Troubles with Shenandoah

[Aleksey Shipilev]

On 4/8/19 10:54 PM, Simone Bordet wrote:
> On Mon, Apr 8, 2019 at 10:43 PM Aleksey Shipilev <shade at redhat.com> wrote:
>> This was with batch-count=10K. I did run several iterations of it, spanning around an hour (lots of
>> [Enters] pressed), and there are no lost messages. It used to be every 5-th run ending with lost
>> msgs. What would be a good stress test to run overnight? batch-count=10M?
> 
> Just to be clear, you had no problems with heap locked at max size and
> always pre-touch with 12+33?

Correct. I used jdk12 binary [1] for testing:
12-testing+0-builds.shipilev.net-openjdk-jdk12-b158-20190406-jdk-1233jdk-12-ga. Locking the heap
resolved the latency and lost messages problems for me. YMMV, so please try yourself.

> batch count = 1_000 runs for ~10s.
> 8 hrs -> batch count = 2_880_000
> 
> Make it 2_500_00 and you should be good overnight (I once did precise
> math and the "circa" 10s, when multiplied, yielded hours more :)

Okay!

>> I'd say for evaluation use, you can continue with jdk12u binary and always-committed-heap, unless
>> you experience another problem.
> 
> Do you mean the tip of http://hg.openjdk.java.net/jdk-updates/jdk12u/?

Yes. Our jdk12u nightlies [1] are built from there. You can build it yourself, or just use the
existing binary.

-Aleksey

[1] https://builds.shipilev.net/openjdk-jdk12/