[Pgpool-general] hanging pgpool processes during massive UPDATEs

[Yoshiyuki Asaba]

Hi,

From: "Felix J. Ogris" <fjo-lists at ogris.de>
Subject: Re: [Pgpool-general] hanging pgpool processes during massive UPDATEs
Date: Thu, 11 Oct 2007 09:38:11 +0200

> > Could you show us pgpool.conf and backtrace?
> > 
> >   % gdb -p <PID> --batch --ex bt
> Backtrace as with `gdb -p <PID>` and manually "bt" cmd (gdb -p <PID> --batch
> --ex bt doesn't work with my 'GNU gdb Red Hat Linux (6.3.0.0-1.143.el4rh)'):
> 
> (gdb) bt
> #0  0x0000000000443e87 in base_yylex () at scan.c:1080
> #1  0x000000000042d638 in sql_yylex () at parser.c:87
> #2  0x00000000004393aa in yyparse () at gram.c:11084

Thanks for the report. I can reproduce with the following query.

  UPDATE up SET "a=1

If there is an unterminated double-quote identifier, pgpool enters
infinite-loop. In this case, there is the identifier in WHERE clause.
Could you modify your query? I will fix tomorrow, sorry.


> > Could you show us a result of the following command?
> > 
> >   % grep 559 log | grep -A1 statement2
> 
> Ok, once again I have scrambled all statements for the sake of our
> customers:

Thanks. I will research it.

--
Yoshiyuki Asaba
y-asaba at sraoss.co.jp