Re: Re : Memory leak in Tomcat 6.0.35 ( 64 bit)
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Saumil, Please try to keep discussions on the mailing list to everyone can benefit. On 4/2/13 6:48 PM, saumil shah wrote: For some reason ...I do not see Java process in Task Manager in Windows, just Tomcat6 process. I am assuming killing Tomcat6 kills JVM internally...is this correct assumption ? I'm not sure what the Tomca6 process is, but if you kill the service (and nothing goes wrong), the JVM should terminate. Also, I am NOT re-deploying the application WAR's for some reason...when I deployed the application using Tomcat Manager ... for some reason it put WAR in the WebApps folder and then exploded them. That is entirely expected behavior. If you don't want your webapps to be deployed on startup, you'll need to set deployOnStartup=false in your Host. If you set deplyOnStartup=false, you'll have to use the Manager webapp to deploy your webapps after Tomcat has started. Now it seems every time I start Tomcat , it tries to Re-deploy the application because it finds .WAR there ? I wouldn't use the term re-deploy since the webapp isn't running prior to that. I usually use the term re-deploy to mean that the webapp is taken out of service in a running Tomcat, then a new webapp is deployed in its place (usually on the same context path). - -chris -BEGIN PGP SIGNATURE- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBCAAGBQJRXD5GAAoJEBzwKT+lPKRYq0gQAJZ9E4OSrYj+ypDTmHUV5ADa GdUbo/TYfyZNPFjmHzl/GYeZlYTirs24dA95VacIiAdJhtZ2g0J44Oc6PnT9pqmR RxcGtqNF+kN+AV5i0Lf4ewAZE8MQBvWmBHumqt75ETEsYYAlEv0NaVza9EcM7DNi RTpR9MrD4QUS7jhrKsyvQiagL7hF8xJgDij9CY5Bc5j6wQjuh6nttW4JtJXOTxb1 iiMbQqndmX0RkZHY3Dw3BJOuFU/NjAmSGB5pfmDBrA+z6jasH4SZd0KOy3DKcSgX EV8ja57U161yJMdH7Bt7ap9C2mpAaoFKMvANaPYCy29oRcUgQ3qMB7lofRCy2NZ5 JDWDnVaKa4UdXoCK4pUvt/noo4EZwUhHI9y8IAtCOC+5xgEDA65FPeTnXBBWYqyu uASODSRe4DdQYfLJjMw6rWmGFCqM+aPICeZKcXlC+UR8eqp3pmoLL1I5Y15xth2y mW9dA/qVAUNLUXlW1oWw7b58UAyCi7nVSASC4p3LcQ1K+cnQ8j8/VB0GtMMgA2gL Bpk7RR19NlzeUXmK//AN/BthZY6oK09qz6n+yNTi2tiV4T2XCjlRO4dsLOoW73nP FPTMSFfzYaeBKkn0pHsORZLCtWz8bZ060dYrfKkKLSKhk11CzwGNeP4A5So9xa+J ncBMVYTGyKnanbDE3Fj9 =dZYT -END PGP SIGNATURE- - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: how to monitor the traffic through the connector
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Jeffrey, On 4/1/13 3:52 PM, Jeffrey Janner wrote: -Original Message- From: Daniel Mikusa [mailto:dmik...@vmware.com] Sent: Monday, April 01, 2013 1:09 PM To: Tomcat Users List Subject: Re: how to monitor the traffic through the connector On Apr 1, 2013, at 1:40 PM, a wrote: Hi, Thank you for reply. The servlet of the web application does not response to the request after running serveral months. The problem can be solved by restarting the tomcat. I want to troubleshoot by monitoring the number of avaliable threads or occupied threads. Thank you First, don't top post (i.e. reply at the top of your email). Either reply below the previous response or as typically works best, inline. Second, when you encounter a problem with your servlet / Tomcat, take three or four thread dump, separated by 15 - 20 seconds each. Each