On Tue, 21 Sep 2021 04:54:24 GMT, David Holmes <dhol...@openjdk.org> wrote:
>> Roger Riggs has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> The switch from a Java child to /bin/sleep caused another test
>> to fail on Linux. The cleanup for a test used /usr/bin/pkill "sleep 60".
>> A race between that cleanup and subsequent tests that used sleep 60 or 600
>> could kill the sleep before the test of waitFor completed.
>> Changing the test using pkill to use 59 seconds makes the test cleanup
>> selective to the sleep spawned for that test.
>> The test that was failing (lines 2626-2624) passes consistently.
>
> test/jdk/java/lang/ProcessBuilder/Basic.java line 2217:
>
>> 2215: // A unique (59s) time is needed to avoid killing
>> other sleep processes.
>> 2216: final String[] cmd = { "/bin/bash", "-c", "(/bin/sleep
>> 59)" };
>> 2217: final String[] cmdkill = { "/bin/bash", "-c",
>> "(/usr/bin/pkill -f \"sleep 59\")" };
>
> Maybe future RFE but why do we even need pkill here when we can get the PID
> of the sleep process we create and kill only that process?
I thought of that too, but notice the parens "()" around that /bin/sleep; that
creates and extra level of forked processes and its harder to get that pid.
There probably is a way to traverse the hierarchy but I'll keep it as is for
now.
-------------
PR: https://git.openjdk.java.net/jdk/pull/5239
