Dmitri Pivkine

> First occurrence: > OpenJ9: [150f183](https://github.com/eclipse/openj9/commit/150f183f1276da57e9a699de0bc9e78995e096fb) > OMR: 79c250b > OpenJDK: 39a43ea > > Previous build: > OpenJ9: [150f183](https://github.com/eclipse/openj9/commit/150f183f1276da57e9a699de0bc9e78995e096fb) > OMR: 79c250b > OpenJDK: 39a43ea looks like mentioned builds are...

Softmx option works: according GC verbose log heap contraction occur as expected

Test checks actual memory available in OS. It is sensitive to running environment. We use to have similar failures in the past.

Also there are OOM "heap memory" at the end of the test. Test instructed to contract heap to ~84m: ``` 1STHEAPTYPE Object Memory NULL id start end size space/region 1STHEAPSPACE...

@llxia FYI: Test has requested heap contraction and it succeeds. However seems size of heap after contraction is not enough to complete test framework run, so we got OOM at...

the only check we can do here to differentiate possible JVM problem from environment behaviour it look to gc verbose log to confirm heap is contracted actually. I believe these...

Heap has been contracted below 256m as requested, however test can not see this memory returned back to OS. I think this is environment problem. In general there is no...

> @dmitripivkine this failed on both jdk8 and 11 last night, may be a new problem. I checked both failures - heap is contracted as expected

I just double check again: Heap has been contracted as requested The difference between old and new behaviour before heap allowed to expand above 512m (because `-Xmx1024m`), now it never...

This assertion means `Card` for `srcObject` in `MM_CopyForwardScheme::updateMarkMapAndCardTableOnCopy()` has unexpected state (garbage?). Unfortunately there is no system core stored in artifacts.