Обсуждение: Help with SI buffer overflow error

Hello, I have a nightly cron script that runs: vacuumdb -a -z. The output
of the cron script is emailed to me every night. For the last several
weeks, about 50% of the time I get the following error at least once but
sometimes more:

NOTICE: RegisterSharedInvalid: SI buffer overflow
NOTICE: InvalidateSharedInvalid: cache state reset

I am wondering if this is just something that will happen sometimes, or if
it implies some type of problem. I have not noticed any problems in using
the system. Any input would be appreciated.

I am running PG7.0.3 from RPMs (I plan on upgrading to 7.1.2 soon) on RedHat
7.0, 800Mhz Athlon w/ 256M. Most of the databases are reasonably small (a
few Meg) but some of them are over 1Gig.

The Postmaster command is: /usr/bin/postmaster -i -B 5000 -N 48 -o -S 16384

The output of the vacuum command looks like:

Vacuuming template1
VACUUM
... < Several more databases > ...
Vacuuming postgres
NOTICE: RegisterSharedInvalid: SI buffer overflow
NOTICE: InvalidateSharedInvalid: cache state reset
VACUUM
Vacuuming jpplanning
VACUUM
... < Several more databases> ...
Vacuuming OEA
NOTICE: RegisterSharedInvalid: SI buffer overflow
NOTICE: InvalidateSharedInvalid: cache state reset
VACUUM
Vacuuming nutil
VACUUM
Vacuuming feedback
VACUUM
Vacuuming ctlno
VACUUM
0.21user 0.08system 21:37.58elapsed 0%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (5074major+1093minor)pagefaults 0swaps

Matt O'Connor

Matthew <matt@ctlno.com> writes:
> NOTICE: RegisterSharedInvalid: SI buffer overflow
> NOTICE: InvalidateSharedInvalid: cache state reset

These are normal; at most they suggest that you've got another backend
sitting around doing nothing (but in an open transaction) while VACUUM
runs.

I think we finally got around to downgrading them to DEBUG messages
for 7.2.
        regards, tom lane