Обсуждение: Re: [COMMITTERS] pgsql: Fix pg_receivexlog --synchronous

Simon Riggs <simon@2ndQuadrant.com> writes:
> Fix pg_receivexlog --synchronous

The buildfarm says you broke the 9.5 branch.

In general, pushing inessential patches just a few hours before a wrap
deadline is a dangerous business.  Pushing them without any testing
is close to irresponsible.
        regards, tom lane

On Mon, Aug 29, 2016 at 8:34 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Simon Riggs <simon@2ndQuadrant.com> writes:
>> Fix pg_receivexlog --synchronous
>
> The buildfarm says you broke the 9.5 branch.
>
> In general, pushing inessential patches just a few hours before a wrap
> deadline is a dangerous business.  Pushing them without any testing
> is close to irresponsible.

This area of the code has faced some refactoring from Magnus lately,
so you need this on REL9_5_STABLE:
--- a/src/bin/pg_basebackup/receivelog.c
+++ b/src/bin/pg_basebackup/receivelog.c
@@ -534,7 +534,7 @@ ReceiveXlogStream(PGconn *conn, XLogRecPtr
startpos, uint32 timeline,
    }
    else
    {
-       if (stream->synchronous)
+       if (synchronous)
            reportFlushPosition = true;
        else
            reportFlushPosition = false;
--
Michael

On 2016-08-29 07:34:52 -0400, Tom Lane wrote:
> Simon Riggs <simon@2ndQuadrant.com> writes:
> > Fix pg_receivexlog --synchronous
> 
> The buildfarm says you broke the 9.5 branch.
> 
> In general, pushing inessential patches just a few hours before a wrap
> deadline is a dangerous business.  Pushing them without any testing
> is close to irresponsible.

And the comment change doesn't actually seem an improvement, because it
makes it harder to understand why a slot forces this to be enabled.

Do we want to revert this until the release, or does somebody want to
push the fix?

Andres

Andres Freund <andres@anarazel.de> writes:
> Do we want to revert this until the release, or does somebody want to
> push the fix?

If this had broken the 9.6 branch I would have already summarily
reverted it.  Since it didn't, my only real concern vis-a-vis today's
release is that the build failure in 9.5 calls into question the
quality of the testing that happened in 9.6.  9.6 is still pretty
close to HEAD, but not so close that it's a good idea to push patches
you have not tested in that branch.
        regards, tom lane

On Mon, Aug 29, 2016 at 07:34:52AM -0400, Tom Lane wrote:
> Simon Riggs <simon@2ndQuadrant.com> writes:
> > Fix pg_receivexlog --synchronous
> 
> The buildfarm says you broke the 9.5 branch.
> 
> In general, pushing inessential patches just a few hours before a wrap
> deadline is a dangerous business.  Pushing them without any testing
> is close to irresponsible.

Not being around to fix the breakage after the commit isn't great
either.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com

+ As you are, so once was I. As I am, so you will be. +
+                     Ancient Roman grave inscription +

On 29 August 2016 at 12:34, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Simon Riggs <simon@2ndQuadrant.com> writes:
>> Fix pg_receivexlog --synchronous
>
> The buildfarm says you broke the 9.5 branch.
>
> In general, pushing inessential patches just a few hours before a wrap
> deadline is a dangerous business.  Pushing them without any testing
> is close to irresponsible.

Sorry about that everybody. Thanks to Alvaro for doing that in my absence.

I pushed to 9.5 because of a misunderstanding that the author was
saying to me they had also tested it for 9.5. It was not knowingly
untested, but responsibility and mistake was mine in not confirming
that with my own eyes before pushing.

-- 
Simon Riggs                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services