Re: IllegalStateException - org.apache.coyote.tomcat4.CoyoteRequest.doGetSession()
Thanks guys! I went through the code last night, after sending the e-mail. I think I found the problem. Our developer uses dynamically included JSP declaring session=true (by not declaring anything) while the parent declares session=false. We'll go through a set of tests anyway. I'm ashamed to doubt about Tomcat now ;-). Take care all! - Yagiz Erkan - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: IllegalStateException - org.apache.coyote.tomcat4.CoyoteRequest.doGetSession()
[EMAIL PROTECTED] wrote: Hi all, We're running Tomcat 4.1.24 on Win2K sp4. After a few days of use (usually 4-5 days), Tomcat comes to a semi-halt mode and it refuses to process some of the pages. Even the thrown exceptions are not logged. It looks like it's a thread pool related problem because, apparently, the connector runs out of threads. I know that if an error occurs, the thread is supposed to be discarded from the pool to keep the consistency and the integrity however I don't understand why they aren't created again... Please have a look at the stack trace below. Any help/advice would be greatly appreciated! ;-) Thanks, Stack Trace follows: 2003-08-31 01:37:30 JspFactoryImpl: Exception initializing page context java.lang.IllegalStateException: Cannot create a session after the response has been committed Well, this is a valid error, and doesn't indicate a problem. As for TC running out of threads, you should do a thread dump to see what's going on (ctrl+break on Windows in the TC console; I don't know how to do it if TC is running as a service). If the issues persist, you could test TC 5.0.9 beta, which has scalability improvements. Remy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: IllegalStateException - org.apache.coyote.tomcat4.CoyoteRequest.doGetSession()
Well, this is a valid error, and doesn't indicate a problem. Remy, Thanks for your prompt answer. Please don't get me wrong, I don't want to sound arrogant or something, but how come an error may not indicate a problem. ;-) We've already gotten the thread dump. I'll have a look into that as well. Any more ideas? Anybody else maybe? :-) Thanks again, - Yagiz Erkan - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: IllegalStateException - org.apache.coyote.tomcat4.CoyoteRequest.doGetSession()
The stack trace you posted has to do with committing the reponse (which actually means sending data back to the client instead of sitting in a buffer), then trying to allocate a session. Allocating a NEW session involves sending a cookie back to the client to let the client be aware of the new session (so there may be a session). Well, cookies can NOT be created after the response has been committed since the header has already been sent to the client. So ... look into your code and determine where you you may create a new session. Then when that may occur, see if the state may have been previously committed for some reason. The servlet spec has more details about this. -Tim [EMAIL PROTECTED] wrote: Well, this is a valid error, and doesn't indicate a problem. Remy, Thanks for your prompt answer. Please don't get me wrong, I don't want to sound arrogant or something, but how come an error may not indicate a problem. ;-) We've already gotten the thread dump. I'll have a look into that as well. Any more ideas? Anybody else maybe? :-) Thanks again, - Yagiz Erkan - 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: IllegalStateException - org.apache.coyote.tomcat4.CoyoteRequest.doGetSession()
[EMAIL PROTECTED] wrote: Thanks for your prompt answer. Please don't get me wrong, I don't want to sound arrogant or something, but how come an error may not indicate a problem. ;-) It indicates there's likely a bug in the app, not in TC (that's what I meant). You can't create a session if the response has already been committed (a flush or writing a relatively large amount of data would do that). We've already gotten the thread dump. I'll have a look into that as well. Any more ideas? Anybody else maybe? :-) Remy - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]