Neo4j: disabled - Githubissues

hipster-labs / jhipster-daily-builds

:construction_worker: :rocket: Additional builds for generator-jhipster repository

Apache License 2.0

19 stars 25 forks source link

Neo4j: disabled #45

Closed pascalgrimaud closed 4 years ago

pascalgrimaud commented 4 years ago

Disabled at https://github.com/hipster-labs/jhipster-daily-builds/commit/f5dbe944f0ac606c90b8617fa07898d8e8ac40e6

Waiting to be fixed in main-generator JHipster. cc @atomfrede -> when it's fixed, we can reactivate it

atomfrede commented 4 years ago

Hope to have some time tomorrow.

atomfrede commented 4 years ago

The main generator will be fixed by https://github.com/jhipster/generator-jhipster/pull/11581 Would like to make a dedicated sample for neo4j (until we have fixed the performance problem with the bidirectional relationships in the bank account example).

atomfrede commented 4 years ago

Test failed with failing assertion. I tried the same configuration locally and it works fine. Will check if I have different application options selected.

atomfrede commented 4 years ago

Okay I found a very strange thing. I used the exact same configuration from the daily builds and mvn verify really failed with the same error as in ci. I run the test in IDEA and it was green. Afterwards it was also green via mvn verify :confounded: I have uploaded the generated sample here if someone can try it https://github.com/atomfrede/neo-daily-build-sample

pascalgrimaud commented 4 years ago

@atomfrede : I'll try it soon

pascalgrimaud commented 4 years ago

@atomfrede :

I manage to easily reproduce the same issue in the CI
sometimes, it failed, sometimes not -> random failures
we got a similar issue with Cassandra: https://github.com/jhipster/generator-jhipster/issues/11169
@SudharakaP tried to fix it with https://github.com/jhipster/generator-jhipster/pull/11204

But in the end, the problem is global

for NoSQL database, each integration tests of our entities depends on "deleteAll" before
the test needs a clean database to pass
we don't have the problem with SQL as we used @Transactional
for unknow reason, the deleteAll doesn't executed well and sometimes, as the order is different, the test can fail
for me, the main refactoring should be done for each tests, which should NOT depends on a clean database

atomfrede commented 4 years ago

Interesting didn't think of that. I cloned the sample on my laptop and each time different tests fail. In fact delete all is not called. It looks like adding that to the beforeEach method seems to fix that.

atomfrede commented 4 years ago

And finally its green :relieved:

atomfrede commented 4 years ago

Will look into the failing webflux build tonight.

pascalgrimaud commented 4 years ago

thanks @atomfrede

pascalgrimaud commented 4 years ago

already fixed