Collect Metrics on BonnyCI

[missaugustina]

Start implementing checks for different things we want to check in BonnyCI. This is different from monitoring checks but not necessarily mutually exclusive (there will likely be overlap).

Currently, there is research to collect metrics from projects on TravisCI. We should collect similar metrics from BonnyCI. Obviously this won't really be useful until we have users, but it's something we can start putting into place now.

See TravisCI research here: https://github.com/BonnyCI/ci-plunder/wiki/Data-Points-Brainstorm Update BonnyCI metrics ideas: https://github.com/BonnyCI/ci-plunder/wiki/BonnyCI-Metrics

• [ ] Count of jobs per (day? hour? minute? author?)
• [ ] Time to complete job (end to end from start to finish)
• [ ] Time to complete individual activity chunks (this may be available within the individual services, we should also see if we can get a start to finish time from whatever outside process is triggering it)
• [ ] Longest and mean dependency chain length (how heavily are people using depends-on?)
• [ ] Gate reset frequency (failures in a gate pipeline that cause other jobs to restart)

[ConsoleCatzirl]

We may also want to look at:

• Wait time (time between job trigger and job start)
• Number of test nodes used per hour? day?

And a couple of things that make sense for Bonny but not for Travis are:

• Longest and mean dependency chain length (how heavily are people using depends-on?)
• Gate reset frequency (failures in a gate pipeline that cause other jobs to restart)

[missaugustina]

Even though we're still figuring out what metrics we want to collect, efforts can be made to start publishing them. Please talk to me if you're interested in working on this.