RFR: 8304835 jdk/jfr/event/oldobject/TestArrayInformation.java fails with "Could not find event with class ... as (leak) object"

[Aleksey Shipilev]

On Wed, 14 Jun 2023 13:16:53 GMT, Erik Gahlin <egahlin at openjdk.org> wrote:

> > So now the "failing" test would timeout? This does not look right.
> > How many tries it usually takes to get a lucky sample? Take that estimate and do about 10x more iterations and fail, if still not found?
> 
> The test will timeout, not sure that is a problem. Looking at the log, it's obvious why. I don't know how many times are needed, so why not run as many as possible? This is the pattern we have used for other jfr tests.

Well, my concern is that timeouts are mostly treated as the infrastructural problems. I don't think anyone expects the test failure to manifest as timeout: the expectation is that tests complete in reasonable time, whether they pass or fail. Running until timeout is reached also makes the test dependent on TIMEOUT_FACTOR and other infrastructural tunings.

But you're right, JFR tests already do this, so it makes little sense in trying to address it in this PR. It would be nice to see if JFR tests can avoid timeouting on failure, some time later, though.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/14467#issuecomment-1591271354