On 06/28/2011 05:11 PM, Jacques Le Roux wrote:
> Adam,
>
> Did you see Scott message?
I don't see Scott's mail, not even in my junk folder. I was planning
on softening that assert to more of a warning, as there are other
tests in that class I'd like to keep running.
The JVM, as documented, is supposed to clear all soft/weak refs,
before an OOM is thrown. However, I don't know if any queues will be
given notice about the cleared references, and that could be the problem.
However, what Scott is talking about has no bearing on the test case
that failed. That test doesn't use all memory, and doesn't use
UtilCache, which is the only thing that would care about object
references.
I'm looking at the test now. I thought I had solved all race
conditions, it used to be *way* worse.
>
> Scott Gray wrote:
>>> So from what I can tell the ttl works fine on the testExpire test
>>> case but the call to useAllMemory() doesn't successfully cause
>>> the cache to be cleared. I'm guessing it works fine locally because
>>> our systems are slower that the test server and the other
>>> startup threads running cause the ttl to expire the cache rather than
>>> the OOME condition. Haven't yet tested that theory by
>>> increasing the ttl on the memory portion of the test.
>>>
>>> Questions, if anyone knows:
>>> 1. Does the cache actually support expiring entries when the heap
>>> space is running out? If not then this test seems a bit
>>> strange.
>>> 2. If it does support that, where can I find the code to debug it?
>>>
>>> I'd like to get this fixed or disable the test, the false(?)
>>> negatives coming from this isn't doing our CI reporting any good.
>>>
>>> Thanks
>>> Scott
>>>
>>> HotWax Media
>>>
http://www.hotwaxmedia.com>>>
>>
>
> Jacques
>
> Adam Heath wrote:
>> On 06/28/2011 06:19 AM,
[hidden email] wrote:
>>> The Buildbot has detected a new failure on builder ofbiz-trunk while
>>> building ASF Buildbot.
>>> Full details are available at:
>>>
http://ci.apache.org/builders/ofbiz-trunk/builds/1918>>>
>>> Buildbot URL:
http://ci.apache.org/>>>
>>> Buildslave for this Build: isis_ubuntu
>>>
>>> Build Reason: scheduler
>>> Build Source Stamp: [branch ofbiz/trunk] 1140520
>>> Blamelist: erwan
>>>
>>> BUILD FAILED: failed compile_1
>>
>> bother, this is actually my problem. There's a race condition in
>> framework/base, AsyncTTLObjectTest. grumble...
>
Locked