thread dump will show you the current status of all the threads in the JVM. You can use the thread dumps to check for block threads and long running processes. Instructions for taking a thread dump. https://wiki.apache.org/tomcat/HowTo#How_do_I_obtain_a_thread_dump_of_m y_running_webapp_.3F Dan FWIW, I've found jconsole immensely helpful in sorting out blocked threads. And none of that manually searching through dumps. Jconsole is notoriously difficult to get running on a remote server (unless it's running Microsoft Windows, but then again I can't imagine a Windows server that stays running for several months where the webapp turns out to be the problem ;). - -chris -BEGIN PGP SIGNATURE- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBCAAGBQJRXEfkAAoJEBzwKT+lPKRYKPUQAJIshzZVPoxPaqbBGLb6IzXe ChSiwK40Jo/LtniAxMkpgtraXXCatDCLK9TZKrc7QsnaYePcJTTuJylYF/cbDox9 6SvFMOWlvbm4ZX+Qj7jnKj5UkNIbuws7PHEFtD5s+2HyuXe5DSl/xTwlbW25cAi6 GwXI/HbhwwFAqLWkgf9AMyfA8SQT0YEYmC8hhBtz6lfJYP0xiN6oqeIuOgnJAFJf YU+ITSyiKyqHboNjnoz4OtTgQGbxVuGjeAeJKVjTJVL2qpHmVWHekV53/4Qps/nk vW55K3ah7K+e0xETZpHE1Wdbse189DDM2GEWvoEGANHjLnvJ19ugvs08IQ4xIR19 orZmSBf7RSFqR6MSwT2N6YrFTKnGij/yQXYg6oxJkF0/0Ep6frWxf8xfUis5+Muo /WcsW5nNFbrFHDWuAifPNT1HPXVtftJ/VxZs43Fsht6618pfFZMpQStqmUUWmbVb 31uTaZk7karGhdQuMFqQNMLhUcj6jhsZEQJPiozO/oc2GUznQfz8TcaRVhtiElxf PEUICU23hBVOHqOhfozmC5a1opFVbpSTvH2KHcdAkaJB1NrBtsKcA0GPTH8S4E5K y6Ld3TdrLXvpuu6d1tyfLtkaMixT5GnqbALiOPLvZFOKoSw99VQCUpj+NBeMvn5M RRrBwOu10/oV64H9SNvP =mPr8 -END PGP SIGNATURE- - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
RE: Re : Memory leak in Tomcat 6.0.35 ( 64 bit)
Thanks again Chris I did change in Tomcat 6.0/Conf/server.xml unpackWARs=false and autoDeploy=false but the logs still complaint about the Deploying web applications org.apache.catalina.startup.HostConfig deployWARINFO: Deploying web application archive CmcAppActions.warorg.apache.catalina.core.StandardContextaddApplicationListenerINFO: The listener com.sun.faces.config.ConfigureListener is alreadyconfigured for this context. The duplicate definition has been ignored. 2. Also for a 64 bit JVM , could I then just bump up my Tomcat Java params as : -Xms1024m -Xmx4096m instead of my default as -Xms512m -Xmx1024m -XX:MaxPermSize=1024m instead of my default -XX:MaxPermSize=512m Appreciate all your help. Thanks again. Date: Wed, 3 Apr 2013 10:35:50 -0400 From: ch...@christopherschultz.net To: users@tomcat.apache.org CC: saumil...@hotmail.com Subject: Re: Re : Memory leak in Tomcat 6.0.35 ( 64 bit) -BEGIN PGP SIGNED MESSAGE- Hash: SHA256 Saumil, Please try to keep discussions on the mailing list to everyone can benefit. On 4/2/13 6:48 PM, saumil shah wrote: For some reason ...I do not see Java process in Task Manager in Windows, just Tomcat6 process. I am assuming killing Tomcat6 kills JVM internally...is this correct assumption ? I'm not sure what the Tomca6 process is, but if you kill the service (and nothing goes wrong), the JVM should terminate. Also, I am NOT re-deploying the application WAR's for some reason...when I deployed the application using Tomcat Manager ... for some reason it put WAR in the WebApps folder and then exploded them. That