Re: decrease idle threads in the pool faster...
Carlos Rodríguez Colino wrote: Hi all... I have just installed Tomcat 4.1 on a Solaris box and I have realized a surprising effect: Thread pool creates new threads quickly when a request burst arrives, but the freeing when they become idle is very slow (besides, the computer is almost full cpu loaded). Due to this, several bursts carry out reaching the max number of processes, and Tomcat stops working. My intention is to force idle recent-used threads (lwp in Solaris) to be killed in a shorter period of time in order to mantain almost constant the number of lwps in the pool. So, I was going to dive into CoyoteConnector code to discover and decrease any freeing timeout of the idle threads, but perhaps some of you can guide me better to get what I need. I didn't look in the code recently, but doesn't the threads do a wait() and get a notify() when there is a new request/connection? In that case you can do a wait(timeout) and after wait returns check if it is woken up for a new request/connection or because of the timeout. If it's the timeout, you can end the thread. Just a thought. Greetings, Ronald. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: decrease idle threads in the pool faster...
ERROR: There is no such list IDLE here. SOLUTION: Send a message to [EMAIL PROTECTED] with a subject of 'info' (no quotes) for a list of available mailing lists. -- Sincerely, the Minimalist - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
decrease idle threads in the pool faster... (II)
Hi everybody I have been searching previous mails about killing idle threads, and I found lots of them describing the same problem (lots of requests carry out new threads that never are released, so, in a long-term, max number of proccesses is reached and Tomcat stops). Some surprising answers were such as: Yes, no attempt is being made at killing processing threads that werecreated. OTOH, they should be reused if some high load situation occurs again (no new threads will be created). Unless you're really short on OS resources, I don't see that being a huge problem. Remy (12/2001... about TC4.0.1) Do you keep thinking the same? It seems to be a serious problem from my point of view!! Burst of requests increase the number of threads used by Tomcat (I work nowadays with TC4.1.12), but I think they should be released later on. Must they grow until infinite? Must I set maxProcessors to 1 for avoiding Max number of processes reached??? :-P A very complex system relies on Tomcat and a lot of servlets, so I need a solution. Some mails, like mine, proposed to dive in the code to change the management of idle threads and kill them. If there isn't any scheduled action about this problem, could anybody give me a hint about where to look in the code? Thanks in advance and best regards, Carlos. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: decrease idle threads in the pool faster... (II)
- Original Message - From: Carlos Rodríguez Colino [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, June 30, 2003 1:21 AM Subject: decrease idle threads in the pool faster... (II) Hi everybody I have been searching previous mails about killing idle threads, and I found lots of them describing the same problem (lots of requests carry out new threads that never are released, so, in a long-term, max number of proccesses is reached and Tomcat stops). Some surprising answers were such as: Yes, no attempt is being made at killing processing threads that were created. OTOH, they should be reused if some high load situation occurs again (no new threads will be created). Unless you're really short on OS resources, I don't see that being a huge problem. Remy (12/2001... about TC4.0.1) Do you keep thinking the same? It seems to be a serious problem from my point of view!! Burst of requests increase the number of threads used by Tomcat (I work nowadays with TC4.1.12), but I think they should be released later on. Must they grow until infinite? Must I set maxProcessors to 1 for avoiding Max number of processes reached??? :-P A very complex system relies on Tomcat and a lot of servlets, so I need a solution. Some mails, like mine, proposed to dive in the code to change the management of idle threads and kill them. If there isn't any scheduled action about this problem, could anybody give me a hint about where to look in the code? Well, the place to look is j-t-c/util/java/org/apache/tomcat/net/PoolTcpEndpoint.java. With my experience, it works very well with the stand-alone connector. On one very old (and brain-dead) Linux system, I had to set a soTimeOut for the Jk-Connector to get it to kill-off Threads (but this is really a work-around to a problem with that Linux version). Thanks in advance and best regards, Carlos. - 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]
decrease idle threads in the pool faster...
Hi all... I have just installed Tomcat 4.1 on a Solaris box and I have realized a surprising effect: Thread pool creates new threads quickly when a request burst arrives, but the freeing when they become idle is very slow (besides, the computer is almost full cpu loaded). Due to this, several bursts carry out reaching the max number of processes, and Tomcat stops working. My intention is to force idle recent-used threads (lwp in Solaris) to be killed in a shorter period of time in order to mantain almost constant the number of lwps in the pool. So, I was going to dive into CoyoteConnector code to discover and decrease any freeing timeout of the idle threads, but perhaps some of you can guide me better to get what I need. Thanks in advance and best regards, Carlos Rodríguez - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
