RFR(S): 8235454: [TESTBUG] Basic test for JFR event streaming for jdk_jfr_sanity

[Igor Ignatyev]

Hi Misha,

the only problem I have w/ the test is its failure mode, other than by crash, the test can fail only by timeout, which isn't the nicest way to fail. I understand that there is no specific guarantee when an event is received by consumer, nor there are such guarantees for on-close, but I'd really like us to find a better way to detect and report failures. for example, you can replace no-arg CountDownLatch::await calls w/ one which takes timeout value (await(long, j.u.c.TimeUnit)), use some reasonable timeout value and report test failure if the return value isn't true.

Thanks,
-- Igor

> On Dec 9, 2019, at 5:36 PM, mikhailo.seledtsov at oracle.com wrote:
> 
> Please review this change that introduces a very simple basic/sanity test for JFR streaming, and adds this test to jdk_jfr_sanity (which runs in tier-1).
> 
>     JBS: https://bugs.openjdk.java.net/browse/JDK-8235454
>     Webrev: http://cr.openjdk.java.net/~mseledtsov/8235454.00/
>     Testing:
>         1. Ran this test over 200 times on multiple platforms in the test cluster: ALL PASS
> 
>          2. Ran the group jdk_jfr_sanity 80 times: ALL PASS
> 
> Thank you,
> Misha
>