Thread: Fwd: Administrative Processing for csitech: getrows

Fwd: Administrative Processing for csitech: getrows

From
Naomi Walker
Date:
Every night I VACUUM ANALYZE our databases.  I've asked in the past, and
was told not to worry about this.  One database never returns a
message.  This database consistently returns this message.  What odd thing
am I tripping over?

>Before Vacuum
>Sat May 11 02:00:54 MST 2002
>psql:/usr/local/postgres/etc/csitech/daily/sql/getrows.sql:49:
>NOTICE:  RegisterSharedInvalid: SI buffer overflow
>psql:/usr/local/postgres/etc/csitech/daily/sql/getrows.sql:49:
>NOTICE:  InvalidateSharedInvalid: cache state reset
>VACUUM
>Sat May 11 02:01:10 MST 2002
>After Vacuum


Re: Fwd: Administrative Processing for csitech: getrows

From
Tom Lane
Date:
Naomi Walker <nwalker@eldocomp.com> writes:
> This database consistently returns this message.  What odd thing
> am I tripping over?

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

These messages are completely, utterly ignorable.  We've downgraded
them to DEBUG level in 7.2, and may well eliminate them entirely
someday.

The reason they're there is that in ancient history (6.4 or so ;-))
the SI overflow code had a lot of bugs.  But AFAIK there is not any
longer a reason to worry.

(Now, if you are still running 6.anything, you are past due to upgrade.
If you're on 7.1 or later, don't worry about it...)

            regards, tom lane