Horton

Re: libpq debug log

Kyotaro Horiguchi <horikyota.ntt@gmail.com>

View thread View on postgr.es
From: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
To: iwata.aya@fujitsu.com
Cc: alvherre@alvh.no-ip.org, tgl@sss.pgh.pa.us, tsunakawa.takay@fujitsu.com, k.jamison@fujitsu.com, pgsql-hackers@lists.postgresql.org
Date: 2021-04-02T05:45:13Z
Lists: pgsql-hackers 
At Fri, 02 Apr 2021 14:40:09 +0900 (JST), Kyotaro Horiguchi <horikyota.ntt@gmail.com> wrote in 
> At Fri, 2 Apr 2021 02:56:44 +0000, "iwata.aya@fujitsu.com" <iwata.aya@fujitsu.com> wrote in 
> > Hi Alvaro san
> > 
> > Thank you for your fix of trace log code.
> > 
> > > From: 'alvherre@alvh.no-ip.org' <alvherre@alvh.no-ip.org>
> > > Sent: Friday, April 2, 2021 11:30 AM
> > ...
> > > It still didn't fix it!  Drongo is now reporting a difference in the expected trace --
> > > and the differences all seem to be message lengths.
> > > Now that is pretty mysterious, because the messages themselves are printed
> > > identically.  Perl's output is pretty unhelpful, but I wrote them to a file and diffed
> > > manually; it's attached.
> >
> > Do ErrorResponse and NoticeResponse vary from test to test ...?
> > If so, it seemed difficult to make the trace logs available for regression test. 
> > I will also investigate the cause of this issue.
> 
> The redacted fields, F, L and R contained source file, souce line and
> source function respectively. It is reasonable guess that the
> difference comes from them but I'm not sure how they make a difference
> of 50 bytes in length...
> 
> Anyway if the length is wrong, we should get an error after emitting
> the log line.
> 
> >	if (logCursor - 1 != length)
> >		fprintf(conn->Pfdebug,
> >				"mismatched message length: consumed %d, expected %d\n",
> >				logCursor - 1, length);
> 
> So, the cheapest measure for regression test would be just making the

So, the cheapest measure for regression test would be just *masking* the

> length field, while regress is true...

tired?

-- 
Kyotaro Horiguchi
NTT Open Source Software Center

Commits

  1. Rename PQtraceSetFlags() to PQsetTraceFlags().

  2. Suppress length of Notice/Error msgs in PQtrace regress mode

  3. Strip file names reported in error messages on Windows, too.

  4. Fix setvbuf()-induced crash in libpq_pipeline

  5. libpq_pipeline: Must strdup(optarg) to avoid crash

  6. Remove setvbuf() call from PQtrace()

  7. Initialize conn->Pfdebug to NULL when creating a connection

  8. Disable force_parallel_mode in libpq_pipeline

  9. libpq_pipeline: add PQtrace() support and tests

  10. Improve PQtrace() output format

  11. Re-simplify management of inStart in pqParseInput3's subroutines.