Re: Re: [COMMITTERS] pgsql: Properly set relpersistence for fake relcache entries.

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> Definitions aside, I think it's a pretty scary issue. It basically means that if you have a recovery (crash or archive) during which you read a buffer into memory, the buffer won't be checkpointed. So if, before the buffer is next evicted, you have a crash, and if at least one checkpoint has intervened between the most recent WAL-logged operation on the buffer and the crash, you're hosed. That's not a terribly unlikely scenario.

This is only an issue on standby slaves or when doing a PITR recovery, no?
As far as I can tell from the discussion, it would *not* affect crash
recovery, because we don't do restartpoints during crash recovery.

> While I can't claim to understand exactly what our standards for forcing an immediate minor release are, I think this is pretty darn bad. I certainly don't want my customers running with this for a minute longer than necessary, and I feel really bad for letting it get into a release, let alone go undetected for this long. :-(

There's been some discussion about it among -core already. The earliest
we could possibly do anything would be a release this coming week (that
is, wrap Thursday for release Monday 9/24). However, considering that
a lot of key people will be attending PG Open between now and Thursday,
I'm not sure how practical that really would be. Waiting a week might
be better, and it would give more time for initial bug reports against
9.2.0 to filter in.

regards, tom lane