On Fri, Mar 18, 2011 at 7:56 AM, Dag H. Wanvik <[email protected]> wrote:
> Kathey Marsden <[email protected]> writes:
>
>>> DERBY-5109 - InterruptResilienceTest.testRAFWriteInterrupted
>>> DERBY-5081 - Intermittent failure in InterruptResilienceTest: IO thread 
>>> times out waiting for another thread to recover channel
>>> DERBY-4922 - DboPowersTest.testReEncrypt
>>> DERBY-5140 - Intermittent failure in InterruptResilienceTest: test thread 
>>> doing join on worker thread gets interrupted
>
> I am volunteering to look at these if somebody can help me reproduce
> them with any kind of frequency.. I haven't been able to reproduce them
> myself yet.
>
> Thanks,
> Dag
>
>

Hi Dag,

Thank you for volunteering!

DERBY-5109 I've not seen in the last 3 weeks with anything other than
ibm 1.4.2. With ibm 1.4.2. on my laptop (with windows XP) I could
reproduce this consistently by just running the test with trunk
builds. Because of this, I've disabled the the entire test with that
jvm...I expect to be busy with the 10.8 release next week, so I'm not
sure when I can experiment with the suggestions you made...Do you have
access to this jvm? Judging from the description/stack trace, do you
think there's a possible derby bug there? Otherwise I'm inclined to
let it sit until later, or maybe even close with 'won't fix', because
I think it rather unlikely anyone will try to use the thread
interrupts with that jvm and a fix in the jvm is unlikely unless
someone has a really critical situation.

re DERBY-4922; I happened to notice each of the two failures described
in this bug pop up in the last two days (each on a different day) in
the nightly tests on windows with the 10.7 branch, with ibm 1.5. I'll
spend some time trying to reproduce this more consistently. If I
cannot, I'll look into possible machine specific issues on the machine
where these tests run, like, number of processors, speed, timing of
the test run (i.e. some other process interfering, why else
specifically with ibm 1.5?) - that sort of thing. I'll post comments
in this bug when I have anything to report...

I've no further input on DERBY-5081 or DERBY-5140...

Myrna

Reply via email to