tsearch - Regression tests fail

"Markus Wollny" <Markus.Wollny@computec.de> writes:

I tried the regression tests for contrib/tsearch as suggested, and this
leads to exactly the same symptoms as described with the real-life test.

Hm, it works here. What is your platform exactly? Can you provide a
stack trace from the core file that the crashed backend leaves?

regards, tom lane

Hi!

Platform is SuSE Linux 7.3 (Linux 2.4.10-64GB-SMP #1 SMP Tue Sep 25
12:36:09 GMT 2001 i686 unknown); I think it doesn't dump core, so I
cannot provide a stack trace. In the logfile it just says: "DEBUG:
server process (pid 22137) was terminated by signal 11". I cannot find
any core-files.

Regards,

Markus

Marcus,

we need version of postgresql and sample data for testing.

Oleg
On Mon, 29 Jul 2002, Markus Wollny wrote:

Hi!

Platform is SuSE Linux 7.3 (Linux 2.4.10-64GB-SMP #1 SMP Tue Sep 25
12:36:09 GMT 2001 i686 unknown); I think it doesn't dump core, so I
cannot provide a stack trace. In the logfile it just says: "DEBUG:
server process (pid 22137) was terminated by signal 11". I cannot find
any core-files.

Regards,

Markus

-----UrsprО©╫ngliche Nachricht-----
Von: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Gesendet: Montag, 29. Juli 2002 16:46
An: Markus Wollny
Cc: pgsql-general@postgresql.org; Oleg Bartunov
Betreff: Re: [GENERAL] tsearch - Regression tests fail

"Markus Wollny" <Markus.Wollny@computec.de> writes:

I tried the regression tests for contrib/tsearch as

suggested, and this

leads to exactly the same symptoms as described with the

real-life test.

Hm, it works here. What is your platform exactly? Can you provide a
stack trace from the core file that the crashed backend leaves?

regards, tom lane

Regards,
Oleg
_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83

"Markus Wollny" <Markus.Wollny@computec.de> writes:

Platform is SuSE Linux 7.3 (Linux 2.4.10-64GB-SMP #1 SMP Tue Sep 25
12:36:09 GMT 2001 i686 unknown); I think it doesn't dump core, so I
cannot provide a stack trace. In the logfile it just says: "DEBUG:
server process (pid 22137) was terminated by signal 11". I cannot find
any core-files.

Signal 11 is SIGSEGV on most Unixen (certainly including Linux), so the
backend is certainly crashing. If you are not getting core files in
the $PGDATA/base/yourdbnumber/ directory, then I'd bet that the
postmaster is being started under "ulimit -c 0" conditions. Restart
it with "ulimit -c unlimited" environment and you should get a core.

regards, tom lane

Hi!

You live, you learn - thanks for your help :) Version is PostgreSQL 7.2.
The backend crashed when I do "select txt2txtidx(text) from article
limit 1;". This is what I get when I do "gdb psql ./core":

Core was generated by `postgres: postgres pcgames [local] SELECT
'.
Program terminated with signal 11, Segmentation fault.
#0 0x40126853 in ?? ()
#1 0x50e280fb in ?? ()
#2 0x50e290a5 in ?? ()
#3 0x080c90bd in ?? ()
#4 0x080c9194 in ?? ()
#5 0x080c9720 in ?? ()
#6 0x080c9a3a in ?? ()
#7 0x080c9cd6 in ?? ()
#8 0x080c9dba in ?? ()
#9 0x080cfb77 in ?? ()
#10 0x080c8099 in ?? ()
#11 0x080d0978 in ?? ()
#12 0x080c8159 in ?? ()
#13 0x080c6fce in ?? ()
#14 0x080c65f7 in ?? ()
#15 0x0811796b in ?? ()
#16 0x08116421 in ?? ()
#17 0x081173d0 in ?? ()
#18 0x080fd0f0 in ?? ()
#19 0x080fc91f in ?? ()
#20 0x080fbac9 in ?? ()
#21 0x080fb635 in ?? ()
#22 0x080d95bd in ?? ()
#23 0x400cb7ee in ?? ()

