-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Balvindar,
On 9/3/12 9:12 AM, balvindar dhaliwal wrote: > We are using tomcat 7.0.19, for a jsf(richfaces),seam, jpa web > application. Here are the software versions and hardware > infrastructure details. Apache Tomact # 7.0.19 Java run time # > 1.6.0_12 If possible, it would be best to upgrade to the latest version of both of these packages: Java 1.6.0_36 and Tomcat 7.0.30. There are several important security fixes in both of those (relative to where you are now). > We have noticed after about 2 months, the web requests stop being > processed. We had observed this issue twice in the last 5 months. > We are using HTTP/1.1 Connector ( BIO ). The heap and permgen > memory seem fine. There is no out of memory errors in the logs. We > have used jvisualvm to monitor heap, permgen, threads and thread > dump, AND THEY ALL LOOK OK. The thread dump looks fine and there > are no deadlocks/stuck threads observed. Your thread dump attachment was stripped from the list. Can you copy/paste it into a follow post? > Please let me know if you need more information. It's hard to guess without the thread dump, but my initial reaction is that it's connection-pool exhaustion, but you said that your connection pool seems healthy. I see from your server.xml that you are using UserDatabaseRealm: is that actually in use in production? It's probably an inappropriate component to use for anything but the most trivial configuration (like maybe the Tomcat manager app). Are you using any kind of <Realm> or <Resource> specifically configured in in your webapp's context.xml? That would also be good to know. If you are using JDBCRealm, you may have hit an edge case that causes some kind of deadlock (but again, you say there are no deadlocks). There were some fixes to the NIO and APR connectors recently (and your version of Tomcat is now more than a year out-of-date) that fixed apparent deadlocks but I'm not sure if those extended into the BIO realm. I can't see anything in the changelog that would indicate a fix to the BIO connector that would affect you. Again, it's very difficult to speculate without some more data, so sending that thread dump (especially if you can give us several thread dumps over a few minutes) will definitely help. - -chris -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.17 (Darwin) Comment: GPGTools - http://gpgtools.org Comment: Using GnuPG with Mozilla - http://www.enigmail.net/ iEYEARECAAYFAlBKEXEACgkQ9CaO5/Lv0PA4dACgjDfHeAMIJClgb9LrwH/Lv2Oy wIsAoKsu4c/ilXoEppKUX2YB2GgsMS1R =X/Hd -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
