Re: GitHub Actions and integration tests, we have a problem!

[Guangning E]

A little advice on getting logs, I'm not sure I can use this in the
bookkeeper, This is how I get the logs of integration test in pulsar:

https://github.com/apache/pulsar/blob/master/tests/integration/src/test/java/org/apache/pulsar/tests/integration/containers/ChaosContainer.java
https://github.com/apache/pulsar/blob/master/tests/integration/src/test/java/org/apache/pulsar/tests/integration/utils/DockerUtils.java

```
 @Override
public void start() {
    super.start();
    this.tailContainerLog();
    if (this.getContainerName().contains("pulsar-broker")) {
            DockerUtils.runCommandAsync(this.dockerClient,
this.getContainerId(), "tail", "-f", "/var/log/pulsar/broker.log");
        } else if (this.getContainerName().contains("bookie")) {
            DockerUtils.runCommandAsync(this.dockerClient,
this.getContainerId(), "tail", "-f", "/var/log/pulsar/bookie.log");
        } else if (this.getContainerName().contains("functions-worker")) {
            DockerUtils.runCommandAsync(this.dockerClient,
this.getContainerId(), "tail", "-f",
"/var/log/pulsar/functions_worker.log");
        }
}
```

Thanks,
Guangning

Enrico Olivelli <eolive...@gmail.com> 于2020年2月11日周二 下午3:56写道：

> Hi Bookkeepers,
> I think that we have a big problem with GitHub Actions and integration
> tests.
>
> I have sent a dummy PR but Integration Tests fail with a timeout
> https://github.com/apache/bookkeeper/pull/2262
>
> Many times such Integration tests fail with the same result.
> I should add that with github actions only committers can re-run
> checks and this is a problem for contributors
>
> AFAIK Integration tests spawn a few docker containers for bookies
> (different versions) and zookeeper.
>
> Honestly I am not able to run all of them on some machine with low
> resources (like only 2 cores) and the error is the same (see below)
>
> Questions:
> - is GitHub actions a right place for Integration Tests ?
> - is there any way to improve our integrations tests (I don't know,
> the cost is about simply starting the full cluster, not in the tests
> themselves) ? is it worth ?
> - how can we fix this blocker problem ? we could increase the timeout
> ? in my PR I have increased it of 100% without success
> - is there anyway to see docker container logs on github actions ?
>
>
>
> Enrico
>
>
> Tests run: 0, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.25 sec
> 301There are no tests to run.
> 302
> 303Results :
> 304
> 305Tests run: 0, Failures: 0, Errors: 0, Skipped: 0
> 306
> 307[ERROR] Failed to execute goal
> org.apache.maven.plugins:maven-surefire-plugin:2.21.0:test
> (default-test) on project smoke: There was a timeout or other error in
> the fork -> [Help 1]
> 308[ERROR]
> 309[ERROR] To see the full stack trace of the errors, re-run Maven
> with the -e switch.
> 310[ERROR] Re-run Maven using the -X switch to enable full debug logging.
> 311[ERROR]
> 312[ERROR] For more information about the errors and possible
> solutions, please read the following articles:
> 313[ERROR] [Help 1]
> http://cwiki.apache.org/confluence/display/MAVEN/MojoFailureException
> 314[ERROR
> <http://cwiki.apache.org/confluence/display/MAVEN/MojoFailureException314%5BERROR>
> ]
> 315[ERROR] After correcting the problems, you can resume the build
> with the command
> 316[ERROR] mvn <args> -rf :smoke
>