Horton
Menu

RE: Perform streaming logical transactions by background workers and parallel apply

Hayato Kuroda (Fujitsu) <kuroda.hayato@fujitsu.com>

View thread
From: "kuroda.hayato@fujitsu.com" <kuroda.hayato@fujitsu.com>
To: "houzj.fnst@fujitsu.com" <houzj.fnst@fujitsu.com>
Cc: Amit Kapila <amit.kapila16@gmail.com>, Dilip Kumar <dilipbalaut@gmail.com>, Masahiko Sawada <sawada.mshk@gmail.com>, "shiy.fnst@fujitsu.com" <shiy.fnst@fujitsu.com>, Peter Smith <smithpb2250@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, "wangw.fnst@fujitsu.com" <wangw.fnst@fujitsu.com>
Date: 2022-09-29T09:50:01Z
Lists: pgsql-hackers 
Dear Hou,

Thanks for updating patch. I will review yours soon, but I reply to your comment.

> > 04. applyparallelworker.c - LogicalParallelApplyLoop()
> >
> > ```
> > +               shmq_res = shm_mq_receive(mqh, &len, &data, false);
> > ...
> > +               if (ConfigReloadPending)
> > +               {
> > +                       ConfigReloadPending = false;
> > +                       ProcessConfigFile(PGC_SIGHUP);
> > +               }
> > ```
> >
> >
> > Here the parallel apply worker waits to receive messages and after dispatching
> > it ProcessConfigFile() is called.
> > It means that .conf will be not read until the parallel apply worker receives new
> > messages and apply them.
> >
> > It may be problematic when users change log_min_message to debugXXX for
> > debugging but the streamed transaction rarely come.
> > They expected that detailed description appears on the log from next
> > streaming chunk, but it does not.
> >
> > This does not occur in leader worker when it waits messages from publisher,
> > because it uses libpqrcv_receive(), which works asynchronously.
> >
> > I 'm not sure whether it should be documented that the evaluation of GUCs may
> > be delayed, how do you think?
> 
> I changed the shm_mq_receive to asynchronous mode which is also consistent
> with
> what we did for Gather node when reading data from parallel query workers.

I checked your implementation, but it seemed that the parallel apply worker will not sleep
even if there are no messages or signals. It might be very inefficient.

In gather node - gather_readnext(), the same way is used, but I think there is a premise
that the wait-time is short because it is related with only one gather node.
In terms of parallel apply worker, however, we cannot predict the wait-time because
it is related with the streamed transactions. If such transactions rarely come, parallel apply workers may spend many CPU time.

I think we should wait during short time or until leader notifies, if shmq_res == SHM_MQ_WOULD_BLOCK.
How do you think?


Best Regards,
Hayato Kuroda
FUJITSU LIMITED

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Fix invalid memory access during the shutdown of the parallel apply worker.

  2. Fix assertion failure in apply worker.

  3. Use elog to report unexpected action in handle_streamed_transaction().

  4. Use appropriate wait event when sending data in the apply worker.

  5. Allow the logical_replication_mode to be used on the subscriber.

  6. Rename GUC logical_decoding_mode to logical_replication_mode.

  7. Display the leader apply worker's PID for parallel apply workers.

  8. Improve the code to decide and process the apply action.

  9. Document the newly added wait events added by commit 216a784829.

  10. Perform apply of large transactions by parallel workers.

  11. Wake up a subscription's replication worker processes after DDL.

  12. Add copyright notices to meson files

  13. Better document logical replication parameters

  14. Add a common function to generate the origin name.

  15. Harmonize parameter names in storage and AM code.

  16. Avoid using list_length() to test for empty list.

  17. Improve two comments related to a boolean DefElem's value

  18. Fix partition table's REPLICA IDENTITY checking on the subscriber.

  19. Fix data inconsistency between publisher and subscriber.

  20. Fix cache look-up failures while applying changes in logical replication.