Thanks for the reminder. I'd forgotten about the random element in how
Hudson runs the tests.
I am strongly in favor of controlling which system we run on. For one
thing, I think a release candidate should be run in as many different
environments as possible, and not being able to control which OS Hudson
uses limits that.
Since you obviously remember this problem, but voted for javaspace as my
next bug hunt target, I assume you consider this lower priority the
javaspace failures. Of course, that does not mean it has no priority -
maybe I'll get to this next after javaspace.
Patricia
Jonathan Costers wrote:
This error has happened since day one, when Hudson builds and tests on the
solaris1 instance.
It is specific to Solaris, and does not happen on the ubuntu instances.
I highlighted this happening many times before.
You can check the build output for all solaris builds and compare to the
ubuntu builds if you need confirmation of my statement.
This error is the whole reason why I would like to setup separate builds:
- one running only on solaris instances
- one running only on ubuntu instances
2010/10/24 Patricia Shanahan <[email protected]>
On 10/24/2010 8:55 AM, Apache Hudson Server wrote:
See<https://hudson.apache.org/hudson/job/River-trunk-QA/41/changes>
Changes:
[pats] Add comment documenting use of "+," to embed a comma in a test JVM
argument.
The failure was:
[java]
com/sun/jini/test/spec/lookupdiscovery/MulticastMonitorAllChange.td
[java] Test Failed: Test Failed: com.sun.jini.qa.harness.TestException:
change failed -- waited 870 seconds (14 minutes) -- 3 change event(s)
expected, 0 change event(s) received
This result is more likely to be an intermittent failure due to a
concurrency bug than caused by adding a comment to qaDefaults.properties.
I'll put a VirtualBox to work running the test repeatedly to see if it will
fail for me, and if so whether I can make it happen more often to aid debug.
Patricia
