"Cory Vokey" <cvokey@rim.com> writes:
> Hi all, I'm seeing a checkpoint process lingering around for ~ 24hrs on
> one of my test systems and WAL logs are accumulating. It's running
> postgres 7.4.6.
Is the checkpoint process actually doing anything, or just sitting?
(strace'ing it would reveal this) If it seems to be just sitting,
could you attach to it with gdb and get a stack trace to find out
where it's sitting?
The only known cause of checkpoints not proceeding in 7.4.* is that
long btree CREATE INDEX operations can block them. But 24 hours
sounds pretty long for a CREATE INDEX too ...
regards, tom lane