[jdk17] RFR: 8269281: java/foreign/TestUpcall.java times out
The previous fix to this test which used removed the `VerifyDependency` flag worked well, but we're still observing rare timeouts. Looking at the test reports, it seems likely that slightly increasing the timeout would do the trick. ------------- Commit messages: - Increase timeout for TestUpcall Changes: https://git.openjdk.java.net/jdk17/pull/198/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk17&pr=198&range=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8269281 Stats: 1 line in 1 file changed: 0 ins; 0 del; 1 mod Patch: https://git.openjdk.java.net/jdk17/pull/198.diff Fetch: git fetch https://git.openjdk.java.net/jdk17 pull/198/head:pull/198 PR: https://git.openjdk.java.net/jdk17/pull/198
On Thu, 1 Jul 2021 21:11:44 GMT, Maurizio Cimadamore <mcimadamore@openjdk.org> wrote:
The previous fix to this test which used removed the `VerifyDependency` flag worked well, but we're still observing rare timeouts. Looking at the test reports, it seems likely that slightly increasing the timeout would do the trick.
Hi Maurizio, Sorry but I don't see why you think increasing the timeout will help here. From what I can see a normal successful run takes less than a minute, or just over. I see a couple of outliers at 4 and 6 minutes, then you have a timeout after 10 minutes - that is 10x slower than expected. I think something unexpected is happening here to cause it to take so long. David ------------- PR: https://git.openjdk.java.net/jdk17/pull/198
On Thu, 1 Jul 2021 21:11:44 GMT, Maurizio Cimadamore <mcimadamore@openjdk.org> wrote:
The previous fix to this test which used removed the `VerifyDependency` flag worked well, but we're still observing rare timeouts. Looking at the test reports, it seems likely that slightly increasing the timeout would do the trick.
After an offline discussion we agreed to leave the test unchanged at the moment. It is not immediately clear if the problem is the test (which runs in ~40 seconds) or with specific machines, as we're seeing some very wild spikes. The test uses some logic to prod the GC (so that cleaners are run and off-heap resources are released) - but the same logic is also present in the dual TestDowncall which has never timed out. ------------- PR: https://git.openjdk.java.net/jdk17/pull/198
On Thu, 1 Jul 2021 21:11:44 GMT, Maurizio Cimadamore <mcimadamore@openjdk.org> wrote:
The previous fix to this test which used removed the `VerifyDependency` flag worked well, but we're still observing rare timeouts. Looking at the test reports, it seems likely that slightly increasing the timeout would do the trick.
This pull request has been closed without being integrated. ------------- PR: https://git.openjdk.java.net/jdk17/pull/198
participants (2)
-
David Holmes
-
Maurizio Cimadamore