RE: Small connection problems with JK 1.2.15 and IIS 5
Hi, I experienced the same problem as described earlier in this thread. I'm also using Tomcat 5.5 (5.5.15 to be exact), JK 1.2.15 and IIS 5 (Win2k server). I'm running a single Tomcat instance, and it's on the same server as IIS. I was also having these messages, really thousands every day: - [Thu Feb 02 08:21:08 2006] [warn] jk_ajp_common.c (2138): Unable to get the free endpoint for worker ajp13 from 10 slots [Thu Feb 02 08:21:08 2006] [info] jk_ajp_common.c (2152): can't find free endpoint - Increasing the cachesize in worker.properties to 100 indeed fixes these problems! Many tanx to Mladen Turk for his replies earlier in this thread! I've been having this problem since september 2005, and generated A LOT OF page not found errors. These have now disappeared :-) However, the JK logs still show a whole lot of (about 10.000 every day!!) these errors: - [Wed Feb 08 13:03:49 2006] [error] jk_isapi_plugin.c (639): WriteClient failed with 2745 [Wed Feb 08 13:03:49 2006] [info] jk_ajp_common.c (1384): Connection aborted or network problems [Wed Feb 08 13:03:49 2006] [info] jk_ajp_common.c (1731): Receiving from tomcat failed, because of client error without recovery in send loop 0 [Wed Feb 08 13:03:49 2006] [info] jk_lb_worker.c (711): unrecoverable error 400, request failed. Client failed in the middle of request, we can't recover to another instance. - Here's my worker.properties file: - ps=\ workers.tomcat_home=C:\Tomcat5.5 workers.java_home=C:\Java\jre5 worker.list=ajp13,jkstatus worker.ajp13w.type=ajp13 worker.ajp13w.host=localhost worker.ajp13w.port=8009 #note: Tomcat AJP maxThreads is currently at 125 worker.ajp13w.cachesize=100 worker.ajp13w.lbfactor=1 worker.ajp13w.redirect=catch worker.catch.type=ajp13 worker.catch.host=localhost worker.catch.port=8009 worker.catch.lbfactor=1 worker.catch.disabled=true worker.ajp13.type=lb worker.ajp13.balance_workers=ajp13w,catch worker.jkstatus.type=status - I'm not sure yet if these errors actually have any consequences for my site-visitors, since I only just fixed the can't find free endpoint problem. But I do know, those errors must be there for a reason... Anyone? Tanx a lot! Paul Hamer - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: Small connection problems with JK 1.2.15 and IIS 5
Sorry, I just noticed I forgot a piece of the error :) The errors I'm still having are almost always immediately preceeded by this one: [Wed Feb 08 13:27:53 2006] [error] jk_isapi_plugin.c (549): HSE_REQ_SEND_RESPONSE_HEADER failed Tanx, Paul Hamer -Original Message- From: Paul Hamer [mailto:[EMAIL PROTECTED] Sent: Wednesday, 08 February, 2006 13:23 To: users@tomcat.apache.org Subject: RE: Small connection problems with JK 1.2.15 and IIS 5 Hi, I experienced the same problem as described earlier in this thread. I'm also using Tomcat 5.5 (5.5.15 to be exact), JK 1.2.15 and IIS 5 (Win2k server). I'm running a single Tomcat instance, and it's on the same server as IIS. I was also having these messages, really thousands every day: - [Thu Feb 02 08:21:08 2006] [warn] jk_ajp_common.c (2138): Unable to get the free endpoint for worker ajp13 from 10 slots [Thu Feb 02 08:21:08 2006] [info] jk_ajp_common.c (2152): can't find free endpoint - Increasing the cachesize in worker.properties to 100 indeed fixes these problems! Many tanx to Mladen Turk for his replies earlier in this thread! I've been having this problem since september 2005, and generated A LOT OF page not found errors. These have now disappeared :-) However, the JK logs still show a whole lot of (about 10.000 every day!!) these errors: - [Wed Feb 08 13:03:49 2006] [error] jk_isapi_plugin.c (639): WriteClient failed with 2745 [Wed Feb 08 13:03:49 2006] [info] jk_ajp_common.c (1384): Connection aborted or network problems [Wed Feb 08 13:03:49 2006] [info] jk_ajp_common.c (1731): Receiving from tomcat failed, because of client error without recovery in send loop 0 [Wed Feb 08 13:03:49 2006] [info] jk_lb_worker.c (711): unrecoverable error 400, request failed. Client failed in the middle of request, we can't recover to another instance. - Here's my worker.properties file: - ps=\ workers.tomcat_home=C:\Tomcat5.5 workers.java_home=C:\Java\jre5 worker.list=ajp13,jkstatus worker.ajp13w.type=ajp13 worker.ajp13w.host=localhost worker.ajp13w.port=8009 #note: Tomcat AJP maxThreads is currently at 125 worker.ajp13w.cachesize=100 worker.ajp13w.lbfactor=1 worker.ajp13w.redirect=catch worker.catch.type=ajp13 worker.catch.host=localhost worker.catch.port=8009 worker.catch.lbfactor=1 worker.catch.disabled=true worker.ajp13.type=lb worker.ajp13.balance_workers=ajp13w,catch worker.jkstatus.type=status - I'm not sure yet if these errors actually have any consequences for my site-visitors, since I only just fixed the can't find free endpoint problem. But I do know, those errors must be there for a reason... Anyone? Tanx a lot! Paul Hamer - 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: Small connection problems with JK 1.2.15 and IIS 5
