On 14 May 2012 23:46, Jörg Schaible <[email protected]> wrote: > Hilco Wijbenga wrote: > >> On 11 May 2012 15:56, Hilco Wijbenga <[email protected]> wrote: >>> I have noticed that in a fairly large reactor build the build dies >>> close to the end (about 10 modules to go out of around 90). It simply >>> runs out of memory. Restarting it with -rf allows it to finish without >>> issues. There is a limit to how much memory I can give Maven... :-) >>> >>> So apparently, Maven does not release memory before continuing on with >>> the next module? Or would this be something that is specific to a >>> particular plugin (in my case, it's undoubtedly the GWT Maven plugin)? >> >> I changed a number of things so I'm not sure exactly which one did the >> trick ... but I think we can make a good guess. :-) Oh, and this is >> all with Maven 3.0.3. >> >> I finally upgraded my system from 32bit to 64bit (GNU/Linux). I also >> went from the Sun JDK (1.6.0.32) to IcedTea (7.2.1). Finally, I >> upgraded the Maven Compiler plugin to 2.4. I believe it is mainly the >> first two changes that fixed things. > > I'd bet against it.
Count yourself lucky that you never got the chance because you would have lost that bet. :-) >> Strangely enough, I can now run >> the full build with only 1GB (and I could probably use even less). I >> had 2GB before and that was still not enough... > > Use an older compiler plugin and your memory problem is back ;-) After your challenge, I decided to go back to 2.3.2 and actually try it. :-) The build still runs fine: my memory problem did not come back. So while I am sure 2.4 is an improvement over 2.3.2, it does not seem to have anything to do with my memory problems. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
