RFR: 8222533 - jtreg test jdk/internal/platform/cgroup/TestCgroupMetrics.java fails on SLES12.3 linux ppc64le machine

mikhailo.seledtsov at oracle.com mikhailo.seledtsov at oracle.com
Tue May 7 20:55:24 UTC 2019


Please make sure to update copyright year on the SubSystem.java before 
integration.

Thank you,

Misha


On 5/7/19 1:51 PM, mikhailo.seledtsov at oracle.com wrote:
> Looks good,
>
> Misha
>
>
> On 5/7/19 1:25 PM, Bob Vandette wrote:
>> Please review this simple fix for a TestCgroupMetrics.java test failure.
>>
>> --- 
>> a/src/java.base/linux/classes/jdk/internal/platform/cgroupv1/SubSystem.java
>> +++ 
>> b/src/java.base/linux/classes/jdk/internal/platform/cgroupv1/SubSystem.java
>> @@ -108,7 +108,7 @@
>>           try {
>>               List<String> lines = 
>> Files.readAllLines(Paths.get(subsystem.path(), param));
>>               for (String line: lines) {
>> -                if (line.contains(match)) {
>> +                if (line.startsWith(match)) {
>>                       retval = conversion.apply(line);
>>                       break;
>>                   }
>>
>> Under docker we typically only see a single block I/O device so
>> the test passed since both lines containing “Total” are the same value.
>>
>> 8:16 Read 4452352
>> 8:16 Write 0
>> 8:16 Sync 0
>> 8:16 Async 4452352
>> 8:16 Total 4452352
>> Total 4452352
>>
>> It’s possible and likely that there will be multiple devices causing 
>> failures
>> since the test and the Metrics API are not examining the same lines.
>>
>> 249:0 Read 10477568
>> 249:0 Write 294431895552
>> 249:0 Sync 17292476416
>> 249:0 Async 277149896704
>> 249:0 Total 294442373120 <——— The API was returning this
>> 8:16 Read 19017216
>> 8:16 Write 326780178432
>> 8:16 Sync 17398915072
>> 8:16 Async 309400280576
>> 8:16 Total 326799195648
>> 8:0 Read 27092992
>> 8:0 Write 31070281728
>> 8:0 Sync 10728873984
>> 8:0 Async 20368500736
>> 8:0 Total 31097374720
>> Total 652338943488 <—— Test test was comparing to this
>>
>> We are after the line that “startsWith” “Total”.
>>
>> Bob.
>>
>



More information about the hotspot-runtime-dev mailing list