Failures running Javac for DDR_Test

[andrewcraik]

PR builds are failing regularly trying to run the DDR_Tests. Failures report a compile step failed related to javac with cryptic messages similar to:

10:48:33 
10:48:33 Total time: 1 second
10:48:33 makefile:82: recipe for target 'compile' failed
10:48:33 make: *** [compile] Error 1
[Pipeline] }

see https://ci.eclipse.org/openj9/job/PullRequest-Sanity-JDK11-aix_ppc-64_cmprssptrs-OpenJ9/85/console

[andrewcraik]

@DanHeidinga @pshipton FYI. @keithc-ca could this be related to any of the recent DDR changes? Not trying to place blame just trying to figure out if this is a problem with javac or a problem with DDR - any thoughts?

[andrewcraik]

FYI @0xdaryl and @liqunl due to interest on the OpenJ9 triage slack channel

[DanHeidinga]

I think I've seen it on non-DDR builds as well - looking for a link now.

Similar failure for IllegalAccessError_for_protected_method in https://ci.eclipse.org/openj9/job/PullRequest-Sanity-JDK8-osx_x86-64_cmprssptrs-OpenJ9/65/console

[gacholio]

Could the problem here be that the javac is the one just built from the PR. Any flakiness in the code could case the compiler to fail before getting to actual testing (which is a serious error that we continue to propagate).

[keithc-ca]

Yes, I think the test code is being compiled with the brand new JDK. I agree it's serious if javac in that JDK doesn't work.

[pshipton]

I think we found the culprit
#4017

[DanHeidinga]

@pshipton Is this fixed now that #4017 has been merged? If so we should close it out.

[AdamBrousseau]

No, 4017 is what caused this issue. I am still determining why though. I would like to figure it out so we can get in a fix rather than simply reverting the change.

[DanHeidinga]

@AdamBrousseau What's the downside of reverting now and re-introducing the change later? Currently PR builds are broken which affects all PRs.