| Re: [jakartaee-tck-dev] workaround for GlassFish 6.0 not always starting correctly... |
Hi,We had our first perfect Full Platform TCK run last night with zero failures when running with the [4] fix! [5] was run with the GlassFish zip [6] that was built for testing the change.
This is very encouraging and may mean that the TCK reliability problem is solved.
Thank you Jonathan Coustick for addressing the failure via [4]! Very much appreciated!
Scott [4] https://github.com/eclipse-ee4j/glassfish/pull/23264[5] https://ci.eclipse.org/jakartaee-tck/job/jakartaee-tck-scottmarlow/job/darranl-unauthorized_status_code/2/testReport [6] https://ci.eclipse.org/glassfish/job/glassfish_build-and-test-using-jenkinsfile/job/PR-23264/1/artifact/bundles/glassfish.zip
On 10/15/20 12:31 PM, Scott Marlow wrote:
Hi,As best as we can tell, the "Glassfish start-domain fails after waiting for long" stability problem [1] that we have been hitting occasionally during TCK testing does have a user workaround as is mentioned in the comments [2].I think that we will look to apply this workaround in the Platform TCK, changing the first time that we start GlassFish, to instead start it twice (once without the glassfish/modules/monitoring-core.jar file which we will move out of GlassFish) and then again with the monitoring-core.jar file back in place, which seems to work around the failure for me (on my local machine) and others.Perhaps we could make this change in the various scripts under docker folder [3] match with `grep -ri asadmin.*start-domain -l`.When the [1] issue is fixed later, we can remove the double running of GlassFish (note the first run will fail but the second run of GlassFish should start correctly more consistently).Thoughts? Scott [1] https://github.com/eclipse-ee4j/glassfish/issues/23191[2] https://github.com/eclipse-ee4j/glassfish/issues/23191#issuecomment-708977368[3] https://github.com/eclipse-ee4j/jakartaee-tck/tree/master/docker