On Thu, May 31, 2012 at 8:05 PM, Robert Muir <[email protected]> wrote:
> On Thu, May 31, 2012 at 5:51 PM, Michael McCandless
> <[email protected]> wrote:
>> I think the best option is to ignore the OOME from this test case...?
>>
>> Mike McCandless
>>
>
> I think thats fine for now, but I'm not convinced there is no problem
> at all. However, its not obvious the problem is us, either.
>
> Its easy to see this OOM is related to G1 garbage collector.
>
> This test has failed 3 times in the past couple days (before it never
> failed: i suspect packed ints changes sent it over the edge).
>
> https://builds.apache.org/job/Lucene-Solr-tests-only-trunk-java7/2707/
> https://builds.apache.org/job/Lucene-Solr-tests-only-trunk-java7/2719/
> https://builds.apache.org/job/Lucene-Solr-tests-only-trunk-java7/2723/
>
> All 3 cases are java 7, and all 3 cases uses -XX:+UseG1GC. (Uwe turned
> on GC randomization at lucene revolution)

Aha!  Nice sleuthing :)

So maybe this means G1 isn't as good when heap is limited...

Can we somehow detect / pass property to the JVM when G1 is in use?
Then we can scope down the "ignore OOME" I committed to only when G1
is in use...

Mike McCandless

http://blog.mikemccandless.com

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to