On 09/01/2019 03:07, Sergey Bylokhov wrote:
On 07/01/2019 23:31, Alan Bateman wrote:
Tests that consume all the heap and trigger OOME will often have
reliability issues, sometimes this happens when the tests are run
with different combinations of VM options or when there is concurrent
activity as the OOME can be triggered
An existed tests in for beans which use OOM are run as tier3 tests,
and as far as I know do not cause some unexpected failures?
If you do not suggest to rewrite an existed tests, then it will be
better in this case to use the same solution.
If someone has cycles then it would be good to cleanup the other tests
too. You mentioned agentvm in one of your mails and that may be one
motivation to avoid triggering OOME in random places.
-Alan
