tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christopher Schultz <>
Subject Re: AW: AW: Tomcat Cluster 6.x and HTTP/1.1 503 Service Temporarily Unavailable
Date Mon, 01 Feb 2010 18:22:57 GMT
Hash: SHA1


I took at look at your log file, and everything looks good to me,
honestly. There are no errors that I can see, other than the
setsockopt(SO_UPDATE_ACCEPT_CONTEXT) warning that apparently accompanies
every child process shutdown.\

On 2/1/2010 6:17 AM, Alexander Diedler wrote:
> In httpd.conf:
> ThreadLimit 1920
> ThreadsPerChild 1920
> MaxRequestsPerChild 10000

Okay, but this doesn't really say anything about your AJP connections,
other than you might expect to see 1920 concurrent requests that might
be sent to Tomcat.

> worker.loadbalancer.type=lb
> worker.loadbalancer.balance_workers=worker1 worker2

Great. Where are the settings for worker1 and worker2? These can be
quite important. Essential, some might say.

> server.xml: <Connector 	port="8010" 
> 			protocol="AJP/1.3" 
> 			connectionTimeout="30000"
> 			redirectPort="8444" 
> 			request.useSecret="true" 
> 			request.secret="XXXXX"
> 			maxThreads="10000"
> 			tcpNoDelay="true" />

Okay, so you have 10000 threads per server, and 2 app servers. That's
max 20000 threads configured for a maximum load of 3040 possible
concurrent incoming connections. I would review all those settings at
some point. It doesn't looks like mod_jk is allocating too many connections:

> 44 requests currently being processed, 1656 idle workers

...and Tomcat won't allocate a bunch of threads until it needs them, so
it never will. So, this isn't really a problem... just an indication of
sloppy resource management decisions.

- -chris
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla -


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message