Horton

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

Amit Kapila <amit.kapila16@gmail.com>

View thread
From: Amit Kapila <amit.kapila16@gmail.com>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Dilip Kumar <dilipbalaut@gmail.com>, Michael Paquier <michael@paquier.xyz>, Tomas Vondra <tomas.vondra@2ndquadrant.com>, Peter Eisentraut <peter.eisentraut@2ndquadrant.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2019-12-13T09:16:20Z
Lists: pgsql-hackers

Commits

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

  1. Tighten the concurrent abort check during decoding.

  2. Improve hash_create()'s API for some added robustness.

  3. Use HASH_BLOBS for xidhash.

  4. Fix initialization of RelationSyncEntry for streaming transactions.

  5. Remove unused function declaration in logicalproto.h.

  6. Add additional tests to test streaming of in-progress transactions.

  7. Fix inline marking introduced in commit 464824323e.

  8. Add support for streaming to built-in logical replication.

  9. Fix the SharedFileSetUnregister API.

  10. Fix comment in procarray.c

  11. Suppress compiler warning in non-cassert builds.

  12. Extend the BufFile interface.

  13. Mark a few logical decoding related variables with PGDLLIMPORT.

  14. Implement streaming mode in ReorderBuffer.

  15. Extend the logical decoding output plugin API with stream methods.

  16. WAL Log invalidations at command end with wal_level=logical.

  17. Immediately WAL-log subtransaction and top-level XID association.

  18. Allow logical replication to transfer data in binary format.

  19. Only superuser can set sslcert/sslkey in postgres_fdw user mappings

  20. Track statistics for spilling of changes from ReorderBuffer.

  21. Add logical_decoding_work_mem to limit ReorderBuffer memory usage.

  22. logical decoding: process ASSIGNMENT during snapshot build

  23. Emit invalidations to standby for transactions without xid.

On Wed, Dec 11, 2019 at 11:46 PM Robert Haas <robertmhaas@gmail.com> wrote:
>
> On Mon, Dec 2, 2019 at 3:32 AM Dilip Kumar <dilipbalaut@gmail.com> wrote:
> > I have rebased the patch set on the latest head.
>
> 0001 looks like a clever approach, but are you sure it doesn't hurt
> performance when many small XLOG records are being inserted? I think
> XLogRecordAssemble() can get pretty hot in some workloads.
>
> With regard to 0002, logging a separate WAL record for each
> invalidation seems painful; I think most operations that generate
> invalidations generate a bunch of them all at once. Perhaps you could
> just queue up invalidations as they happen, and then force anything
> that's been queued up to be emitted into WAL just before you emit any
> WAL record that might need to be decoded.
>

I feel we can log the invalidations of the entire command at one go if
we log at CommandEndInvalidationMessages.  We already have all the
invalidations of current command in
transInvalInfo->CurrentCmdInvalidMsgs.  This can save us the effort of
maintaining a new separate list/queue for invalidations and to a good
extent, it will ameliorate your concern of logging each invalidation
separately.

>
> 0006 contains lots of XXX comments that look like real issues. I guess
> those need to be fixed. Also, why don't we do the thing that the
> commit message for 0006 says we could "theoretically" do? I don't
> understand why we need the k-way merge at all,
>

I think we can do what is written in the commit message, but then we
need to maintain two paths (one for streaming contexts and other for
non-streaming contexts) unless we want to entirely get rid of storing
subtransaction changes separately which seems like a more fundamental
change.  Right now, also to some extent such things are there, but I
have already given a comment to minimize it.  Having said that, I
think we can go either way.  I think the original intention was to
avoid doing more stuff unless it is really required as this is already
a big patchset, but maybe Tomas has a different idea about this.

-- 
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com