Provide script to detect jobs that are finishing on default github timeout

[apupier]

which means usually that there was a deadlock or infinite loop.
By default it takes 6 hours to cancel the job, in case of jobs which are supposed to run faster, we could determine and configure a lower timeout.

good idea coming from apache/mina#58

[ppkarwasz]

We could also add setting a timeout of about 2× or 3× the usual run time as best practice in:

https://cwiki.apache.org/confluence/display/INFRA/GitHub+Actions+Recommended+Practices

What do you think? The two approaches are complementary and in a repository where maintainers often check the CI, runaway jobs are rare and get fixed rather soon. Maybe a notification to private@ if jobs time out often is the best approach (nobody follows the GitHub notification mailing lists).

[apupier]

We could also add setting a timeout of about 2× or 3× the usual run time as best practice in:

https://cwiki.apache.org/confluence/display/INFRA/GitHub+Actions+Recommended+Practices

agree

. Maybe a notification to private@ if jobs time out often is the best approach (nobody follows the GitHub notification mailing lists).

yes would be nice. Not sure how to easily implement that though

[apupier]

We could also add setting a timeout of about 2× or 3× the usual run time as best practice in:
https://cwiki.apache.org/confluence/display/INFRA/GitHub+Actions+Recommended+Practices

agree

thinking out loud. Does the timeout include the "queue" time or only the "in progress" time?