Lin Hu

Hi @pshipton Could you please review the changes, Thanks

@pshipton I will take a look the issue, Thanks

base on example (https://ci.adoptopenjdk.net/view/Test_system/job/openjdk12_hs_systemtest_x86-64_windows/71/tapResults/), the error is for OpenJDK hotspot G1, please provide the error case on OpenJ9, Thanks CL1 stderr j> 2019/06/10 11:07:07.499 Peak usage smaller than current usage...

Hi @Mesbah-Alam，Could you please share test code or point the test src code location? if it has the issue on both Hotspot and OpenJ9, sounds like issue is on JCL...

another thought it might relate nonheap memory pool(such as codecache?), wait the grinder to catch the issue on OpenJ9 for investigating the details.

> @LinHu2016 : The error is intermittent. Started a Grinder for OpenJ9 here: https://ci.adoptopenjdk.net/view/Test_grinder/job/Grinder/1664/console @Mesbah-Alam the Grinder failed to run the tests.

Hi @lumpfish Could you please provide a failure case? I would like investigate the case. In general for heap memory pools, the peakUsage would be updated at the begin of...

also getUsage(), getPeakUsage() are two independent api, heap usage could be changed between api calls, so for realtime you should call getUsage before calling getPeakUsage if you want to compare...

so above the validation code should be like this ``` long currentUsageUsed = current_usage.getUsed(); long peakUsageUsed = peak_usage.getUsed(); if (peakUsageUsed < currentUsageUsed) { Message.logErr(" Peak usage smaller than current usage...

> so above the validation code should be like this > > ``` > long currentUsageUsed = current_usage.getUsed(); > long peakUsageUsed = peak_usage.getUsed(); > > if (peakUsageUsed < currentUsageUsed) {...