Re: PATCH: logical_work_mem and logical streaming of large in-progress transactions

On Sat, Mar 28, 2020 at 03:29:34PM +0530, Amit Kapila wrote:
>On Sat, Mar 28, 2020 at 2:19 PM Dilip Kumar <dilipbalaut(at)gmail(dot)com> wrote:
>>
>> On Sat, Mar 28, 2020 at 11:56 AM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>> >
>> >
>> > I have looked at the solution proposed and I would like to share my
>> > findings. I think calling ProcArrayApplyXidAssignment for each
>> > subtransaction is not a good idea for a couple of reasons:
>> > (a) It will just beat the purpose of maintaining KnowAssignedXids
>> > array which is to avoid looking at pg_subtrans in
>> > TransactionIdIsInProgress() on standby. Basically, if we remove it
>> > for each subXid, it will consider the KnowAssignedXids to be
>> > overflowed and check pg_subtrans frequently.
>>
>> Right, I also think this is a problem with this solution. I think we
>> may try to avoid this by caching this information. But, then we will
>> have to maintain this in some dimensional array which stores
>> sub-transaction ids per top transaction or we can maintain a list of
>> sub-transaction for each transaction. I haven't thought about how
>> much complexity this solution will add.
>>
>
>How about if instead of writing an XLOG_XACT_ASSIGNMENT WAL, we set a
>flag in TransactionStateData and then log that as special information
>whenever we write next WAL record for a new subtransaction? Then
>during recovery, we can only call ProcArrayApplyXidAssignment when we
>find that special flag is set in a WAL record. One idea could be to
>use a flag bit in XLogRecord.xl_info. If that is feasible then the
>solution can work as it is now, without any overhead or change in the
>way we maintain KnownAssignedXids.
>

Ummm, how is that different from what the patch is doing now? I mean, we
only write the top-level XID for the first WAL record in each subxact,
right? Or what would be the difference with your approach?

Anyway, I think you're right the ProcArrayApplyXidAssignment call was
done too early, but I think that can be fixed by moving it until after
the RecordKnownAssignedTransactionIds call, no? Essentially, right
before rm_redo().

You're right calling ProcArrayApplyXidAssignment() may be an issue,
because it exclusively acquires the ProcArrayLock. I've actually hinted
that might be an issue in my original message, suggesting we might add a
local cache of assigned XIDs (a small static array, doing essentially
the same thing we used to do on the upstream node). I haven't done that
in my WIP patch to keep it simple, but AFACS it'd work.

regards

--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services