[rt-users] Fulltext search error 500 on first try, then ok

[Michele Bergonzoni]

[rt 4.0.9, postgres 9.1 with full text search, ubuntu 12.04.1, Apache 
2.2.22, mod_fastcgi-SNAP-0910052141, Perl v5.14.2]

I have a sporadic annoyance.

When I try a fulltext search on all tickets ("any" keyword), it waits 10 
seconds and then I get a 500 Internal server error. Apache error log says:

[warn] FastCGI: (dynamic) server "/path/to/rt-4.0.9/sbin/rt-server.fcgi" 
(pid 10410) terminated due to uncaught signal '13' (Broken pipe)

And postgresql log:

  LOG:  could not receive data from client: Connection reset by peer
  LOG:  unexpected EOF on client connection
  LOG:  could not receive data from client: Connection reset by peer
  LOG:  unexpected EOF on client connection

(four lines in the same second as the Apache line)

If i retry immediately after, all goes well and the results is displayed 
in maybe 2-3 seconds. Even if I purposely try a search that I know will 
return thousands of tickets, all goes well in a few seconds and tickets 
are retrieved.

The rt-fulltext-indexer is run every ten minutes and completes in a few 
seconds. Postgres statement timeout is 0 (disabled), mod_fastcgi 
-idle-timeout is set to 300.

Besides telling users to keep calm and carry on hitting F5, does anybody 
have any troubleshooting path to suggest? Please keep in mind that 
reproducing this is not easy, today I saw it only once.

Thanks in advance,
				Bergonz

-- 
Ing. Michele Bergonzoni - Laboratori Guglielmo Marconi S.p.a.
Phone:+39-051-6781926 e-mail: bergonz at labs.it
alt.advanced.networks.design.configure.operate