Thanks Özgür... This email application is not exposed to any real users with Inbox. Its an internal email processing application which will recieve emails routed from our Enterprise Email server. So each of our James nodes will receive an email and process it (extract some info from subject and body and call couple of web services)..Let me try to adjust the parameters as you specified. Just a clarification.. In a 4 CPU machine, how far we can go ahead with those parameters ? Can you suggest some figures based on your experience ? The reason is I can't do a trial and error thing in production..
Thanks Mahesh On Wed, Apr 1, 2015 at 2:36 PM, Özgür EROĞLU <oeroglu.c...@gmail.com> wrote: > Sorry, I could not get your point in using two servers which are getting > mail for the same domain, without having a common db for them. Of course > as I do not know you project details, it might have special use case here. > I just wondered. > > > And I meant; > > increase Spool Manager threads > decrease smtp connectiontimeout > decrease connection manager idle timeout > and increase connection manager max-connections. > > > Özgür Eroğlu > > > On 04/01/2015 11:43 AM, Mahesh Sivarama Pillai wrote: > >> Hi, >> >> Thanks for the inputs. We don't have any DB configurations. Its all file >> based. There are two servers which are already behind the load balancer. >> We >> were getting Too Many Open files issues.. We have increased the ulimit to >> the maximum two days back. Did you mean "decrease the timeout and increase >> the max-connections ?" In a 4CPU machine, can I increase the >> max-connections to say 100 if we came to know that its about load ? >> >> Thanks >> Mahesh >> >> On Wed, Apr 1, 2015 at 1:16 PM, Özgür EROĞLU <oeroglu.c...@gmail.com> >> wrote: >> >> Hi Mahesh, >>> >>> All you have asked may be the cause, I can not give clear answer right >>> now. But if your system have sufficient resources try to increase >>> max-connections and timeout parameters. And observe whether any >>> differences >>> occur or not. >>> >>> But I recommend you to use more than one James instances which are >>> connected to the same and replicated database server( howto do it depends >>> on your DB choice) and put them all behind a load balancer, so that you >>> can >>> reduce the load on one server (of course this will be solution only if >>> the >>> problem is the actual load). >>> >>> Özgür Eroğlu >>> >>> >>> On 04/01/2015 10:22 AM, Mahesh Sivarama Pillai wrote: >>> >>> Hi All, >>>> >>>> We have rolled out James 2.3.2 to production for our email processing >>>> application. I see that James is no longed accepting connetions after >>>> few >>>> days of run. It processes around 100K email a day and sends a good >>>> amount >>>> of Notification emails through RemoveDelivery. James is running on a 4 >>>> CPU >>>> machine with 8GB RAM. Heapsize of James is set to 4GB with Openjdk 1.7. >>>> >>>> INITIALLY WE THOUGHT the server was getting killed/terminated. As per >>>> the >>>> latest information, server is not killed.. Our support team used to >>>> restart >>>> the server when we get "Connection Refused" error from port 25..We have >>>> a >>>> monitoring tool which Connects to James server every few seconds, and >>>> issue >>>> a QUIT command... This monitoring tool is getting Connection Refused >>>> error. >>>> Hence the team thought the server is down and followed the routine Stop, >>>> Start commands... >>>> >>>> I have the following configurations in various places in the >>>> config.xml. I >>>> have INFO logging enabled. I couldnt find any WARN messages about >>>> connections. >>>> >>>> SMTP Server (the only process enabled): >>>> <connectiontimeout>360000</connectiontimeout> >>>> >>>> Spool Manager: <threads> 10 </threads> >>>> >>>> Connections Manager >>>> >>>> <connections> >>>> <idle-timeout>300000</idle-timeout> >>>> <max-connections>30</max-connections> >>>> </connections> >>>> >>>> Thread Manager >>>> >>>> <thread-manager> >>>> <thread-group> >>>> <name>default</name> >>>> <priority>5</priority> >>>> <is-daemon>false</is-daemon> >>>> <max-threads>100</max-threads> >>>> <min-threads>20</min-threads> >>>> <min-spare-threads>20</min-spare-threads> >>>> </thread-group> >>>> </thread-manager> >>>> >>>> The total number of threads (spool+remotedelivery etc ) are under very >>>> much >>>> under 100. We don't have any DB configuration in config.xml as well. Do >>>> you >>>> think the timeout values might cause the connection refused errors ?. >>>> Especially the idle-timeout ? Isn't 5 minutes too high ? If say 30 >>>> clients >>>> are taking few minutes, this will be more than enough to raise an alert >>>> from the monitoring tool...even if all the connections are being used, >>>> it >>>> should be release after the processing done right ? Is there any chance >>>> of >>>> the connections not being released and going back to the pool ? Please >>>> provide your comments.. >>>> >>>> Thanks >>>> Mahesh >>>> >>>> >>>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: server-user-unsubscr...@james.apache.org >>> For additional commands, e-mail: server-user-h...@james.apache.org >>> >>> >>> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: server-user-unsubscr...@james.apache.org > For additional commands, e-mail: server-user-h...@james.apache.org > >
