Mailing-List: contact users-help@tomcat.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Tomcat Users List" <users@tomcat.apache.org>
Received-SPF: neutral (athena.apache.org: local policy)
Message-ID: <4D25FA96.3050602@christopherschultz.net>
Date: Thu, 06 Jan 2011 12:23:34 -0500
From: Christopher Schultz <chris@christopherschultz.net>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US;
 rv:1.9.2.13) Gecko/20101207 Thunderbird/3.1.7
MIME-Version: 1.0
To: Tomcat Users List <users@tomcat.apache.org>
Subject: Re: Tomcat 6 busy threads issue
References: <1294331219.2694.37.camel@kbiton-lap-ub>
	 <4D25F66E.4070304@christopherschultz.net>
 <1294334067.2694.45.camel@kbiton-lap-ub>
In-Reply-To: <1294334067.2694.45.camel@kbiton-lap-ub>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Kobi,

On 1/6/2011 12:14 PM, Kobi Biton wrote:
> -Tomcat 6.0.22

Great. Any opportunities to upgrade? Your version was never released and
is over a year old.

> - yes they are load balanced simple Round Robin Linux LVS  , (LB
> distributes the requests evenly)

So, you expect each server to serve 1/N of the incoming requests per
unit time. Any session stickiness?

> - All the servers (But 1) had the following thread pool confguration
> prior the change:
>    
>    MaxThreads=1000 , minSpareThread=25,maxSpateThreads=75
> 
> 1 server had:
> 
>   MaxThreads=300 , minSpareThread=25,maxSpateThreads=75

Okay, that will mean that the second server will handle, at most, 300
simultaneous connections. Do you have any per-backend-server connection
limits set at the load balancer?

> For some reason the server that had MaxThreads=300 report x2
> threadCount / threadBusy Via JMX polling , only after changing its
> maxThreads to 1000 the threadCount/threadBusy count was decreased x2 and
> matched all the other servers

So, let's do an example so I can get my head around it:

N = 3

Server 1: maxThreads=1000, busy=100
Server 2: maxThreads=1000, busy=100
Server 3: maxThreads=300,  busy=200

Is that roughly what you observed? Was that consistent over a long
period of time? Did you observe what kinds of requests were being served
on server 3? If some of them were long-running, you may just be
observing a natural build-up of work on a server that reliably gets 33%
of incoming requests but has some older requests that are still processing.

> - Indeed busyThreads <= maxThreads

Good. If that weren't true, something would be terribly wrong :)

> Just trying to figure out why did the maxThreads was Impacting the
> thread busy/count ...

I can't imagine why it would, other than putting a cap on the number of
busy threads (because busyThreads <= maxThreads).

What kind of lb are you using?

- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk0l+pYACgkQ9CaO5/Lv0PBuogCgitvm9D01NSi84oO0tVO+itdn
ByMAmwfOVpq59BhVIlUtx/EswxsBm3lJ
=AzZU
-----END PGP SIGNATURE-----

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