pgsql: Add pg_partition_root to display top-most parent of a partition

On 2019-Feb-08, Michael Paquier wrote:

Add pg_partition_root to display top-most parent of a partition tree

This is useful when looking at partition trees with multiple layers, and
combined with pg_partition_tree, it provides the possibility to show up
an entire tree by just knowing one member at any level.

I noticed days ago that if you call pg_partition_root on the topmost
partitioned table, the server crashes :-)

--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On Fri, Mar 22, 2019 at 12:26:12AM -0300, Alvaro Herrera wrote:

I noticed days ago that if you call pg_partition_root on the topmost
partitioned table, the server crashes :-)

It's when you think that the thing is actually done that another issue
pops up. The attached fixes the issue, I suggest to return the
top-most parent as result if the input is the top-most parent itself.
Returning NULL does not make sense as in this case the relation can be
part of a partition tree.
--
Michael

Hi,

On 2019/03/22 12:55, Michael Paquier wrote:

On Fri, Mar 22, 2019 at 12:26:12AM -0300, Alvaro Herrera wrote:

I noticed days ago that if you call pg_partition_root on the topmost
partitioned table, the server crashes :-)

I thought we already fixed that last month, but that was pg_partition_tree().

It's when you think that the thing is actually done that another issue
pops up. The attached fixes the issue, I suggest to return the
top-most parent as result if the input is the top-most parent itself.
Returning NULL does not make sense as in this case the relation can be
part of a partition tree.

Thanks for writing the patch.

/* Fetch the top-most parent */
ancestors = get_partition_ancestors(relid);

Maybe, the patch should update this comment to say "Fetch the list of
ancestors".

+
+	/*
+	 * If the listed partition is already the top-most parent, just
+	 * return itself.
+	 */

What does "listed partition" mean? Maybe, you mean "If the input
relation..."? How about write this as:

If the input relation is itself the top-most parent, just return itself.

Regards,
Amit

On Fri, Mar 22, 2019 at 01:09:23PM +0900, Amit Langote wrote:

/* Fetch the top-most parent */
ancestors = get_partition_ancestors(relid);

Maybe, the patch should update this comment to say "Fetch the list of
ancestors".

Perhaps so. Note that actually the list is fetched to grab the
top-most parent from it, so it sort of makes sense?

What does "listed partition" mean? Maybe, you mean "If the input
relation..."? How about write this as:

Input relation sounds better.
--
Michael

Sorry about the messed up subject string of my previous reply. I failed
to notice that our internal mailing software occasionally adds that to
email headers before the emails get to my machine. I've been asked to be
careful before, but I didn't notice it again today. :-(

Thanks,
Amit

On 2019/03/22 13:12, Michael Paquier wrote:

On Fri, Mar 22, 2019 at 01:09:23PM +0900, Amit Langote wrote:

/* Fetch the top-most parent */
ancestors = get_partition_ancestors(relid);

Maybe, the patch should update this comment to say "Fetch the list of
ancestors".

Perhaps so. Note that actually the list is fetched to grab the
top-most parent from it, so it sort of makes sense?

It looked fine before, but the new lines added by patch makes it look
wrong/misplaced somehow.

Thanks,
Amit

On Fri, Mar 22, 2019 at 01:28:19PM +0900, Amit Langote wrote:

It looked fine before, but the new lines added by patch makes it look
wrong/misplaced somehow.

Okay, what do you think about the attached then?
--
Michael

On 2019/03/22 14:13, Michael Paquier wrote:

On Fri, Mar 22, 2019 at 01:28:19PM +0900, Amit Langote wrote:

It looked fine before, but the new lines added by patch makes it look
wrong/misplaced somehow.

Okay, what do you think about the attached then?

Thanks, looks fine.

Regards,
Amit

On Fri, Mar 22, 2019 at 04:05:20PM +0900, Amit Langote wrote:

Thanks, looks fine.

OK, committed.
--
Michael