On Fri, 17 Jun 2022 11:50:08 GMT, Kevin Walls <[email protected]> wrote:
>> The failure is very intermittent and I wasn't able to reproduce it.
>>
>> I suspect that the swap might be changed during test execution so I add the
>> corresponding check.
>>
>> If test still fails after fix we should at least exclude this potential
>> problem.
>
> test/jdk/com/sun/management/OperatingSystemMXBean/TestTotalSwap.java line 76:
>
>> 74:
>> 75: // if swap data from OS chnaged re-read OS and MBean data
>> 76: while (expected_swap_size != getSwapSizeFromOs()) {
>
> Can we put a limit on this?
> OS swap usage could be continually changing, this seems worth a try but we
> want to escape and give up after some number of tries?
> If there is some variation that's making us fail, we could also mask off some
> number of low bits, or a byte or more even, to do an approximate comparison?
> Might want to try that if retrying does not succeed.
I doubt that swap might change so often. The failure is reproduced once in
several months.
Let's check the logs if the test failed.
-------------
PR: https://git.openjdk.org/jdk/pull/9174
