tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From bugzi...@apache.org
Subject DO NOT REPLY [Bug 44454] - busy count reported in mod_jk inflated, causes incorrect balancing
Date Tue, 19 Feb 2008 21:31:24 GMT
DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG·
RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT
<http://issues.apache.org/bugzilla/show_bug.cgi?id=44454>.
ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND·
INSERTED IN THE BUG DATABASE.

http://issues.apache.org/bugzilla/show_bug.cgi?id=44454





------- Additional Comments From rainer.jung@kippdata.de  2008-02-19 13:31 -------
Are we sure, that the growing busy numbers are wrong?

Asked differently: can you do Thread Dumps of your backends to confirm, that
they are *not* having that many requests busy? In theory some requests might got
stuck in the backend. If you don't have a reaply_timeout, mod_jk will wait
indefinitely for a response.

By using "kill -QUIT PID" (PID is the process id of your backend=romcat process)
you will get a stack for each thread in Tomcat in your catalina.out. This
procedure does not influence your running Tomcat negatively. It will pause for a
few milliseconds and then resume the usual work.

In the dump, the threads responsible for requests coming in via AJP13 (from
mod_jk) can be identified as TP-ProcessorXXX. Some of the will only have a very
short stack, which indicates, that they are idle in the thread pool. Some have a
slightly longer stack (about 10 lines) with a top level method being a socket
read, which tells us, that mod_jk is connected to them, but they are not actualy
processing a request, and few will have a very long stack, usually containing
webapp methods, a service method etc. Those ar the ones that process a request
and should be equivalnt to "busy" (at least of Tomcat only gets requests from
one httpd. Otherwise you need to add up all the busy for this Tomcat in the
various httpd).

Altough in principal there's a chance, that busy could go wrong, I never
detected it in practice and the feature is out in the wild for quite some time
now, without problem reports. That's why I first want you to check, taht the
numbers are really wrong.

Furthermore: are there errors in the JK error log?

Last: if you want to track how and when busy increases, you can add
"%{JK_LB_LAST_BUSY}n" to your LogFormat in httpd. Then your access log will
contain the busy number after each request in the request log line.

Regards,

Rainer


-- 
Configure bugmail: http://issues.apache.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org


Mime
View raw message