Tomcat 5 cluster spins out of control periodically
I have two machines running Tomcat 5.0 with java 1.4. Both are clustered together. Every once in a while, one or the other just starts spinning out of control and clocking wall time. This is on hp ux version 11. New machines, lots of resources (16 Gig of RAM) Any ideas? Thanks, Micky - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5 cluster spins out of control periodically
No, not using Apache to front end. Haven't done any thread dumps...yet. I just was throwing feelers out to see if there is any history of this. Tim Funk wrote: Are you using apache in front of tomcat? Have you done thread dumps when in a good state vs a 100% cpu state? Sounds like an inifinite loop. -Tim Micky Williamson wrote: I have two machines running Tomcat 5.0 with java 1.4. Both are clustered together. Every once in a while, one or the other just starts spinning out of control and clocking wall time. This is on hp ux version 11. New machines, lots of resources (16 Gig of RAM) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5 cluster spins out of control periodically
Filip, thanks for the information. I did notice that on our test servers, it doesn't happen as frequently. And in the production server.xml we had: Sender className=org.apache.catalina.cluster.tcp.ReplicationTransmitter replicationMode=pooled/ In the test one it was: Sender className=org.apache.catalina.cluster.tcp.ReplicationTransmitter replicationMode=asynchronous/ What is the better way? Filip Hanik - Dev wrote: probably is the multi cast receiver that freaks out. This can happen if the network cable is unplugged. I will add in a sleep on the receiver so that even in this case it wont freak out. so in the scenario above, its a known problem, checking in a fix right now. if you do provide a dump however, we will know more. or even better, profiling it. Filip - Original Message - From: Micky Williamson [EMAIL PROTECTED] To: Tomcat Users List tomcat-user@jakarta.apache.org Sent: Thursday, February 10, 2005 1:24 PM Subject: Re: Tomcat 5 cluster spins out of control periodically No, not using Apache to front end. Haven't done any thread dumps...yet. I just was throwing feelers out to see if there is any history of this. Tim Funk wrote: Are you using apache in front of tomcat? Have you done thread dumps when in a good state vs a 100% cpu state? Sounds like an inifinite loop. -Tim Micky Williamson wrote: I have two machines running Tomcat 5.0 with java 1.4. Both are clustered together. Every once in a while, one or the other just starts spinning out of control and clocking wall time. This is on hp ux version 11. New machines, lots of resources (16 Gig of RAM) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat 5 cluster spins out of control periodically
hmm is there any difference in performance with pooled vs async? The OS is HPUX 11 Filip Hanik - Dev wrote: ah, you might be having problems with the java.nio package, (TCP) as the conf below you are mentioning, are using those settings. asynch doesn't require an ack message, and some VM's on some platforms have problems sending data back on a NIO channel. linux had this issue unless you set LD_ASSUME_KERNEL for example, Filip - Original Message - From: Micky Williamson [EMAIL PROTECTED] To: Tomcat Users List tomcat-user@jakarta.apache.org Sent: Thursday, February 10, 2005 2:44 PM Subject: Re: Tomcat 5 cluster spins out of control periodically Filip, thanks for the information. I did notice that on our test servers, it doesn't happen as frequently. And in the production server.xml we had: Sender className=org.apache.catalina.cluster.tcp.ReplicationTransmitter replicationMode=pooled/ In the test one it was: Sender className=org.apache.catalina.cluster.tcp.ReplicationTransmitter replicationMode=asynchronous/ What is the better way? Filip Hanik - Dev wrote: probably is the multi cast receiver that freaks out. This can happen if the network cable is unplugged. I will add in a sleep on the receiver so that even in this case it wont freak out. so in the scenario above, its a known problem, checking in a fix right now. if you do provide a dump however, we will know more. or even better, profiling it. Filip - Original Message - From: Micky Williamson [EMAIL PROTECTED] To: Tomcat Users List tomcat-user@jakarta.apache.org Sent: Thursday, February 10, 2005 1:24 PM Subject: Re: Tomcat 5 cluster spins out of control periodically No, not using Apache to front end. Haven't done any thread dumps...yet. I just was throwing feelers out to see if there is any history of this. Tim Funk wrote: Are you using apache in front of tomcat? Have you done thread dumps when in a good state vs a 100% cpu state? Sounds like an inifinite loop. -Tim Micky Williamson wrote: I have two machines running Tomcat 5.0 with java 1.4. Both are clustered together. Every once in a while, one or the other just starts spinning out of control and clocking wall time. This is on hp ux version 11. New machines, lots of resources (16 Gig of RAM) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]