AW: Installation rt-2-0-15 with fastcgi

Warnke_Andreas · February 24, 2003, 5:55pm

Thanks John for your answer,
but I guess, this must have a different cause:
this happens even on the login-page - and on every other page. And five minutes later everything is fine again - without server-restart or anything else.

And on our testmachine with mod_perl, everything is fine too.

There is an other anomalie: from time to time (every two weeks), the mysql database needs too much processing time: 90% of the 2GHz CPU are gone for the mysql process. A restart of mysql doesn’t solve the problem, a RT2 restart is necessary. I don’t know, if these two problems are linked somehow - but also this problem doens’n occur on our test-machine.

Kind regards
AndreasVon: John Gedeon [mailto:jgedeon@qualcomm.com]
Gesendet: Montag, 24. Februar 2003 18:29
An: Warnke, Andreas
Cc: rt-users@fsck.com
Betreff: Re: [rt-users] Installation rt-2-0-15 with fastcgi

Hi Andreas,

The time I ran into that error with rt-2-0-15 on Solaris with fastcgi

is when my scripts were trying to access a file that didn’t exist. The
situation was that we have several Solaris machines and the file systems
are all shared. One of the computers was down there for when RT tried to go
to …/computer-name/path/to/file via the web it hung and I started getting
those time out errors. Maybe you have a file that RT is trying to access
that it cannot. I can’t say exactly though since you are also getting the
incomplete header errors which I did not get.

 Try and see what those requests have in common, ie if they are trying

to access the same page? the same ticket? etc. and maybe that will give you
a lead on the problem.

HTH
John

At 09:01 AM 2/24/2003, you wrote:

Hello,

I have installed rt-2-0-15 with fastcgi.
Now - everything works fine since weeks

except for some requests : ist about 1%:

It seems there is a communication problem
betwenn fastcgi and mason_handler.fcgi.
The error log says:

[Mon Feb 17 12:57:07 2003] [error] [client 143.164.98.120] FastCGI: comm
with server “/usr/local/rt2/bin/mason_handler.fcgi” aborted: idle
timeout (300 sec)
[Mon Feb 17 12:57:07 2003] [error] [client 143.164.98.120] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”
[Mon Feb 17 12:59:00 2003] [error] [client 212.86.197.105] FastCGI: comm
with server /usr/local/rt2/bin/mason_handler.fcgi" aborted: idle
timeout (300 sec)
[Mon Feb 17 12:59:00 2003] [error] [client 212.86.197.105] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”
[Mon Feb 17 12:59:16 2003] [error] [client 143.164.98.120] FastCGI: comm
with server “/usr/local/rt2/bin/mason_handler.fcgi” aborted: idle
timeout (300 sec)
[Mon Feb 17 12:59:16 2003] [error] [client 143.164.98.120] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”
[Mon Feb 17 13:00:00 2003] [error] [client 212.114.211.42] FastCGI: comm
with server “/usr/local/rt2/bin/mason_handler.fcgi” aborted: idle
timeout (300 sec)
[Mon Feb 17 13:00:00 2003] [error] [client 212.114.211.42] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”

The user gets an ‘Internal Server Error’ message.
This is independently of the requested page.
This happens independently of the socket-type:
unix-pipes as well as tcp connections
between fastcgi and perl don’t work.

Has anyone a solution? I’d be eager for suggestions.

Kind Regards
Andreas

–
Andreas Warnke
3SOFT GmbH, Frauenweiherst. 14, 91058 Erlangen
Tel.: +49-9131-7701-274 mailto:Andreas.Warnke@3SOFT.de
Fax: +49-9131-7701-333 http://www.3SOFT.de

rt-users mailing list
rt-users@lists.fsck.com
http://lists.fsck.com/mailman/listinfo/rt-users

Have you read the FAQ? The RT FAQ Manager lives at http://fsck.com/rtfm

<>< Proverbs 3:5 “Trust in the Lord with all your heart and lean not on
your own understanding;”

John_Gedeon · February 24, 2003, 7:44pm

I’m not using msql any more but if it is hanging or taking too long to
return data for RT that may cause your errors, What do you other people on
the list think? Maybe checking into the differences between the mysql
installations on the two machines or maybe RT and mysql hang when doing acl
checks?

But thats just an educated guess.
John
At 09:55 AM 2/24/2003, you wrote:

Thanks John for your answer,
but I guess, this must have a different cause:
this happens even on the login-page - and on every other page. And five
minutes later everything is fine again - without server-restart or
anything else.

And on our testmachine with mod_perl, everything is fine too.

There is an other anomalie: from time to time (every two weeks), the mysql
database needs too much processing time: 90% of the 2GHz CPU are gone for
the mysql process. A restart of mysql doesn’t solve the problem, a RT2
restart is necessary. I don’t know, if these two problems are linked
somehow - but also this problem doens’n occur on our test-machine.

Kind regards
Andreas

-----Ursprüngliche Nachricht-----
Von: John Gedeon [mailto:jgedeon@qualcomm.com]
Gesendet: Montag, 24. Februar 2003 18:29
An: Warnke, Andreas
Cc: rt-users@fsck.com
Betreff: Re: [rt-users] Installation rt-2-0-15 with fastcgi

Hi Andreas,
The time I ran into that error with rt-2-0-15 on Solaris with fastcgi
is when my scripts were trying to access a file that didn’t exist. The
situation was that we have several Solaris machines and the file systems
are all shared. One of the computers was down there for when RT tried to go
to …/computer-name/path/to/file via the web it hung and I started getting
those time out errors. Maybe you have a file that RT is trying to access
that it cannot. I can’t say exactly though since you are also getting the
incomplete header errors which I did not get.
 Try and see what those requests have in common, ie if they are trying
to access the same page? the same ticket? etc. and maybe that will give you
a lead on the problem.