is entirely expected behavior. If you don't want your webapps to be deployed on startup, you'll need to set deployOnStartup=false in your Host. If you set deplyOnStartup=false, you'll have to use the Manager webapp to deploy your webapps after Tomcat has started. Now it seems every time I start Tomcat , it tries to Re-deploy the application because it finds .WAR there ? I wouldn't use the term re-deploy since the webapp isn't running prior to that. I usually use the term re-deploy to mean that the webapp is taken out of service in a running Tomcat, then a new webapp is deployed in its place (usually on the same context path). - -chris -BEGIN PGP SIGNATURE- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBCAAGBQJRXD5GAAoJEBzwKT+lPKRYq0gQAJZ9E4OSrYj+ypDTmHUV5ADa GdUbo/TYfyZNPFjmHzl/GYeZlYTirs24dA95VacIiAdJhtZ2g0J44Oc6PnT9pqmR RxcGtqNF+kN+AV5i0Lf4ewAZE8MQBvWmBHumqt75ETEsYYAlEv0NaVza9EcM7DNi RTpR9MrD4QUS7jhrKsyvQiagL7hF8xJgDij9CY5Bc5j6wQjuh6nttW4JtJXOTxb1 iiMbQqndmX0RkZHY3Dw3BJOuFU/NjAmSGB5pfmDBrA+z6jasH4SZd0KOy3DKcSgX EV8ja57U161yJMdH7Bt7ap9C2mpAaoFKMvANaPYCy29oRcUgQ3qMB7lofRCy2NZ5 JDWDnVaKa4UdXoCK4pUvt/noo4EZwUhHI9y8IAtCOC+5xgEDA65FPeTnXBBWYqyu uASODSRe4DdQYfLJjMw6rWmGFCqM+aPICeZKcXlC+UR8eqp3pmoLL1I5Y15xth2y mW9dA/qVAUNLUXlW1oWw7b58UAyCi7nVSASC4p3LcQ1K+cnQ8j8/VB0GtMMgA2gL Bpk7RR19NlzeUXmK//AN/BthZY6oK09qz6n+yNTi2tiV4T2XCjlRO4dsLOoW73nP FPTMSFfzYaeBKkn0pHsORZLCtWz8bZ060dYrfKkKLSKhk11CzwGNeP4A5So9xa+J ncBMVYTGyKnanbDE3Fj9 =dZYT -END PGP SIGNATURE- - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: Re : Memory leak in Tomcat 6.0.35 ( 64 bit)
Chris, On Tue, Apr 2, 2013 at 5:12 PM, Christopher Schultz ch...@christopherschultz.net wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA256 I understand its not a premanent solution but as a stopgap for now ? If so , we can put it as part of daily cycle to bounce tomcat6. If you don't re-deploy your webapp, then daily rolling Tomcat restarts are not necessary. I wonder why you are re-deploying your web application so many times? As a new tomcat user and still somewhat junior java/jsf developer, I restart tomcat whenever I have new software changes to deploy-and-want-to-run-on the production server. sometimes, I deploy-and-restart multiple times per day, but sometimes, I'm able to let tomcat/tomee run for days without restart. We run several Tomcats in parallel with modest heaps (less than 512MiB each) and they can run for months before we stop them for upgrades. It *is* possible to run JVMs without running out of memory... I too, have not experienced any OOME, and recently, per what I have seen-and-read of other (more senior java developers than myself), I have decreased memory settings in my java options on tomcat7w.exe (see below). -Xmx1024m -XX:MaxPermSize=384m -XX:+UseTLAB -XX:+UseConcMarkSweepGC -XX:+CMSClassUnloadingEnabled On Java tab of tomcat7w.exe, have the following specified: Initial memory pool = 256 Maximum memory pool = Thread stack size = My Windows 2008 R2 Server (64bit 32GB RAM) never seems to get higher than 1% CPU, and I think I do have memory leaks somewhere in the app, but FWIW (in heap dump in java visual vm), the memory leaks seem to be tomee