Mladen, Thank you very much for your reply. I didn't catch cachesize relates to concurrent users, in this case I will have to change a lot of workers. So I assume that there are (currently) no recommendations regarding a maximum total worker.cachesize for a single IIS5 server, correct? Also, a little comment: it would be useful to have a default higher than 10 for cachesize, or maybe a global setting in the isapi_redirect.properties file. Thanks again, br1. - Original Message - From: Mladen Turk [EMAIL PROTECTED] To: Tomcat Users List users@tomcat.apache.org Sent: Wednesday, January 11, 2006 5:29 PM Subject: Re: Small connection problems with JK 1.2.15 and IIS 5 br1 wrote: Hello, I am new to the list but I must say I used it a lot to find things in the past years. This time I am facing a problem or two but not finding a solution.. small problems luckily. (2) [Fri Jan 06 15:17:20 2006] [warn] jk_ajp_common.c (2138): Unable to get the free endpoint for worker MYWORKER from 20 slots [Fri Jan 06 15:17:20 2006] [info] jk_ajp_common.c (2152): can't find free endpoint Increase the cachesize for the worker. The value should reflect the number of concurrent users on the system. The above messages means that you have more then 20 concurrent users connected. According to a few posts I found I added connectionTimeout=0 to some tomcats and raised the cachesize of some workers that were giving problems (worker.MYWORKER.cachesize=20) Now I still get the error (1) a couple of times a second and the error (2) a few times the day. My questions at this point would be: 1- Are the above entries caused by the new JK 1.2.15 or I did have the same problems and I didn't notice? (the log level was set to emerg) See above comments. 2- I applied connectionTimeout=0 to Tomcats 5.5.x. Is this also needed for older Tomcat versions 5.0/4.1.7/4.0.6 - though in the docs the parameter was not documented for the AJP connector? You don't need this for 5.x (by default connectionTimeout is disabled), but you need that for 4.x (connectionTimeout defaults to 1 minute). 4- In the AJP documentation the min and maxProcessors are deprecated. What should one use instead? They saved my life with Tomcat 4. maxThreads. 6- Too many questions? :-) No :) Regards, Mladen. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Small connection problems with JK 1.2.15 and IIS 5
br1 wrote: Mladen, Thank you very much for your reply. I didn't catch cachesize relates to concurrent users, in this case I will have to change a lot of workers. So I assume that there are (currently) no recommendations regarding a maximum total worker.cachesize for a single IIS5 server, correct? Also, a little comment: it would be useful to have a default higher than 10 for cachesize, or maybe a global setting in the isapi_redirect.properties file. No, because XP (Workstation) versions of IIS are limited to 10 concurrent connections. You will need a Server version of Windows for higher number. That's probably a Microsoft trick to sell the Servers :) Regards, Mladen. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Small connection problems with JK 1.2.15 and IIS 5
myrealbruno wrote: Sorry - the message was cut for some reason. I was also saying that one that decides to use XP Pro for a production machine should be shot (and you could reply that everyone that decides to use Windows should be shot :-) In the end I was only suggesting a global setting for the JK redirector, I had to add cachesize=something to every one of my workers (30). Why do you have 30 workers? The worker should reflect a physical connection to the backend server. It would only make sense if you'd have 30 Tomcats behing. And also wondering what the maximum would be for an average (production) server, I now have around 30*30 connections. It depends on your client load, but I would say that 200 is more then enough for a single box and an average application. See: http://people.apache.org/~mturk/docs/article/ftwai.html Regards, Mladen. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: Small connection problems with JK 1.2.15 and IIS 5
Why do you have 30 workers? The worker should reflect a physical connection to the backend server. It would only make sense if you'd have 30 Tomcats behing. Yes, I do, that's why I was asking.. :-) It depends on your client load, but I would say that 200 is more then enough for a single box and an average application. See: http://people.apache.org/~mturk/docs/article/ftwai.html Thanks a lot, br1. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]