RFR: 8236106: [TESTBUG] Shenandoah: Make TestThreadFailure more resilient

Zhengyu Gu zgu at redhat.com
Tue Dec 17 17:34:57 UTC 2019


Sure.

-Zhengyu

On 12/17/19 12:21 PM, Roman Kennke wrote:
> I'll also backport it to sh/jdk11 right away, because that's where I 
> need it most. Ok?
> 
> Roman
> 
>> Good.
>>
>> Thanks,
>>
>> -Zhengyu
>>
>> On 12/17/19 11:31 AM, Roman Kennke wrote:
>>> Currently, we see frequent timeouts with TestThreadFailure. On my 
>>> local machine, this test standalone can run 1:30, on CI box it can 
>>> exceed the 8minutes limit. Problem seems to be that it's running into 
>>> back-to-back-GCs but not into OOM fast enough.
>>>
>>> Decreasing the max heapsize of the launched process seems to complete 
>>> much faster (in under 2 minutes on the CI box) and thus more resilient.
>>>
>>> http://cr.openjdk.java.net/~rkennke/JDK-8236106/webrev.00/
>>>
>>> Testing: hotspot_gc_shenandoah, several runs of the test standalone
>>>
>>> Ok?
>>>
>>
> 



More information about the shenandoah-dev mailing list