ARTEMIS-2716 + ARTEMIS-3340 Pluggable Quorum + Sequential activation tracking by franz1981 · Pull Request #3680 · apache/artemis

franz1981 · 2021-07-30T16:40:26Z

https://issues.apache.org/jira/browse/ARTEMIS-2716

and

https://issues.apache.org/jira/browse/ARTEMIS-3340

franz1981 · 2021-08-03T07:55:49Z

Some of the test failures are due to Zookeeper URL/port assignment (there must be something happening at random while starting testable ZK server), but nothing that prevent the feature to work: I'm now giving another round of doc/code/test check and this can be ready to be merged 👍

clebertsuconic · 2021-08-03T22:15:59Z

@franz1981 can you rebase it please?

franz1981 · 2021-08-04T10:21:40Z

@clebertsuconic running CI again after rebased

...java/org/apache/activemq/artemis/tests/integration/cluster/distribution/ClusterTestBase.java

...quorum-ri/src/test/java/org/apache/activemq/artemis/quorum/file/FileDistributedLockTest.java

...src/main/java/org/apache/activemq/artemis/core/server/impl/ReplicationPrimaryActivation.java

clebertsuconic · 2021-08-04T19:11:31Z

artemis-commons/src/test/java/org/apache/activemq/artemis/utils/ThreadLeakCheckRule.java

+         return true;
+      } else if (threadName.contains("RMI Scheduler")) {
+         return true;
+      } else if (threadName.contains("RMI RenewClean")) {


Are you sure you're not hiding a thread leakage? can't you just stop the RMI Stub on this case?

Nope, already checked in the first PR on the quorum vote

artemis-quorum-api/src/main/java/org/apache/activemq/artemis/quorum/MutableLong.java

...orum-ri/src/main/java/org/apache/activemq/artemis/quorum/file/FileBasedPrimitiveManager.java

...st/java/org/apache/activemq/artemis/tests/smoke/quorum/ZookeeperPluggableQuorumPeerTest.java

…ication policies + peer Co-authored-by: franz1981 <nigro.fra@gmail.com>

…og output length, try redirecting output to file for now

gemmellr · 2021-08-18T10:33:30Z

The huge amount of log output generated by zookeeper bits during the tests caused the Travis CI jobs to exceed the allowed log size and abort mid run (e.g https://app.travis-ci.com/github/apache/activemq-artemis/builds/234627453 for the merge).

I worked around it by the above commit b8c779c redirecting the output to file, and also 9baa9a4 to stop maven only outputting error details (as that then meant there wasnt output for so long it thought the build died).

I would wonder if the amount of output can be reduced though, e.g there are lots of warnings about not finding Jetty that seem like they could be removed one way or another (e.g suppressing them, or adding it). Presumably the many many connection refused etc errors could be harder.

franz1981 changed the title ~~ARTEMIS-2716 Pluggable Quorum API + Curator RI and ARTEMIS-3340 Sequential activation tracking for pluggable quorum replication~~ ARTEMIS-2716 + ARTEMIS-3340 Pluggable Quorum + Sequential activation tracking Jul 30, 2021

franz1981 mentioned this pull request Jul 30, 2021

ARTEMIS-3340 layered over ARTEMIS-2716 - activation sequence tracking to protect the journal #3646

Closed

franz1981 force-pushed the ARTEMIS-3340 branch 4 times, most recently from b994449 to 332059c Compare August 2, 2021 11:30

franz1981 force-pushed the ARTEMIS-3340 branch from 332059c to 86c7068 Compare August 3, 2021 12:47

franz1981 force-pushed the ARTEMIS-3340 branch 3 times, most recently from c34b348 to 5454fa3 Compare August 4, 2021 09:35