Обсуждение: using index "pg_toast_..." despite IgnoreSystemIndexes

Hi

After an upgrade to 8.4.12 from 8.4.10 we vacuum/analyzed the db.
Postgres is running in standalone mode at this point.


vacuumdb --echo --analyze --all --verbose -U sysdba

07/31/12 04:09:57    INFO:  analyzing
"information_schema.sql_features"
07/31/12 04:09:57    INFO:  "sql_features": scanned 7 of 7 pages,
containing 649 live rows and 0 dead rows; 649 rows in sample, 649
estimated total rows
07/31/12 04:09:57    WARNING:  using index "pg_toast_2619_index"
despite IgnoreSystemIndexes
07/31/12 04:09:57    WARNING:  using index "pg_toast_2619_index"
despite IgnoreSystemIndexes
07/31/12 04:09:57    WARNING:  using index "pg_toast_2619_index"
despite IgnoreSystemIndexes
07/31/12 04:09:57    WARNING:  using index "pg_toast_2619_index"
despite IgnoreSystemIndexes
07/31/12 04:09:57    INFO:  vacuuming
"information_schema.sql_implementation_info"
07/31/12 04:09:57    INFO:  "sql_implementation_info": found 0
removable, 12 nonremovable row versions in 1 out of 1 pages
07/31/12 04:09:57    DETAIL:  0 dead row versions cannot be removed
yet.
07/31/12 04:09:57    There were 1 unused item pointers.
07/31/12 04:09:57    0 pages are entirely empty.
07/31/12 04:09:57    CPU 0.00s/0.00u sec elapsed 0.00 sec.
07/31/12 04:09:57    INFO:  vacuuming "pg_toast.pg_toast_11452"


I see the warning generated in systable_beginscan_ordered as a
warning, and then it proceeds to do the work anyway.

It appears as if this is benign.  Is that the case?

Thanks

lsq@nym.hush.com writes:
> After an upgrade to 8.4.12 from 8.4.10 we vacuum/analyzed the db.
> Postgres is running in standalone mode at this point.

Why were you using standalone mode?  And why were you using
ignore_system_indexes?  This whole procedure seems like overkill
for a routine minor-version update.

> 07/31/12 04:09:57    WARNING:  using index "pg_toast_2619_index"
> despite IgnoreSystemIndexes

> I see the warning generated in systable_beginscan_ordered as a
> warning, and then it proceeds to do the work anyway.
> It appears as if this is benign.  Is that the case?

It is unless you have some reason to think that that index is corrupt...

            regards, tom lane