[incubator-kie-issues-1324] Live reload does not work for KN CLI plugin

[treblereel]

Closes: apache/incubator-kie-issues#1324

[treblereel]

The cause of this issue is that we continue to send the heartbeat when the CDI container is already shutting down and the PostgreSqlJobServiceManagementRepository is unable to obtain the PgPool. The fix involves stopping the heartbeat first (by checking isLeader()).

This issue can be easily reproduced using the examples/jbpm-compact-architecture-example with the command mvn clean package quarkus:dev -Pdevelopment. To trigger the rebuild just edit CandidateData.java or Offer.java and reload the http://localhost:8080/q/dev-ui/extensions

To verify the fix, build incubator-kie-kogito-apps followed by incubator-kie-tools, replacing 999-20240509-SNAPSHOT with 999-SNAPSHOT.

[treblereel]

@ricardozanini ^^

[kie-ci3]

PR job #1 was: UNSTABLE
Possible explanation: This should be test failures

Reproducer

build-chain build full_downstream -f 'https://raw.githubusercontent.com/${AUTHOR:apache}/incubator-kie-kogito-pipelines/${BRANCH:main}/.ci/buildchain-config-pr-cdb.yaml' -o 'bc' -p apache/incubator-kie-kogito-apps -u #2074 --skipParallelCheckout

NOTE: To install the build-chain tool, please refer to https://github.com/kiegroup/github-action-build-chain#local-execution

Please look here: https://ci-builds.apache.org/job/KIE/job/kogito/job/main/job/pullrequest_jobs/job/kogito-apps-pr/job/PR-2074/1/display/redirect

Test results:

• PASSED: 1664
• FAILED: 1

Those are the test failures:

org.kie.kogito.index.ProcessDataIndexMongoDBIT.testProcessInstanceEvents

1 expectation failed.
JSON path errors doesn't match.
Expected: null
Actual: <[{message=Exception while fetching data (/UserTaskInstances[0]/schema) : Runtime service URL not defined, please review the kogito.service.url system property to point the public URL for this runtime., locations=[{line=1, column=85}], path=[UserTaskInstances, 0, schema], extensions={classification=DataFetchingException}}]>

[kie-ci3]

PR job #2 was: UNSTABLE
Possible explanation: This should be test failures

Reproducer

build-chain build full_downstream -f 'https://raw.githubusercontent.com/${AUTHOR:apache}/incubator-kie-kogito-pipelines/${BRANCH:main}/.ci/buildchain-config-pr-cdb.yaml' -o 'bc' -p apache/incubator-kie-kogito-apps -u #2074 --skipParallelCheckout

NOTE: To install the build-chain tool, please refer to https://github.com/kiegroup/github-action-build-chain#local-execution

Please look here: https://ci-builds.apache.org/job/KIE/job/kogito/job/main/job/pullrequest_jobs/job/kogito-apps-pr/job/PR-2074/2/display/redirect

Test results:

• PASSED: 1664
• FAILED: 1

Those are the test failures:

org.kie.kogito.index.service.InfinispanDomainIndexingServiceIT.testConcurrentProcessInstanceIndex

Assertion condition defined as a org.kie.kogito.index.service.AbstractIndexingServiceIT 1 expectation failed.
JSON path data.ProcessInstances[0].state doesn't match.
Expected: is "COMPLETED"
Actual: ACTIVE
within 30 seconds.

[kie-ci3]

PR job #3 was: UNSTABLE
Possible explanation: This should be test failures

Reproducer

build-chain build full_downstream -f 'https://raw.githubusercontent.com/${AUTHOR:apache}/incubator-kie-kogito-pipelines/${BRANCH:main}/.ci/buildchain-config-pr-cdb.yaml' -o 'bc' -p apache/incubator-kie-kogito-apps -u #2074 --skipParallelCheckout

NOTE: To install the build-chain tool, please refer to https://github.com/kiegroup/github-action-build-chain#local-execution

Please look here: https://ci-builds.apache.org/job/KIE/job/kogito/job/main/job/pullrequest_jobs/job/kogito-apps-pr/job/PR-2074/3/display/redirect

Test results:

• PASSED: 1661
• FAILED: 4

Those are the test failures:

org.kie.kogito.index.infinispan.ProcessDataIndexInfinispanHttpIT.testProcessInstanceEvents

1 expectation failed.
JSON path data.UserTaskInstances[0].potentialGroups[0] doesn't match.
Expected: managers
Actual: null

org.kie.kogito.index.mongodb.ProcessDataIndexMongoDBHttpIT.testProcessInstanceEvents

1 expectation failed.
JSON path data.UserTaskInstances[0].potentialGroups[0] doesn't match.
Expected: managers
Actual: null

org.kie.kogito.index.mongodb.ProcessDataIndexMongoDBKafkaIT.testProcessInstanceEvents

Cannot invoke method size() on null object

org.kie.kogito.index.ProcessDataIndexMongoDBIT.testProcessInstanceEvents

Cannot invoke method size() on null object

[treblereel]

@ricardozanini @wmedvede @dgutierr

From what I see, this error is not related to the PR. Moreover, the other PRs also contain the same error (3 out of 4). I agree with @domhanak, who suggests investigating this. WDYT?

[domhanak]

@treblereel where did I say that? :D unrelated flaky tests should be reported in an issue and we can disregard them :)

On the other hand, to test this with kn workflow plugin - do you have steps to build this with the CLI?
I assume Apps -> kie-tools is the order but not sure what is the current state of kie-tools and the weekly snapshot problem. Any clues?

EDIT: Turns out I can't read properly - To verify the fix, build incubator-kie-kogito-apps followed by incubator-kie-tools, replacing 999-20240509-SNAPSHOT with 999-SNAPSHOT.

Will check soon

[domhanak]

@treblereel Worked well, I was able to finish the use case of live reload properly.
One thing that kinda alarms me is that we are not having any unit or integration test here to verify this issue.

Do you think we could add a test to check this cause: The cause of this issue is that we continue to send the heartbeat when the CDI container is already shutting down and the PostgreSqlJobServiceManagementRepository is unable to obtain the PgPool. Let me know please, we have testHeartbeat() and testConflictHeartbeat() in PostgreSqlJobServiceManagementRepositoryTest maybe we can expand it?

[treblereel]

@domhanak here #1530 (comment) :)

hmm, for some unknown reason, it's green today :)

[domhanak]

@treblereel Yea sometimes the CI is slower, depends on the machine and some tests are flaky.

The comment you are referencing point to different PR, is that right?

[treblereel]

@domhanak I don't think we need tests here, all we need is to check the fix is ok. Moreover, fwiv it's not that easy to write tests for this case. But if you think we need it, let me know, maybe we can merge it as is (in case fix is ok) and create a separate issue for tests

[domhanak]

Ack, created one for me for E2E here - apache/incubator-kie-issues#1380
If you have an idea how to check this on lower levels of testing please create something similar.
Thanks!

[treblereel]

@domhanak thank you