Determine whether OpenJ9 builds still need the mingw-cygwin label

[sxa]

All of our machines should be configured the same so the mingw-cygwin label locks it onto the ibmcloud systems unnecessarily (in theory). We should verify whether the openj9 builds will run elsewhere (I think @Haroon-Khel has probably already verified the alibaba windows boxes)

[adamfarley]

Have asked Haroon to confirm about the alibaba machine on Slack.

Running builds on the azure machines now.

[Haroon-Khel]

As per @adamfarley's request, a fresh openj9 build on build-alibaba-win2012r2-x64-1
https://ci.adoptopenjdk.net/job/build-scripts/job/jobs/job/jdk11u/job/jdk11u-windows-x64-openj9/936/console

[sxa]

@Haroon-Khel Isn't that the same test you've already done that worked at adoptium/infrastructure#1818 (comment)?

[Haroon-Khel]

@Haroon-Khel Isn't that the same test you've already done that worked at adoptium/infrastructure#1818 (comment)?

Yes, I mentioned that in the slack thread. Wouldnt hurt to run another one though; more recent

[adamfarley]

The only other two windows machines we could run on (if we remove the mingw tag requirement) are the two azure machines. Here's a pair of jdk8 and jdk16 runs to check those:

build-azure-win2012r2-x64-1 (JDK16) - https://ci.adoptopenjdk.net/job/build-scripts/job/jobs/job/jdk16/job/jdk16-windows-x64-hotspot/57/
build-azure-win2012r2-x64-2 (JDK8) - https://ci.adoptopenjdk.net/job/build-scripts/job/jobs/job/jdk8u/job/jdk8u-windows-x64-openj9/1007/

[adamfarley]

Ok, results:

Azure machines: Builds and tests are green.
Alibaba machine: Build is green, handful of failures in the tests. Checking to make sure that there's nothing new.

Edit: Alibaba test results:

• java_lang_ProcessBuilder_Basicx2
  >'255'< not equal to '0'
  Could be related to this closed issue.

• java_lang_ProcessHandle_OnExitTest
  spawned process should have been in children
  Looks like this closed issue.

• java_lang_ProcessHandle_TreeTest.java x2
  java.lang.AssertionError: Child should be alive: 22788 expected [true] but found [false]
  Looks like this closed issue

• java_lang_StringBuilder_CompactStringBuilderSerialization.java
  JVMJ9VM015W Initialization error for library j9jit29(11): cannot initialize JIT
  Error: Could not create the Java Virtual Machine.
  Error: A fatal exception has occurred. Program will exit.
  Unknown issue. Remote chance it's connected to JDK11 Win64 java/lang/ProcessHandle/TreeTest.java Segmentation error vmState=0x00000000 at J9THR29.dll eclipse-openj9/openj9#10629

• java_lang_reflect_Proxy_ProxyClassAccessTest.java
  JVMJ9VM015W Initialization error for library j9jit29(11): cannot initialize JIT
  Error: Could not create the Java Virtual Machine.
  Error: A fatal exception has occurred. Program will exit.
  Same as above.

Most seem to be "could not start process" problems.

[sxa]

Alibaba machine: Build is green, handful of failures in the tests. Checking to make sure that there's nothing new.

Presumably that was just with the build running on alibaba - the tests were still just running on whatever random machines they happened to land on?

[adamfarley]

Yes. The Alibaba build was tested on test-ibmcloud-win2012r2-x64-1.

[adamfarley]

Re-running the lang tests for the alibaba build on the machine that ran and passed the testing for the azure machines (which was, ironically, build-alibaba-win2012r2-x64-2).

https://ci.adoptopenjdk.net/view/Test_grinder/job/Grinder/7366/console

[adamfarley]

Ok, those all passed.

Looks to me like the sanity.openjdk tests can all pass on builds generated from the alibaba and azure machines, so the mingw label can be safely removed from the list of windows build requirements.

@Haroon-Khel - Your thoughts?