xlog flus not satisfied
While doing
# VACUUM VERBOSE ANALYZE d_trr_dfh;
INFO: vacuuming "xmms.d_trr_dfh"
ERROR: xlog flush request 21F/9F57DF88 is not satisfied --- flushed only to
21F/924CE76C
CONTEXT: writing block 2919652 of relation 17461/17462/17668
I see this in the logs
user= CONTEXT: writing block 2919680 of relation 17461/17462/17668
user= WARNING: could not write block 2919680 of 17461/17462/17668
DETAIL: Multiple failures --- write error might be permanent.
user= LOG: checkpoint starting: time
xlog flush request 21F/9F67DA80 is not satisfied --- flushed only to
21F/924CE76C
Does this mean I have disk issues?
As background, this is a new box mirrored from a separate box via rsync.
I've basically copied/rsync the entire postgresql server and data files over
to create a mirror copy. After which, I've tried to do the vacuum and gotten
into the above trouble.
Any ideas from the PG community? (or is rsync not the way to go for this?)
"Ow Mun Heng" <ow.mun.heng@wdc.com> writes:
As background, this is a new box mirrored from a separate box via rsync.
I've basically copied/rsync the entire postgresql server and data files over
to create a mirror copy. After which, I've tried to do the vacuum and gotten
into the above trouble.
Did you shut down the old postmaster while mirroring its files? I could
believe seeing this type of problem as a consequence of getting
out-of-sync copies of different parts of the database.
regards, tom lane
-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
"Ow Mun Heng" <ow.mun.heng@wdc.com> writes:
As background, this is a new box mirrored from a separate box via rsync.
I've basically copied/rsync the entire postgresql server and data files
over
to create a mirror copy. After which, I've tried to do the vacuum and
gotten
into the above trouble.
Did you shut down the old postmaster while mirroring its files? I could
believe seeing this type of problem as a consequence of getting
out-of-sync copies of different parts of the database.
You've caught me. I'm actually planning to down the master server during
lunch to re-sync it.
Aside : I wonder how pgcluster does it then, cos I thought they use rsync to
do replication.
Thanks. Will report back in couple hours.