using index "pg_toast_..." despite IgnoreSystemIndexes

[l***@nym.hush.com]

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

[Tom Lane]

Post by l***@nym.hush.com
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.

Post by l***@nym.hush.com
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