On 2014-01-22 01:18:36 +0100, Simon Riggs wrote:
> > My understanding is that if it runs out of buffer space while in an
> > XLogInsert, it will be holding one or more buffer content locks exclusively,
> > and unless it can complete the xlog (or scrounge up the info to return that
> > buffer to its previous state), it can never release that lock. There might
> > be other paths were it could get by with an ERROR, but if no one can write
> > xlog anymore, all of those paths must quickly converge to the one that
> > cannot simply ERROR.
>
> Agreed. You don't say it but I presume you intend to point out that
> such long-lived contention could easily have a knock on effect to
> other read-only statements. I'm pretty sure other databases work the
> same way.
>
> Our choice are
>
> 1. Waiting
> 2. Abort transactions
> 3. Some kind of release-locks-then-wait-and-retry
>
> (3) is a step too far for me, even though it is easier than you say
> since we write WAL before changing the data block so a failure to
> insert WAL could just result in a temporary drop lock, sleep and
> retry.
>
> I would go for (1) waiting for up to checkpoint_timeout then (2), if
> we think that is a problem.
How are we supposed to wait while e.g. ProcArrayLock? Aborting
transactions doesn't work either, that writes abort records which can
get signficantly large.
Greetings,
Andres Freund
-- Andres Freund http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training &
Services