Horton

Re: pgsql: Document XLOG_INCLUDE_XID a little better

Dilip Kumar <dilipbalaut@gmail.com>

View thread View on postgr.es
From: Dilip Kumar <dilipbalaut@gmail.com>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
Cc: Amit Kapila <amit.kapila16@gmail.com>, Robert Haas <robertmhaas@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-10-03T11:35:24Z
Lists: pgsql-hackers 
On Sat, Oct 2, 2021 at 8:10 PM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
>
> On 2021-Oct-02, Dilip Kumar wrote:
>
> > I have written two patches, Approach1 is as you described using a
> > static boolean and Approach2 as a local variable to XLogAssembleRecord
> > as described by Amit, attached both of them for your reference.
> > IMHO, either of these approaches looks cleaner.
>
> Thanks!  I haven't read these patches carefully, but I think the
> variable is about assigning the *subxid*, not the topxid.  Amit can
> confirm ...

IIRC, this variable is for logging the top xid in the first WAL by
each subtransaction. So that during logical decoding, while creating
the ReorderBufferTxn for the subtransaction we can associate it to the
top transaction without seeing the commit WAL.

-- 
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com

Commits

  1. Move MarkCurrentTransactionIdLoggedIfAny() out of the critical section.

  2. Replace XLOG_INCLUDE_XID flag with a more localized flag.

  3. Document XLOG_INCLUDE_XID a little better

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