Re: improve performance of pg_dump with many sequences

On Wed, Jan 07, 2026 at 07:24:52PM -0500, Tom Lane wrote:
> Nathan Bossart <nathandbossart(at)gmail(dot)com> writes:
>> On Wed, Jan 07, 2026 at 06:13:48PM -0500, Tom Lane wrote:
>>> That would be a fine argument were it not that collectSequences()
>>> tries to vacuum up the data for every sequence in the DB, whether
>>> the user has asked to dump them all or not.
>
>> I meant that we could teach pg_dump to error in dumpSequenceData() if it
>> sees nulls for the sequence in question.
>
> Ah, gotcha; I thought you were talking about changing
> pg_get_sequence_data() to throw an error instead of returning nulls.

Here is a patch that does this along with what you described upthread,
i.e., teaching pg_get_sequence_data to return nulls for missing sequences.
Apparently pg_dump still runs through dumpSequenceData() for schema-only
dumps, which is a problem for this patch. I've taught it to immediately
return for schema-only dumps to evade this problem. That seems like a win
for older versions, too, as they will no longer run useless queries.

I believe this helps the reporter's case, as their problem involves dumping
one schema while dropping another, which v18 indeed makes worse because (as
you mentioned) we gather data for all sequences in the database.

--
nathan