Travis CI Insights: better build metrics

Travis CI Discussions & Feedback Feature Requests
1
  • What is the total number of builds that your organization has run over the last month?
  • Are most of your builds public and open source? Or are most of your builds private?
  • Are you building more or less than last week? Do most of our builds pass, error or fail?

You can now answer all of these questions by heading to the new Insights tab!

Development is often data-driven, and your CI should be no exception to that. Starting today, you will get more information about your Travis CI usage, from wait times to overall build status results.

Find out more about how Travis CI Insights work in this blog post: https://blog.travis-ci.com/2019-04-03-travis-ci-insights

Is there anything else that you’d like to see included or improved?
Let us know below!

3

I have some reservations about the Average Queue Time metric, it seems to me that it would be desirable to have a lower queue time. Why was the choice made to make the downward percentage red and the upward percentage green? I have attached a few images to show what I mean. In my humble opinion, they should be swapped.
Travis%20CI%20Insights

3 Likes
4

Hi,
It would be good to get these stats per repo.

Also I agree with zPaw’s comment that decreasing average queue time shouldn’t be red.

3 Likes
5

Thank you so much for spotting, @zPaw - we definitely agree with you, having a lower average queue time is a positive trend which is normally associated with green. The upward green percentage was inherited from the total builds chart and needs to be updated for the average queue time.

This behaviour has been updated and it should be displaying properly now :slight_smile:

2 Likes
6

I have to agree that stats per organization does not provide enough insights (hic), about where the resources are wasted or where to focus on.

Also one critical measure seems to be missing from the graph: the average build duration, measured from the moment it was triggered until it did finish (so it would include queues waiting times). Why is important? As we all know we cam move tests in separated stages that may or may not run in parallel. In the end in order to optimize the throughput various decisions are to be made. We need data in order to optimize our CI pipelines.

7

Just started looking at the insights tab.

The data and the graphs are great. Is there a way to download the raw data?

Can I select the week (or month) to show? (I’d love to be able to compare the current week with the previous.) Related to that, I’d like to be able to see a full week of data. (Right now, if I select “week”, it includes today. Unless I do this at midnight, that data is incomplete. I’d like to be able to see a full week of data.)

Is the data all based on UTC? (Our team is in several different time zones, so the TZ matters quite a bit.)

8

Just opened the Insights tab, but I ran into the problem: