On Thu, 2008-09-11 at 15:42 +0300, Heikki Linnakangas wrote:
> One problem with this, BTW, is that if there's a continuous stream of
> medium-length transaction in the slave, each new snapshot taken will
> prevent progress in the WAL replay, so the WAL replay will advance in
> "baby steps", and can fall behind indefinitely.
Why would it fall behind indefinitely ? It only should fall behind to
the "blocking horizon", which should be the start of the longest
currently running transaction... which should be continually advancing
and not too far in the past if there are only medium length transactions
involved. Isn't normal WAL recovery also doing baby-steps, one WAL
record a time ? ;-)
Cheers,
Csaba.