Community
Participate
Working Groups
Mac tests have been failing after moving from 10.11 to 10.13 (Bug 537658). It'll be good if we can try running the tests on a 10.14 machine and see if the same failures happen there.
(In reply to Lakshmi Shanmugam from comment #0) > Mac tests have been failing after moving from 10.11 to 10.13 (Bug 537658). > It'll be good if we can try running the tests on a 10.14 machine and see if > the same failures happen there. @Mikael, Any updates on getting a 10.14 machine?
I'm provisioning a new 10.14 machine. It should avail. by the end of the week. Thanks for you patience
A new mac machine has been provisionned. Name is ta9y0-macos-10.14. I've created a test job https://ci.eclipse.org/releng/job/test-mac10.14-webmaster/ (copy of ep410I-unit-mac64) with the following modifications: <<<< export JAVA_HOME=/Library/Java/JavaVirtualMachines/jdk1.8.0_172.jdk/Contents/Home export ANT_HOME=/shared/common/apache-ant-latest >>>> export JAVA_HOME=$(/usr/libexec/java_home -v 1.8) export ANT_HOME=/usr/local/Cellar/ant/1.10.5/libexec A job is currently running, let's wait for the results https://ci.eclipse.org/releng/job/test-mac10.14-webmaster/4/console
The jobs ran properly, but unfortunately, the same tests are failing on this 10.14 machine than on the previous 10.13 https://ci.eclipse.org/releng/job/test-mac10.14-webmaster/lastCompletedBuild/testReport/ https://ci.eclipse.org/releng/job/ep410I-unit-mac64/lastCompletedBuild/testReport/ I guess this is really some changed introduced in macOS 10.12 or 10.13 that causes the regression. I'll let you some time where you can experiment with both agents, but we will eventually have to shut one of them down. Thanks.
(In reply to Mikaël Barbero from comment #4) > The jobs ran properly, but unfortunately, the same tests are failing on this > 10.14 machine than on the previous 10.13 > Can we try to setup a jnlp connection similar to Windows?
(In reply to Mikaël Barbero from comment #4) > I'll let you some time where you can experiment with both agents, but we > will eventually have to shut one of them down. Thanks. Thanks for setting up the machine, Mikael. May be we can switch to the 10.14 machine for the 4.11 release if there are no other problems. Is it possible to have a jnlp connection setup with the 10.14 machine?
This machine is hosted on an external provider and a JNLP connection would require the same trickery as we do for the windows 10 x64 machine. As we are still experimenting instability with such configuration (see bug 541676), I would prefer we wait for it to be resolved before we do any testing with the macOS test machine. Thanks.
I saw that the test machine was updated to 10.14, so a new 10.14 test machine is not required anymore. Can we please try the jnlp connection now as it's working fine on Windows machine?
(In reply to Lakshmi Shanmugam from comment #8) > Can we please try the jnlp connection now as it's working fine on Windows > machine? Ping!
Has there been any change done for the 10.14 machine in the recent days? I am querying for the period of 10th May 2019 to 12th May 2019 because the Bug 537637 - "15 leak tests fail on Mac after moving to 10.13.5" is not reproducible any more on this machine. But there are other tests that still fail on this test machine. So this question is to attempt to understand the reason for few of the test to be fixed automatically and not the other tests.
AFAICT, nothing has changed on this machine.
Is there still an issue with Mac 10.14 machine? Or can we close this issue?
No reply after five months, therefore closing.
(In reply to Frederic Gurr from comment #12) > Is there still an issue with Mac 10.14 machine? Or can we close this issue? Sorry, didn't see this comment before. We have 16 e4.ui tests and 2 SWT tests still failing. As requested in comment#8, can we try to setup a jnlp connection?
Sorry for the late reply, we won't provide jnlp connections to mac machines.