HTH
John

At 09:01 AM 2/24/2003, you wrote:

Hello,

I have installed rt-2-0-15 with fastcgi.
Now - everything works fine since weeks

except for some requests : ist about 1%:

It seems there is a communication problem
betwenn fastcgi and mason_handler.fcgi.
The error log says:

[Mon Feb 17 12:57:07 2003] [error] [client 143.164.98.120] FastCGI: comm
with server “/usr/local/rt2/bin/mason_handler.fcgi” aborted: idle
timeout (300 sec)
[Mon Feb 17 12:57:07 2003] [error] [client 143.164.98.120] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”
[Mon Feb 17 12:59:00 2003] [error] [client 212.86.197.105] FastCGI: comm
with server /usr/local/rt2/bin/mason_handler.fcgi" aborted: idle
timeout (300 sec)
[Mon Feb 17 12:59:00 2003] [error] [client 212.86.197.105] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”
[Mon Feb 17 12:59:16 2003] [error] [client 143.164.98.120] FastCGI: comm
with server “/usr/local/rt2/bin/mason_handler.fcgi” aborted: idle
timeout (300 sec)
[Mon Feb 17 12:59:16 2003] [error] [client 143.164.98.120] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”
[Mon Feb 17 13:00:00 2003] [error] [client 212.114.211.42] FastCGI: comm
with server “/usr/local/rt2/bin/mason_handler.fcgi” aborted: idle
timeout (300 sec)
[Mon Feb 17 13:00:00 2003] [error] [client 212.114.211.42] FastCGI:
incomplete headers (0 bytes) received from server
“/usr/local/rt2/bin/mason_handler.fcgi”

The user gets an ‘Internal Server Error’ message.
This is independently of the requested page.
This happens independently of the socket-type:
unix-pipes as well as tcp connections
between fastcgi and perl don’t work.

Has anyone a solution? I’d be eager for suggestions.

Kind Regards
Andreas

–
Andreas Warnke
3SOFT GmbH, Frauenweiherst. 14, 91058 Erlangen
Tel.: +49-9131-7701-274 mailto:Andreas.Warnke@3SOFT.de
Fax: +49-9131-7701-333 http://www.3SOFT.de

rt-users mailing list
rt-users@lists.fsck.com
http://lists.fsck.com/mailman/listinfo/rt-users

Have you read the FAQ? The RT FAQ Manager lives at http://fsck.com/rtfm

–
<>< Proverbs 3:5 “Trust in the Lord with all your heart and lean not on
your own understanding;”

rt-users mailing list
rt-users@lists.fsck.com
http://lists.fsck.com/mailman/listinfo/rt-users

Have you read the FAQ? The RT FAQ Manager lives at http://fsck.com/rtfm

<>< Proverbs 3:5 “Trust in the Lord with all your heart and lean not on
your own understanding;”

Bruce_Campbell · February 25, 2003, 4:10pm

There is an other anomalie: from time to time (every two weeks), the
mysql database needs too much processing time: 90% of the 2GHz CPU are
gone for the mysql process. A restart of mysql doesn’t solve the
problem, a RT2 restart is necessary.

What do you mean by a ‘RT2 restart’ ? Is this a restart of the apache
handling RT2, a restart of the actual server, or a restart of a particular
CLI app?

Andreas

                         Bruce Campbell                            RIPE
               Systems/Network Engineer                             NCC
             www.ripe.net - PGP562C8B1B             Operations/Security

Warnke_Andreas · February 28, 2003, 10:25am

Hello,

We had the strangest errors - and we had absolutely no clue to where our problem is - but since 2 days, the system is running and the error log stays clean - may be we have found a valuable hint:

When the system was hanging (no crashes, just 100% CPU time and no results), a restart of mysql did nothing,
a restart of the apache was a good help.
FastCGI was complaining about communication problems with mason_handler.fcgi
MySQL didn’t accept any requests -even from the command line tool mysql.

We exchanged fast_cgi by mod_perl, downloaded the lastest cpan archives or took the perl-modules that come with SuSE 8.1 - nothing helped.
Now, we downgraded mysql from 3.23.51 (or something like that, comes with SuSE 8.1) to 3.23.48-19 - And our problems were gone. At least for 2 days.

Kind Regards
Andreas
Andreas Warnke
3SOFT GmbH, Frauenweiherst. 14, 91058 Erlangen
Tel.: +49-9131-7701-274 mailto:Andreas.Warnke@3SOFT.de
Fax: +49-9131-7701-333 http://www.3SOFT.deVon: Bruce Campbell [mailto:bruce_campbell@ripe.net]
Gesendet: Dienstag, 25. Februar 2003 17:10
An: rt-users@fsck.com
Betreff: Re: AW: [rt-users] Installation rt-2-0-15 with fastcgi

There is an other anomalie: from time to time (every two weeks), the
mysql database needs too much processing time: 90% of the 2GHz CPU are
gone for the mysql process. A restart of mysql doesn’t solve the
problem, a RT2 restart is necessary.

What do you mean by a ‘RT2 restart’ ? Is this a restart of the apache
handling RT2, a restart of the actual server, or a restart of a particular
CLI app?

Andreas

                         Bruce Campbell                            RIPE
               Systems/Network Engineer                             NCC
             www.ripe.net - PGP562C8B1B             Operations/Security

rt-users mailing list
rt-users@lists.fsck.com
http://lists.fsck.com/mailman/listinfo/rt-users

Have you read the FAQ? The RT FAQ Manager lives at http://fsck.com/rtfm