The relevant cell (first row of article, column text) contains this
text: " Sie haben Spaß an PC-Spielen? Und am Schreiben? Sie
verf&uuml;gen zudem &uuml;ber gute Deutsch- und Englischkenntnisse und
streben eine Zukunft als Journalist an? Sie w&uuml;rden sich
dar&uuml;ber freuen, mit Ihren Artikeln monatlich &uuml;ber eine Million
Leser zu erreichen? Dann ist Ihr Platz vielleicht bei PC Games. Wer
Interesse an einem Arbeitsplatz bei uns hat, findet alles Weitere in
unserem Stellenangebot in der Special-Sektion - folgen Sie einfach dem
Link unter dieser Meldung."

Can you make something out of this?

Regards,

Markus

Hmm

just tried

select txt2txtidx('your text here....');

and it worked fine.

test=# select version();
version
---------------------------------------------------------------
PostgreSQL 7.2.1 on i686-pc-linux-gnu, compiled by GCC 2.95.3
(1 row)

I'd suggest to try 7.2.1 (or 7.2.2 from CVS REL7_2_STABLE)

Oleg
On Mon, 29 Jul 2002, Markus Wollny wrote:

Hi!

You live, you learn - thanks for your help :) Version is PostgreSQL 7.2.
The backend crashed when I do "select txt2txtidx(text) from article
limit 1;". This is what I get when I do "gdb psql ./core":

Core was generated by `postgres: postgres pcgames [local] SELECT
'.
Program terminated with signal 11, Segmentation fault.
#0 0x40126853 in ?? ()
#1 0x50e280fb in ?? ()
#2 0x50e290a5 in ?? ()
#3 0x080c90bd in ?? ()
#4 0x080c9194 in ?? ()
#5 0x080c9720 in ?? ()
#6 0x080c9a3a in ?? ()
#7 0x080c9cd6 in ?? ()
#8 0x080c9dba in ?? ()
#9 0x080cfb77 in ?? ()
#10 0x080c8099 in ?? ()
#11 0x080d0978 in ?? ()
#12 0x080c8159 in ?? ()
#13 0x080c6fce in ?? ()
#14 0x080c65f7 in ?? ()
#15 0x0811796b in ?? ()
#16 0x08116421 in ?? ()
#17 0x081173d0 in ?? ()
#18 0x080fd0f0 in ?? ()
#19 0x080fc91f in ?? ()
#20 0x080fbac9 in ?? ()
#21 0x080fb635 in ?? ()
#22 0x080d95bd in ?? ()
#23 0x400cb7ee in ?? ()

The relevant cell (first row of article, column text) contains this
text: " Sie haben SpaО©╫ an PC-Spielen? Und am Schreiben? Sie
verf&uuml;gen zudem &uuml;ber gute Deutsch- und Englischkenntnisse und
streben eine Zukunft als Journalist an? Sie w&uuml;rden sich
dar&uuml;ber freuen, mit Ihren Artikeln monatlich &uuml;ber eine Million
Leser zu erreichen? Dann ist Ihr Platz vielleicht bei PC Games. Wer
Interesse an einem Arbeitsplatz bei uns hat, findet alles Weitere in
unserem Stellenangebot in der Special-Sektion - folgen Sie einfach dem
Link unter dieser Meldung."

Can you make something out of this?

Regards,

Markus

-----UrsprО©╫ngliche Nachricht-----
Von: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Gesendet: Montag, 29. Juli 2002 18:01
An: Markus Wollny
Cc: pgsql-general@postgresql.org; Oleg Bartunov
Betreff: Re: [GENERAL] tsearch - Regression tests fail

"Markus Wollny" <Markus.Wollny@computec.de> writes:

Platform is SuSE Linux 7.3 (Linux 2.4.10-64GB-SMP #1 SMP Tue Sep 25
12:36:09 GMT 2001 i686 unknown); I think it doesn't dump core, so I
cannot provide a stack trace. In the logfile it just says: "DEBUG:
server process (pid 22137) was terminated by signal 11". I

cannot find

any core-files.

