On Mon, Oct 4, 2021 at 8:19 PM Mark Dilger <mark.dilger@enterprisedb.com> wrote:
> I am changing pg_amcheck to filter out indexes as you say. Since the btree check should no longer error in these
cases,the issue of pg_amcheck exit(2) sorts itself out without further code changes.
Cool.
> I am changing verify_heapam to skip unlogged tables during recovery. In testing, checking such a table results in a
simplenotice:
>
> NOTICE: cannot verify unlogged relation "u_tbl" during recovery, skipping
That makes sense to me.
> While testing, I also created an index on the unlogged table and checked that index using bt_index_parent_check, and
wassurprised that checking it using bt_index_parent_check raises an error:
>
> ERROR: cannot acquire lock mode ShareLock on database objects while recovery is in progress
> HINT: Only RowExclusiveLock or less can be acquired on database objects during recovery.
Calling bt_index_parent_check() in hot standby mode is kind of asking
for it to error-out. It requires a ShareLock on the relation, which is
inherently not possible during recovery. So I don't feel too badly
about letting it just happen.
> So I am changing pg_amcheck to filter out indexes when pg_is_in_recovery() is true and relpersistence='u'. Does that
soundright to you?
Yes, that all sounds right to me.
Thanks
--
Peter Geoghegan