On Tue, Sep 14, 2010 at 12:35 AM, Heikki Linnakangas
<heikki.linnakangas@enterprisedb.com> wrote:
> Hmm, looking at this more closely, I'm a bit confused. We already rename
> away backup_label at the beginning of recovery. Under what circumstances do
> we have a problem? This starts to seem like a complete non-issue to me.
A checkpoint record is read before backup_label file is renamed, so
the problem happens if shutdown is requested while the standby server
is waiting for the WAL containing a checkpoint record to arrive from
the master. This happened some times in my box when testing HS/SR.
To avoid the problem, we should rename backup_label before reading
any WAL record?
Regards,
--
Fujii Masao
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center