Re: Re: [COMMITTERS] pgsql: Make standby server continuously retry restoring the next WAL

Aidan Van Dyk wrote:
> * Heikki Linnakangas <heikki.linnakangas@enterprisedb.com> [100211 09:17]:
> 
>> If the file is just being copied to the archive when restore_command
>> ('cp', say) is launched, it will copy a half file. That's not a problem
>> for PITR, because PITR will end at the end of valid WAL anyway, but
>> returning a half WAL file in standby mode is a problem.
> 
> But it can be a problem - without the last WAL (or at least enough of
> it) the master switched and archived, you have no guarantee of having
> being consistent again (I'm thinking specifically of recovering from a
> fresh backup)

You have to wait for the last WAL file required by the backup to be
archived before starting recovery. Otherwise there's no guarantee anyway.

>> We could well just document that you should do that, ie. make sure the
>> file appears in the archive atomically with the right size.
> 
> I have to admit, today was the first time I went and re-read the PITR
> docs, and no, the docs don't seem to talk about that... Maybe it was
> just plain obvious to me because it (the atomic apperance) is something
> unix devloppers have always had to deal with, so it's ingrained in me.
> But I'm *sure* that I've seen that bandied around as common knowledge on
> the lists, and one of the reasons we alway see warnings about using
> rsync instead of plain SCP, etc.
> 
> So ya, we should probably mention that somewhere in the docs.  Section
> 24.3.6. Caveats?

-1. it isn't necessary for PITR. It's a new requirement for
standby_mode='on', unless we add the file size check into the backend. I
think we should add the file size check to the backend instead and save
admins the headache.

--  Heikki Linnakangas EnterpriseDB   http://www.enterprisedb.com