From: | Michael Paquier <michael(dot)paquier(at)gmail(dot)com> |
---|---|
To: | Simon Riggs <simon(at)2ndquadrant(dot)com> |
Cc: | Stas Kelvich <s(dot)kelvich(at)postgrespro(dot)ru>, Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Speedup twophase transactions |
Date: | 2016-01-12 06:35:55 |
Message-ID: | CAB7nPqQ=oVpjQvWnzvRa9_TuhSdK0oXESG87987Qq_f0vmEWeg@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Tue, Jan 12, 2016 at 4:57 AM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
> Performance tests for me show that the patch is effective; my results match
> Jesper's roughly in relative numbers.
>
> My robustness review is that the approach and implementation are safe.
>
> It's clear there are various additional tuning opportunities, but the
> objective of the current patch to improve performance is very, very clearly
> met, so I'm aiming to commit *this* patch soon.
- /* initialize LSN to 0 (start of WAL) */
- gxact->prepare_lsn = 0;
+ /* initialize LSN to InvalidXLogRecPtr */
+ gxact->prepare_start_lsn = InvalidXLogRecPtr;
+ gxact->prepare_end_lsn = InvalidXLogRecPtr;
I think that it would be better to explicitly initialize gxact->ondisk
to false here.
+ xlogreader = XLogReaderAllocate(&logical_read_local_xlog_page, NULL);
+ if (!xlogreader)
+ ereport(ERROR,
+ (errcode(ERRCODE_OUT_OF_MEMORY),
+ errmsg("out of memory"),
+ errdetail("Failed while allocating an
XLog reading processor.")));
Depending on something that is part of logical decoding to decode WAL
is not a good idea. If you want to move on with this approach, you
should have a dedicated function. Even better, it would be nice to
come up with a generic function used by both 2PC and logical decoding.
--
Michael
From | Date | Subject | |
---|---|---|---|
Next Message | Michael Paquier | 2016-01-12 06:41:29 | Re: Speedup twophase transactions |
Previous Message | Michael Paquier | 2016-01-12 05:58:29 | Re: PATCH: add pg_current_xlog_flush_location function |