Steve Loughran commented on YARN-3337:

# For slider we have what we need: the API calls are in the AM and its config 
# What I do need is something generic for other apps, with Spark the one I'm 
currently looking at
# Robert's SSH-in strategy is OK for local-VM systems where I have the SSH key 
and can automated it; I remember doing something similar to test HA NNs in 
Hadoop 1.x. What SSH does well is that you can then issue a {{kill -19}} to 
suspend a process —and so test liveness monitoring.

What I can't do with his code is 
# run tests against clusters that I don't have SSH keys for (possibly including 
the jenkins builds)
# test on windows
# have some re-usable tests which I can get into ASF code for anyone to use.

API wise, force-kill-container would be enough; while my JUnit tests wouldn't 
need a CLI, test runners in different languages might

> Provide YARN chaos monkey
> -------------------------
>                 Key: YARN-3337
>                 URL: https://issues.apache.org/jira/browse/YARN-3337
>             Project: Hadoop YARN
>          Issue Type: New Feature
>          Components: test
>    Affects Versions: 2.7.0
>            Reporter: Steve Loughran
> To test failure resilience today you either need custom scripts or implement 
> Chaos Monkey-like logic in your application (SLIDER-202). 
> Killing AMs and containers on a schedule & probability is the core activity 
> here, one that could be handled by a CLI App/client lib that does this. 
> # entry point to have a startup delay before acting
> # frequency of chaos wakeup/polling
> # probability to AM failure generation (0-100)
> # probability of non-AM container kill
> # future: other operations

This message was sent by Atlassian JIRA

Reply via email to