Signal 11 is SIGSEGV on most Unixen (certainly including
Linux), so the
backend is certainly crashing. If you are not getting core files in
the $PGDATA/base/yourdbnumber/ directory, then I'd bet that the
postmaster is being started under "ulimit -c 0" conditions. Restart
it with "ulimit -c unlimited" environment and you should get a core.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

http://www.postgresql.org/users-lounge/docs/faq.html

Regards,
Oleg
_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83

On Mon, 29 Jul 2002, Markus Wollny wrote:

This is what I get when I do "gdb psql ./core":

Core was generated by `postgres: postgres pcgames [local] SELECT

[..]

It wasn't psql that segfaulted, but the postgres backend.
Try:

$ gdb postgres ./core

instead. It's still not unlikely that you won't have
enough symbol information, though.

Matthew.

Hi!

Thanks for that hint - I tried that and got:

warning: core file may not match specified executable file.
Core was generated by `postgres: postgres pcgames [local] SELECT
'.
Program terminated with signal 11, Segmentation fault.
#0 0x40126853 in ?? ()
(gdb)
(gdb) bt
#0 0x40126853 in ?? ()
#1 0x50e280fb in ?? ()
#2 0x50e290a5 in ?? ()
#3 0x080c90bd in ExecHashJoin ()
#4 0x080c9194 in ExecHashJoin ()
#5 0x080c9720 in ExecHashJoinNewBatch ()
#6 0x080c9a3a in IndexNext ()
#7 0x080c9cd6 in ExecIndexReScan ()
#8 0x080cf934 in _SPI_pquery ()
#9 0x080c8079 in ExecInitAgg ()
#10 0x080c6fce in init_sql_fcache ()
#11 0x080c65f7 in ExecTypeFromTL ()
#12 0x0811796b in DecodeDateDelta ()
#13 0x08116421 in DecodeDateTime ()
#14 0x081173d0 in DecodeNumberField ()
#15 0x080fd0f0 in ldissect ()
#16 0x080fc91f in lmatcher ()
#17 0x080fbac9 in sfast ()
#18 0x080fb635 in sbackref ()
#19 0x080d95bd in _equalAppendPath ()
#20 0x400cb7ee in ?? ()

Can someone make something out of this?

Regards,

Markus

-----Ursprüngliche Nachricht-----
Von: Matthew Kirkwood
Gesendet: Mo 29.07.2002 19:17
An: Markus Wollny
Cc: pgsql-general@postgresql.org; tgl@sss.pgh.pa.us
Betreff: Re: [GENERAL] tsearch - Regression tests fail

On Mon, 29 Jul 2002, Markus Wollny wrote:

This is what I get when I do "gdb psql ./core":

Core was generated by `postgres: postgres pcgames [local]

SELECT
[..]

It wasn't psql that segfaulted, but the postgres backend.
Try:

$ gdb postgres ./core

instead. It's still not unlikely that you won't have
enough symbol information, though.

Matthew.

"Markus Wollny" <Markus.Wollny@computec.de> writes:

Can someone make something out of this?

It's garbage :-( --- that chain of control is quite impossible.

I fear you will need to build from source using --enable-debug
in the configure step to get more meaningful info.

regards, tom lane

Hi!

I thought as much... Well I didn't want to get kneedeep in the mud
experimenting with outdated code, so I just tried the very same with
7.2.1 (same setup as before with 7.2) and was successful - I thought I
might let you know.

The only difficulty I stumbled across was a missing flex. Being a bit
thick sometimes, I didn't realize that I had to reconfigure and rebuild
from scratch after the installation of flex, to compile successfully...
(cd /usr/src/postgresql-7.2.1, make distclean, /configure ..., make, cd
contrib/tsearch, make, make install). I found a hint to this effect in
the archives, though.

So thank you very much for your support!

Regards,

Markus