Re: [wtp-releng] Test failures with latest CBI build master

[Thanh Ha <thanh.ha@xxxxxxxxxxx>]

I'm running into that Runtime Disconnected issue again. This time it's appearing on the HIPP instance [1]. Even clearing my workspace doesn't appear to be helping.
Anyone know what this happens?

What' worse is the error keeps looping instead of failing the build so it generates a lot of log output if no one cancels it. Last time I saw this error was when I was testing the patch in this bug [2].

Thanh

[1] https://hudson.eclipse.org/webtools/job/cbi-build/18/console
[2] https://bugs.eclipse.org/bugs/show_bug.cgi?id=419113


On 04/01/14 05:33 AM, Mickael Istria wrote:

On 01/03/2014 10:07 PM, Thanh Ha wrote:

Hi Everyone,

Hi Thanh,

I noticed I've been running into test failures with the latest CBI build and when I checked with the HIPP instance it's also failing with the same failure [1]. Is this due to a recent change? is there already a bug open for this?

Victor already answered your question, but I'd like to suggest some tweaks to the CI job config to make that failing tests don't block build and are reported as UNSTABLE build instead of failure.
You might like to change the following things in the job configuration:
* Set maven.test.failure.ignore=true and maven.test.error.ignore=true in the Maven executor. Those will make Maven go ahead even if there are some test failures/errors.
* Configure Hudson JUnit report plugin to look at test reports in */*/target/surefire-reports/*.xml,*/*/*/target/surefire-reports/*.xml and maybe other locations (avoid using **/target/surefire-reports/*.xml because ** takes a while and slows down Hudso)
* Set the Maven flag -fae (or --fail-at-end) to let Maven go as far as possible even is something is wrong, and report errors at the end.

HTH

--
Mickael Istria
Eclipse developer at JBoss, by Red Hat
My blog - My Tweets

_______________________________________________
wtp-releng mailing list
wtp-releng@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/wtp-releng