Were you ever able to sort this out? We're having the same problem, any help would be much appreciated.
> I've been all around mod_jrun22 in the last couple days, and I know > way more than I want to. However, I can't get it to work quite right. > > > Environment: > Debian Lenny 5.0.6 > Apache 2.2.9, Worker MPM (ideally) > JRun4 build 108673 > CF 8.0.1 Enterprise in multiserver config > > I've compiled the connector with apxs2. Debian's default MPM is the > worker (threaded) MPM. > > Under load I received a bunch of messages in the apache error log, > along the lines of "startSync - detected broken mutex!"... In digging > through the code, it appears unless APACHE_APR_THREAD is defined when > the connector is compiled, the mutex library call is never made. (At > least for linux... Windows connectors have other means) I find this > interesting because wsconfig doesn't define that flag either, and I > can't find it anywhere in /usr/include/... So, I define the flag and > recompile. All the errors go away and the connector works great. > > Except that a couple requests go out to lunch. Under heavy load, > mod_jrun will drop random connections with the following message: > JRun will not accept request. Check JRun web server configuration and > JRun mappings on JRun server. > > Which responds to the client with an ugly 500 error. > > I understand the error when you send a file extension like .html to > JRun and there's no mapping in the web.xml... However, this is for a > cfm page, whether it's a index.cfm found via DirectoryIndex or some > other cfm page on the server. 99%+ of the requests go through fine, > it's just this one request every couple hours that won't behave that's > driving me crazy. So it can't be a config issue. Turning on tracing > would likely write so much information to the log as to be unusable > with the load required to reproduce this issue. > > Switching to the perfork MPM removes the problem entirely. It does, > however, significantly increase CPU load on the server. > > I've used JMeter to send tens of thousands of requests against the > connector without any issues. It's very hard to create artificial > load to reproduce the problem... but real load makes it happen several > times a day. > > Which brings me to my question... Is anyone out there successfully > using mod_jrun22 in an Apache 2.2 environment with the worker MPM? CF8. > 0.1/CF9/CF9.0.1? Success stories? > > I've considered upgrading JRun/CF to more recent versions, but a quick > scan of the JRun connector code in CF9 and 9.0.1 yields only minor > changes... Looks like mod_deflate fixes finally made it into 9.0.1 and > 9.0 mostly pertained to Winsock2 upgrades. > > Any information would be helpful! > > Thanks. Joe ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Order the Adobe Coldfusion Anthology now! http://www.amazon.com/Adobe-Coldfusion-Anthology/dp/1430272155/?tag=houseoffusion Archive: http://www.houseoffusion.com/groups/cf-talk/message.cfm/messageid:340359 Subscription: http://www.houseoffusion.com/groups/cf-talk/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/groups/cf-talk/unsubscribe.cfm
