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: Dilip Kumar <dilipbalaut@gmail.com>
Cc: Kuntal Ghosh <kuntalghosh.2007@gmail.com>, Tomas Vondra <tomas.vondra@2ndquadrant.com>, Michael Paquier <michael@paquier.xyz>, Peter Eisentraut <peter.eisentraut@2ndquadrant.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2020-05-19T09:04:33Z
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 Mon, May 18, 2020 at 5:57 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
>
>
> 3.
> + /*
> + * If streaming is enable and we have serialized this transaction because
> + * it had incomplete tuple.  So if now we have got the complete tuple we
> + * can stream it.
> + */
> + if (ReorderBufferCanStream(rb) && can_stream && rbtxn_is_serialized(toptxn)
> + && !(rbtxn_has_toast_insert(txn)) && !(rbtxn_has_spec_insert(txn)))
> + {
>
> This comment is just saying what you are doing in the if-check.  I
> think you need to explain the rationale behind it. I don't like the
> variable name 'can_stream' because it matches ReorderBufferCanStream
> whereas it is for a different purpose, how about naming it as
> 'change_complete' or something like that.  The check has many
> conditions, can we move it to a separate function to make the code
> here look clean?
>

Do we really need this?  Immediately after this check, we are calling
ReorderBufferCheckMemoryLimit which will anyway stream the changes if
required.  Can we move the changes related to the detection of
incomplete data to a separate function?

Another comments on v20-0010-Bugfix-handling-of-incomplete-toast-tuple:

+ else if (rbtxn_has_toast_insert(txn) &&
+ ChangeIsInsertOrUpdate(change->action))
+ {
+ toptxn->txn_flags &= ~RBTXN_HAS_TOAST_INSERT;
+ can_stream = true;
+ }
..
+#define ChangeIsInsertOrUpdate(action) \
+ (((action) == REORDER_BUFFER_CHANGE_INSERT) || \
+ ((action) == REORDER_BUFFER_CHANGE_UPDATE) || \
+ ((action) == REORDER_BUFFER_CHANGE_INTERNAL_SPEC_INSERT))

How can we clear the RBTXN_HAS_TOAST_INSERT flag on
REORDER_BUFFER_CHANGE_INTERNAL_SPEC_INSERT action?

IIUC, the basic idea used to handle incomplete changes (which is
possible in case of toast tuples and speculative inserts) is to mark
such TXNs as containing incomplete changes and then while finding the
largest top-level TXN for streaming, we ignore such TXN's and move to
next largest TXN.  If none of the TXNs have complete changes then we
choose the largest (sub)transaction and spill the same to make the
in-memory changes below logical_decoding_work_mem threshold.  This
idea can work but the strategy to choose the transaction is suboptimal
for cases where TXNs have some changes which are complete followed by
an incomplete toast or speculative tuple.  I was having an offlist
discussion with Robert on this problem and he suggested that it would
be better if we track the complete part of changes separately and then
we can avoid the drawback mentioned above.  I have thought about this
and I think it can work if we track the size and LSN of completed
changes.  I think we need to ensure that if there is concurrent abort
then we discard all changes for current (sub)transaction not only up
to completed changes LSN whereas if the streaming is successful then
we can truncate the changes only up to completed changes LSN. What do
you think?

I wonder why you have done this as 0010 in the patch series, it should
be as 0006 after the
0005-Implement-streaming-mode-in-ReorderBuffer.patch.  If we can do
that way then it would be easier for me to review.  Is there a reason
for not doing so?

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