Triage jdk11.0.4_openj9-0.15.1

[adamfarley]

Triage for release runs found here:

https://ci.adoptopenjdk.net/job/build-scripts/job/openjdk11-pipeline/511/

Results of this issue to be copied into this issue by someone who has access:

AdoptOpenJDK/TSC#99 (comment)

[adamfarley]

Windows xl

• sanity.openjdk
  https://ci.adoptopenjdk.net/job/Test_openjdk11_j9_sanity.openjdk_x86-64_windows_xl/2/consoleFull

  • ReuseAddressTest.java - Address in use (Likely non-blocking)
  • BitSetStreamTest.java - NPE. (Doesn't fail until it keeps failing. Repro'd locally. Not seen with the JIT off.)

x86-64_mac_xl

• sanity.openjdk
  https://ci.adoptopenjdk.net/job/Test_openjdk11_j9_sanity.openjdk_x86-64_mac_xl/7/consoleFull

  • BitSetStreamTest.java - Looks to be the same NPE we see on Windows XL (Investigating)
  • AddNodeChangeListener.java, CheckUserPrefsStorage.sh, RemoveReadOnlyNode.java, and RemoveUnregedListener.java - java.util.prefs.BackingStoreException: Synchronization failed for node '/' (Looks to be an OpenJDK regression on JDK-7184287, but only for XL. So it only affects J9.)

x86-64_mac

• sanity.openjdk
  https://ci.adoptopenjdk.net/job/Test_openjdk11_j9_sanity.openjdk_x86-64_mac/7

  • ClassLoader/Assert.java - timeout (Likely non-blocking)
  • Promiscuous.java - SocketException: Exec format error (Known (closed) issue: JTreg test fail - MacOS : java/nio/channels/DatagramChannel/Promiscuous.java eclipse-openj9/openj9#4128)

• extended.system
  https://ci.adoptopenjdk.net/job/Test_openjdk11_j9_extended.system_x86-64_mac/1/consoleFull

  • SharedClasses.SCM23.MultiThread_0 - Overrun, possibly connected to the 30k+ of JIT-QueueSlotMonitors in the javacores (Generating JIT verbose logs for investigation)
  • SharedClassesAPI_0 - Cannot create cache of requested size: Please check your SHMMAX and SHMMIN settings (Possibly machine issue. Investigating)

s390x_linux

• extended.system
  PASSED

• sanity.openjdk

  • java/lang/StringBuffer/CompactStringBuffer.java - Straight fail

java.lang.AssertionError: expected [0] but found [-1]
...
at CompactStringBuffer.testCompactStringBufferForLatinA(CompactStringBuffer.java:62)

• java/net/httpclient/ConnectTimeoutNoProxyAsync.java - Connection timeout (Likely non-blocking)
• java/net/httpclient/ConnectTimeoutNoProxySync.java - ConnectException: Connection refused (Likely non-blocking)
• java/net/httpclient/ConnectTimeoutWithProxyAsync.java - ConnectException: Connection refused (Likely non-blocking)
• java/net/httpclient/ConnectTimeoutWithProxySync.java - ConnectException: Connection refused (Likely non-blocking)
• java/nio/channels/Selector/LotsOfChannels.java - Timeout (Likely non-blocking)
• java/nio/channels/Selector/SelectWithConsumer.java - Straight fail

java.lang.AssertionError: expected [true] but found [false]
...
at SelectWithConsumer.testCancel(SelectWithConsumer.java:566)

(Investigating)

[smlambert]

hi @adamfarley, has this triage effort completed (with relevant issues raised)? If so, we can close this issue.

[adamfarley]

Hi @smlambert - Issues weren't raised for these at the time, as my focus was on identifying blockers for the release.

Many of these issues haven't been seen since, though one or two have. Will raise issues for, and exclude, the latter,

[adamfarley]

Triage progress:

Windows xl

• sanity.openjdk
  • ReuseAddressTest.java - https://bugs.openjdk.java.net/browse/JDK-8225753 (already fixed, to be merged at next tag level).
  • BitSetStreamTest.java - Not seen in the last 5 days of testing.

x86-64_mac_xl

• sanity.openjdk
  • BitSetStreamTest.java - Not seen in the last 5 days of testing.
  • AddNodeChangeListener.java, CheckUserPrefsStorage.sh, RemoveReadOnlyNode.java, and RemoveUnregedListener.java - Not seen in the last 5 days of testing.

x86-64_mac

• sanity.openjdk

  • ClassLoader/Assert.java - JTReg Test Failure: ClassLoader/Assert.java eclipse-openj9/openj9#6668
  • Promiscuous.java - JTreg test fail - MacOS : java/nio/channels/DatagramChannel/Promiscuous.java eclipse-openj9/openj9#6669

• extended.system

  • SharedClasses.SCM23.MultiThread_0 - very infrequent, discussed with j9 team. likely low priority.
  • SharedClassesAPI_0 - Extended system test failure: SharedClassesAPI_0 eclipse-openj9/openj9#6670

s390x_linux

• sanity.openjdk
  • java/lang/StringBuffer/CompactStringBuffer.java - JTReg Test Failure: CompactStringBuffer.testCompactStringBufferForLatinA eclipse-openj9/openj9#6671
  • java/net/httpclient/ConnectTimeoutNoProxyAsync.java - OpenJDK Regression (jtreg) httpclient tests fail with "connection refused" on zLinux #1246
  • java/net/httpclient/ConnectTimeoutNoProxySync.java - OpenJDK Regression (jtreg) httpclient tests fail with "connection refused" on zLinux #1246
  • java/net/httpclient/ConnectTimeoutWithProxyAsync.java - OpenJDK Regression (jtreg) httpclient tests fail with "connection refused" on zLinux #1246
  • java/net/httpclient/ConnectTimeoutWithProxySync.java - OpenJDK Regression (jtreg) httpclient tests fail with "connection refused" on zLinux #1246
  • java/nio/channels/Selector/LotsOfChannels.java - Could be related to infra issue on httpclient. Wait and see.
  • java/nio/channels/Selector/SelectWithConsumer.java - Not seen in the last few days, though s390 has had many problems causing timeouts and such, so it's hard to tell.
    Also, OpenJDK fixed another issue in the same test a few days ago here. When we merge the next tagged level, could this issue be resolved as well? This can be investigated further if needed.

[adamfarley]

Addendum: excluding the following tests from openJ9 runs on the relevant platforms until their issues are resolved:

ClassLoader/Assert.java
Promiscuous.java
java/lang/StringBuffer/CompactStringBuffer.java

PR: #1248

Won't add the http tests until we're sure it's not just a temporary machine issue.
Won't add SharedClassesAPI_0 because it could be a machine issue.