Re: [jruby-dev] JRuby cruise control

On 8/13/07, Charles Oliver Nutter <charles.nutter-xsfywfwIY+M@xxxxxxxxxxxxxxxx> wrote:

Edwin Wong wrote:
> Hi Charles,
>
> I was wondering how Bamboo is giving false positives? Is it reporting builds
> as failed when they completed successfully, or the other way around?
>
> If you could perhaps let us know your setup of JRuby inside Bamboo, we could
> work to make Bamboo any better and more reliable?

I think Nick Sieger has reported a few issues, but in general Bamboo
seems to eventually die off, as though the spawned builds are not being
cleaned up well. It could be related to the memory and resources used
during our test runs, or it could be bamboo. But in general we always
run test runs on our local machines without incident, so the finger
starts to point Bamboo's way.

Currently there's a JVM-version-specific bug I haven't been able to
track down which causes all builds on the CI machine to fail. Once I can
get that resolved, I'll try to monitor Bamboo a bit more closely and see
if it's still having troubles.

I will note here for the record that Bamboo has been working better ( i.e., no builds failed due to  memory/space issues and no bounces required) since I deployed it to Glassfish.  We still have some inconsistencies, but they appear largely to be platform- or machine-dependent.  (The build machine has 4 cores, while most of us develop w/ 2-core laptops.)  With Tomcat we had numerous problems that I tried to report; see http://jira.atlassian.com/browse/BAM-1423.  You guys from Atlassian are probably aware of those.

Cheers,
/Nick