RFR(S): 8213448: [TESTBUG] enhance jfr/jvm/TestDumpOnCrash

[Erik Gahlin]

OK, thanks for fixing. 

My only worry with the test, in general, is that the JVM will be in a state that we can’t recover/dump from, leading to intermittent failures that happens once in a month, or so.

It may make sense to add a mechanism, so it is sufficient if the test succeeds in one out of two/three attempts. The crash dump is a best effort, not a guarantee. That said, it’s early in the release and we may gain some insight into how often those intermittent failures happens if we run it as is.

Erik 

> On 4 Mar 2019, at 02:24, mikhailo.seledtsov at oracle.com wrote:
> 
> I removed the sleeps, and the test still works well. I ran it 25 times on 4 different platforms.
> 
> Here is the updated webrev: http://cr.openjdk.java.net/~mseledtsov/8213448.01/
> 
> 
> Misha
> 
> 
> On 2/27/19 10:54 AM, Mikhailo Seledtsov wrote:
>> Hi Erik,
>> 
>>   I will look into getting rid of thread sleep.
>> 
>> Thank you,
>> Misha
>> 
>> On 2/27/19, 5:26 AM, Erik Gahlin wrote:
>>> Hi Misha,
>>> 
>>> Is there some way we could avoid the thread sleeps?
>>> 
>>> Thanks
>>> Erik
>>> 
>>> 
>>>> On 26 Feb 2019, at 03:37, mikhailo.seledtsov at oracle.com wrote:
>>>> 
>>>> Please review this change that added these enhancements to the test:
>>>>    - added 2 more crash scenarios: crasher via System.halt() and via sending a signal
>>>>    - removed dumponexit=true from the test, since emergency recording should be generated regardless
>>>>    - removed -Xint from test cases, thus testing the default configuration
>>>>    - simplified the logic of identifying the name of the recording file
>>>>    - added diagnostic info to the test
>>>> 
>>>> 
>>>> JBS: https://bugs.openjdk.java.net/browse/JDK-8213448
>>>> Webrev: http://cr.openjdk.java.net/~mseledtsov/8213448.00/
>>>> Testing:
>>>>      1. Locally: exercised the test locally on Linux-x64: PASS
>>>>      2. Test System: ran 20 times on multiple platforms - All PASS
>>>> 
>>>> Thank you,
>>>> Misha
>>>> 
>