Thanks for the advice. I will try your suggestions. I
noticed that from the server status view of the tomcat
manager, I saw some thread with the status 'S' has the
time of '24685295'. Does it mean the thread has hang
there for more than 6 hours (close to 7 hours)?



--- Parsons Technical Services
<[EMAIL PROTECTED]> wrote:

> I think that would be the direction I would take.
> 
> If you need more proof:
> 
> Sniff the connection to the server to confirm the
> lack of available 
> connections. So far it is pointing in that
> direction.
> 
> Another test you might do is create a connection
> from a known IP, close the 
> browser and monitor the connection status. If it is
> hanging open then start 
> digging.
> 
> I think you may also be experiencing  an issue of
> clickatosis. It occurs 
> when the client doesn't think the page is responding
> quick enough and since 
> more is better after 10 or 20 clicks they decide to
> wait or give up. Mean 
> while for each click they cause a request which can
> bog things down. 
> Wrapping this with the held connections you can get
> a scenario as such:
> 
> Client hits site.
> No or slow response.
> Clicks again. Or several times.
> Often closes browser and tries over.
> This creates a new connection request.
> Snowballs from there.
> 
> Think about this, do you really get over a 100k
> users in six hours? Or even 
> 100k page requests? That's a lot of traffic. If the
> client machines abandons 
> the connection before the server, then establishes a
> new one, it compound 
> the problem.
> 
> Find a way to have the OS let Apache handle the
> connection timeout and get 
> it down to a reasonable time.
> 
> Since you are inheriting this, there may be a reason
> that the connection 
> timeout is set so high. If the original creator of
> this wanted the user to 
> be able to sit on a page for a long time without
> losing the session and 
> connection thus causing the user to have to
> re-authenticate. But then there 
> is a trade off. which you are experiencing right
> now.
> 
> Get the time down until the complaint about having
> to re-login get too high 
> or until the connection speed delay is acceptable.
> 
> Just my $0.015
> 
> Doug
> ----- Original Message ----- 
> From: "Jeffrey" <[EMAIL PROTECTED]>
> To: "Tomcat Users List"
> <tomcat-user@jakarta.apache.org>
> Sent: Saturday, May 07, 2005 1:05 PM
> Subject: Re: Tomcat 5 slow, it's in production,
> please help!
> 
> 
> > It seems that, at least in most cases, the main
> menu
> > page after a login is loaded slow. After that
> other
> > pages are loaded pretty quick. Does that mean once
> a
> > connection is established, browser and server
> > communication is ok. If that is the case, maybe I
> > should look more at the connection timeout at the
> OS
> > level (solaris 5.6) since I see a lot of TIME_WAIT
> at
> > the machine using the 'netstat -n' command.
> >
> >
> > --- Jeffrey <[EMAIL PROTECTED]> wrote:
> >
> >> Please see the answer embedded below.
> >>
> >> --- Parsons Technical Services
> >> <[EMAIL PROTECTED]> wrote:
> >>
> >> > 127421 over what period of time?
> >> It's about 6 hours. But most of them are at the
> >> first
> >> hour or so.
> >> >
> >> > Without a profile it is impossible to say if
> >> Tomcat
> >> > can handle the load. But
> >> > if the memory consumption and the CPU usage is
> low
> >> > (as noted) it is fairly
> >> > safe to say that Tomcat itself is not the
> bottle
> >> > neck.
> >> >
> >> > Quick question, why are you using Apache?
> >>
> >> I am using Apache SSL for SSL connections. It is
> the
> >> original setup by someone else.
> >>
> >> >
> >> > Also what is the time out for the connection in
> >> > Apache?
> >>
> >> Apache's Timeout value is 300.
> MaxKeepAliveRequests
> >> is
> >> 100 and KeepAliveTimeout is 15
> >> >
> >> > Do you get this slow reaction if you are
> hitting
> >> > Tomcat directly?
> >>
> >> I did get the slow reaction if I hit Tomcat
> directly
> >> at port 8080 to the same webapp. But the tomcat
> >> manager web pages seem show up fairly quick.
> >>
> >> >
> >> > I may be repeating something that was covered
> >> > because I have not been
> >> > reading this thread in detail. So sorry if
> already
> >> > asked and answered.
> >> >
> >> >
> >> > Doug
> >> >
> >> >
> >> >
> >> > ----- Original Message ----- 
> >> > From: "Jeffrey" <[EMAIL PROTECTED]>
> >> > To: "Tomcat Users List"
> >> > <tomcat-user@jakarta.apache.org>
> >> > Sent: Saturday, May 07, 2005 10:44 AM
> >> > Subject: Re: Tomcat 5 slow, it's in production,
> >> > please help!
> >> >
> >> >
> >> > > This website has many JSP pages of which each
> >> has
> >> > many
> >> > > JSPF included. I wonder whether or not the
> >> system
> >> > is
> >> > > not be able to handle this kind of  setting.
> But
> >> > > strange thing is that when we experienced the
> >> > > slowness, I looked at the usage of CPU and
> find
> >> > that
> >> > > almost 90% of the CPU is idle.
> >> > >
> >> > > Do anyone know what kind of load a tomcat can
> >> > handle.
> >> > > I am thinking that maybe a load balancer will
> >> > help.
> >> > > But I am still puzzled by the fact of low CPU
> >> > usage at
> >> > > the time of slowness.
> >> > >
> >> > > From the server status view in tomcat
> manager, I
> >> > can
> >> > > see 's Request count: 127421'
> >> > >
> >> > > --- Jeffrey <[EMAIL PROTECTED]> wrote:
> >> > >> I have a apache/tomcat environment that has
> >> some
> >> > >> problems recently. The environment has the
> >> > >> following:
> >> > >> 1.  Apache 1.3
> >> > >> 2.  Tomcat 5.0.28 (Max memory 384mb)
> >> > >> 3.  JDK 1.4.2_06
> >> > >>
> >> > >> My apache has MaxClient set to 256 (apache
> >> > default).
> >> > >> The maxThread for tomcat is 500. The site is
> a
> >> > >> database driven site.
> >> > >>
> >> > >> It seems working fine when load is low. But
> >> when
> >> > >> there
> >> > >> are more than 150 threads (from the sever
> >> status
> 
=== message truncated ===



                
__________________________________ 
Do you Yahoo!? 
Make Yahoo! your home page 
http://www.yahoo.com/r/hs

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to