RFR(T) [13] : 8208236 : [TESTBUG] vmTestbase/nsk/stress/stack tests fail by timeout

[David Holmes]

On 13/06/2019 1:45 pm, Igor Ignatyev wrote:
>> On Jun 12, 2019, at 7:25 PM, David Holmes <david.holmes at oracle.com> wrote:
>>
>> On 13/06/2019 10:54 am, Igor Ignatyev wrote:
>>> well, I kinda did. before these tests were converted to jtreg, they had timeout = 30 minutes; my thinking was to set test-specific timeout to a value which will give us 30 minutes w/ the minimal expected timeout, which I assumed is 2. if we are to assume that minimal timeout is 4 (which current default), I can change it to /timeout=450.
>>
>> I see the timeout factor being set between 1 and 16, and for tier3 hotspot it seems to be 1. I can't tell what it is for sure for other tiers. So confusing :(
> timeout factors 1 and 2 are usually used for time-based tests, in other words for the tests which try to spend as much of allocated time budget as possible to do more testing. as of today, our tiers run vmTestbase/nsk/stress/stack tests w/ either default timeout factor 4, 8, 10, or 16, I assume depending on used extra vm flags.
>>
>> Lets leave it as-is.
> "as-is" meaning /timeout=900? or default 120s?

"as-is" meaning "as it is in your webrev" as opposed to reducing to 450.

David

>>
>> Thanks,
>> David
>>
>>> -- Igor
>>>> On Jun 12, 2019, at 5:44 PM, David Holmes <david.holmes at oracle.com> wrote:
>>>>
>>>> Hi Igor,
>>>>
>>>> On 13/06/2019 10:39 am, Igor Ignatyev wrote:
>>>>> http://cr.openjdk.java.net/~iignatyev//8208236/webrev.00/index.html
>>>>>> 19 lines changed: 0 ins; 0 del; 19 mod;
>>>>> Hi all,
>>>>> could you please review this trivial fix which increases timeout of vmTestbase/nsk/stress/stack tests to 15 minutes? the patch is basically 's-main/othervm -main/othervm/timeout=900 -'.
>>>>
>>>> Don't we need to account for the jtreg timeout factor?
>>>>
>>>> Thanks,
>>>> David
>>>>
>>>>> webrev: http://cr.openjdk.java.net/~iignatyev//8208236/webrev.00/index.html
>>>>> JBS: https://bugs.openjdk.java.net/browse/JDK-8208236
>>>>> Thanks,
>>>>> -- Igor
>