Call mixedref testing for all OpenJ9 builds #22
Conversation
|
Thank you for creating a pull request! |
Turn off large heap platforms in the ci builds. More cleanup can be done related to xl packages once the dust settles. Depends adoptium#22 Signed-off-by: Adam Brousseau <[email protected]>
Turn off large heap platforms in the ci builds. More cleanup can be done related to xl packages once the dust settles. Depends adoptium#22 Signed-off-by: Adam Brousseau <[email protected]>
Turn off large heap platforms in the ci builds. More cleanup can be done related to xl packages once the dust settles. Depends adoptium#22 Signed-off-by: Adam Brousseau <[email protected]>
AdamBrousseau
force-pushed
the
openj9_mixed_tests
branch
2 times, most recently
from
f9d19c5 to
d9255a1
Compare
|
run tests |
|
I should also mention, this needs to be a coordinated merge with |
PR TESTER RESULT ✅ All pipelines passed! ✅ |
Turn off large heap platforms in the ci builds. More cleanup can be done related to xl packages once the dust settles. Depends adoptium#22 Signed-off-by: Adam Brousseau <[email protected]>
|
@AdamBrousseau how feasible is it to modify this to only affect certain versions? We are currently blocked on jdk8, but could move jdk11 and 16 to mixedrefs builds. Since I don't know how long it will be until jdk8 is unblocked. it may be better to make the change for jdk11 and 16 to flush out any unexpected problems that will occur once mixedrefs is enabled, giving more time for these problems to be resolved. Not that I would do it on a Friday, but early next week. We can see if jdk8 makes any progress today. |
| context.string(name: 'LABEL_ADDITION', value: additionalTestLabel), | ||
| context.string(name: 'KEEP_REPORTDIR', value: "${keep_test_reportdir}"), | ||
| context.string(name: 'ACTIVE_NODE_TIMEOUT', value: "${buildConfig.ACTIVE_NODE_TIMEOUT}")] | ||
| if (buildConfig.VARIANT == "openj9") { |
There was a problem hiding this comment.
if (buildConfig.VARIANT == "openj9" && versionInfo.major != 8) {
I'll leave it as-is until we decide.
|
@pshipton Were you able to unblock 8 or does @AdamBrousseau need to change this PR? If it's a matter of a few days I'd suggest getting 8 fixed. |
|
Looks like the issue was resolved in the upstream issue (the build is now passing). Hopefully, the J9 PR will be merged today and then we can follow up |
It should be unblocked now. I'll know in the morning based on the overnight test results if we are ready to proceed. If things look good I'll put the PRs at OpenJ9 out for review and once they are merged this will need to be merged as well. This PR and the linked PRs ibmruntimes/openj9-openjdk-jdk11#382
I don't know anything about this, if it was a real problem it must have been unrelated to the blocking jdk8 issue I was referring to, as the fix was only merged within the last few hours. |
|
The overnight testing looks good, I'm discussing with a committer to make the switch to mixedrefs builds within the next few hours. |
|
Seems aarch64 is not yet ready to switch to mixedrefs. |
|
@pshipton I'm setting this to draft so we don't accidentally land it before its deps are ready. |
Turn off large heap platforms in the ci builds. More cleanup can be done related to xl packages once the dust settles. Depends adoptium#22 Signed-off-by: Adam Brousseau <[email protected]>
|
I believe OpenJ9 is ready to proceed. It seems a bad idea to do this now, or on a Friday morning, in case it causes problems at Adopt over the weekend. We'll aim for Monday morning EST. |
Turn off large heap platforms in the ci builds. More cleanup can be done related to xl packages once the dust settles. Depends adoptium#22 Signed-off-by: Adam Brousseau <[email protected]>
| arch = "x86-64" | ||
| } | ||
| def os = buildConfig.TARGET_OS | ||
| testParameters.add(context.string(name: 'PLATFORM', value: "${arch}_${os}_mixed")) |
There was a problem hiding this comment.
As per eclipse-openj9/openj9#11951 (comment), _mixed must not be added for 32-bit platforms.
There was a problem hiding this comment.
Added an exception to the if above for 32. Trying to test it internally but Adopt should test it.
|
As an update is required to handle 32-bit platforms, not sure if this will be ready for merge on Monday. I'll wait for confirmation before we deliver the changes at OpenJ9. |
Change the platform on OpenJ9 tests to be mixed. This will run mixed testing using the existing test jobs. This can be reverted once the default test specs are mixed. Add exception for 32bit (Windows JDK8) as these builds are not mixed. Extracted from adoptium/temurin-build#2441 Resolves adoptium/temurin-build#2426 Depends ibmruntimes/openj9-openjdk-jdk8#477 Signed-off-by: Adam Brousseau <[email protected]>
AdamBrousseau
force-pushed
the
openj9_mixed_tests
branch
from
d9255a1 to
f096aa0
Compare
|
As per eclipse-openj9/openj9#11951 (comment) we are discussing a different approach in TKG, so that this change is not needed at all. |
|
This change is no longer required as it's replaced by adoptium/TKG#155 |
|
@pshipton closing thanks! |
Turn off large heap platforms in the ci builds. More cleanup can be done related to xl packages once the dust settles. Depends #22 Signed-off-by: Adam Brousseau <[email protected]>
…ied_release Fix to ensure Certified bins doesn't get named like Open Edition
Change the platform on OpenJ9 tests to be mixed.
This will run mixed testing using the existing test
jobs. This can be reverted once the default test
specs are mixed.
Add exception for 32bit (Windows JDK8) as these
builds are not mixed.
Extracted from adoptium/temurin-build#2441
Resolves adoptium/temurin-build#2426
Depends ibmruntimes/openj9-openjdk-jdk8#477
Signed-off-by: Adam Brousseau [email protected]
Coordinated merge with
ibmruntimes/openj9-openjdk-jdk11#382
ibmruntimes/openj9-openjdk-jdk16#23
ibmruntimes/openj9-openjdk-jdk8#478
ibmruntimes/openj9-openjdk-jdk#287
eclipse-openj9/openj9#11951