On Sat, Apr 13, 2024 at 4:37 PM Alvaro Herrera wrote:
>
> Hmm, but shouldn't we cancel the sleep after we have completed sleeping
> altogether, that is, until we've determined that we're no longer to
> sleep waiting for this slot? That would suggest to put the call to
> cancel sleep after the for
On 2024-Apr-13, Bharath Rupireddy wrote:
> It looks like there's missing ConditionVariableCancelSleep() in
> InvalidatePossiblyObsoleteSlot() after waiting for the replication
> slot to be released by another process. Although prepare to sleep
> cancels the sleep if the previous one wasn't cancele
