|
From: Leif M. <le...@ta...> - 2006-07-22 00:33:47
|
Mark, The wrapper decides on a working directory long before the JVM is ever launched. Could you post your wrapper.conf file? Maybe one of our Cruise Control users or myself can see the problem. I know that several other users are using the wrapper with Cruise Control. I know that some have had problems initially, but the ones I am aware of have all gotten up and running. Cheers, Leif Mark Modrall wrote: > > Hi… > > I’ve been trying to use JavaServiceWrapper to get CruiseControl > running on windows xp and so far it’s been reasonably painful. Turns > out most of the problems are due to implicit internal dependencies > inside CC using relative paths and assuming a particular working > directory/file system hierarchy. After figuring that out, I found the > > Wrapper.working.dir= > > config element. That was very helpful but there seems to be something > funky in the way JavaServiceWrapper is implementing it. > > For one thing, CC spits out a cruisecontrol.log file for the sysout > output. Even though the code appears to be running in the cc home > directory (it would fall right over if it wasn’t), when it’s run by > JavaServiceWrapper cruisecontrol.log appears in the directory of > wrapper.exe. > > For another thing, CC has some jmx implementation to do remote > controls for the program. Whatever environment the jmx thread’s using, > it appears to be back in the wrapper.exe directory. The cc jmx > implementation is also using a lot of relative paths and when the web > control page tries to invoke a jmx method, it’s crapping out. The log > file says it’s failing because it’s looking for a subdirectory under > the cc home directory it can’t find. > > So how come wrapper.working.dir can get the main thread running using > the right pwd while the jmx thread’s not? > > Thanks > > -Mark > |