doesn't mean what you probably think it means:
> [stuff]
> bool InRecovery = false;
Yes, right. I misunderstood with RecoveryInProgress().
> The assertion actually should be even stricter:
> Assert(!InRecovery || (sessionLock && dontWait));
> i.e. we never acquire a heavyweight lock in the startup process unless
> it's a session lock (as we don't have resource managers/a xact to track
> locks) and we don't wait (as we don't have the deadlock detector
> infrastructure set up).
No problems with this assertion here.
--
Michael
