Limit requeues

[svenfuchs]

This is a draft.

This errors the job after 50 resets and 6 hours after the first reset ("requeue").

/cc @henrikhodne wdyt, any thoughts?

[sarahhodne]

The logic mostly looks right to me, although I'm not super-familiar with the hub codebase anymore. I just have one question about the new helper method to add a log message to the logs, see in-line comment.

[svenfuchs]

Tested this on staging with reduced limits for testing. It errored the job, and made the PUT request to travis-logs:

2016-09-22T19:47:18.900593+00:00 heroku[router]: at=info method=PUT path="/logs/513355" host=travis-logs-staging.herokuapp.com request_id=2fc60658-28af-4519-8170-caf598303c2c fwd="184.73.32.240" dyno=web.1 connect=0ms service=14ms status=204 bytes=124

I guess bytes=124 means the request body was 124 bytes long, and matches the length of the body I have PUT.

However, it didn't set the request body to log.content for some reason:

irb(main):008:0> Job.find(513355).log
> #<Log id: 464479, job_id: 513355, content: "", created_at: "2016-09-22 14:33:32", updated_at: "2016-09-22 19:47:18", aggregated_at: "2016-09-22 19:47:18", archiving: false, ar
hived_at: nil, archive_verified: nil, purged_at: nil, removed_by: nil, removed_at: nil>
rb(main):009:0> Job.find(513355).log.parts
> #<ActiveRecord::Associations::CollectionProxy []>

[svenfuchs]

This seems to work, see https://github.com/travis-pro/team-teal/issues/1284#issuecomment-249155077