leaks. In Java Visual VM, I do see the memory grow over time, as the app is used (without a tomcat restart or re-deploy of app and then restart tomcat), but I still have not seen OOME...'yet'. I should leave the server up and running without restarting tomcat, but when I need to deploy new software changes, i can't wait to deploy. :) My apologies, not trying to hijack the thread. just sharing my response to your comments. Howard - -chris -BEGIN PGP SIGNATURE- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBCAAGBQJRW0m2AAoJEBzwKT+lPKRYt38QAJkMt1+S+h48OdnVU8GVNY+x nhBJMX2THAN9d6LIo7k/8bEUPlbBIbjZVHZeR7tPbt3HgcPaPXLZ+/eiPU7x3tEe ubn/Vp7WVNhrZ3+ZcHQz6AEBz5JfFoHNpwfPvopTxsClhdlbNkQVLoNQX2DHvao6 ARFG3pZWX8HlC/MYMoIsJLD6DxXj9zT0E1ZJzBImHt8r2zE9YPWo35k6RhnEjLut Ol74NejK8kD8orgGfJfz5bM6XXiWaxLm3tBXkukefcEC9Sq5/PMDP61npTBygFgK GhPCoKHfvxtm/oXIFHJrwfibpyoKa5gxdRPRey9PqqKc9zABw8t7oMHb/QxwC6qi qNe9xy5/iSK3NKqQLawXMsFmGfmTA2Rx13I0uP7TnP/2X0bjdjNa/uHx+VKPYY/U RfdqosN02x8BwXkXbDRzeQURiTAPwSKdH8PBL9itnYSLGi6byYynfsvsPeycbpuK zV0qdyvBOeHbz/j5dai7Z451PMxm5ccEZ1B8cPtQiRVXxx5+5lF3q9RFwtkenLbI IpfIrpj4D+96KP66Rn0yfVNEeHosljAOJaYLHuexObna2jjkqwgttzQW4KVajeZm I1BPQOSHCQVg+Qo2ewVpd6YrnygpTRF6XsKhA/gwuHS+Jy5xJVsXMPNO/kl0y/6f k9TjfhdTRUpMqn5siO68 =DjF0 -END PGP SIGNATURE- - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Buggy java native call kills tomcat
Hi all, Saw this recently... Some java native code included in our tomcat deployment contained an error, which caused a JVM segfault when a webapp called the routine. I was startled by the fact that this native code error brought down the entire tomcat instance (all 15+ web apps) due to a fault in a native library (installed in tomcat/lib) called by one app. This would have been very scary if this happened in PROD instead of dev/test. My question: can we protect ourselves from total instance failure, i.e. restrict the damage to the death of just the app calling the native library? Is there a way to make the tomcat VM survive a badly written native component? Thanks Dale
RE: Buggy java native call kills tomcat
From: Dale Ogilvie [mailto:dale_ogil...@trimble.com] Subject: Buggy java native call kills tomcat My question: can we protect ourselves from total instance failure, i.e. restrict the damage to the death of just the app calling the native library? Is there a way to make the tomcat VM survive a badly written native component? Simple answer: nope. In a Windows or Linux environment, native code has full access to the entire address space of the process, and can therefore scribble over anything it feels like, corrupting the JVM. In general, you should try to avoid native code called from Java like the plague; not doing so means you might as well write in C (or worse, C++). If you must use native code, running it in a separate process and communicating with it via pipes or sockets is much safer, but then you have to deal with timeouts and other error recovery issues. Another option is to run a webapp that uses native code in a separate Tomcat instance (using a different port), and forward requests to it from the primary Tomcat. - Chuck THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
