Today we are excited to announce the rollout of the Insights dashboard in CircleCI. We know you all have been patient and excited for this feature, and we’re excited to bring it back better than ever. The new Insights dashboard offers engineering teams access to actionable data for optimizing pipelines and getting even more out of CircleCI.
We will be rolling out starting October 1st (today!) and expand to all our cloud-based customers in the coming days. This is a phased rollout, so keep in mind that you may not have it available right away. Once available, the dashboard can be accessed via the left nav in the UI, the Pipelines page, or any Workflow page.
The insights dashboard allows users to:
Track status: See which jobs are failing, which workflows have failing tests, and prioritize efforts for pipeline improvement.
Monitor duration: Find out which workflows or jobs are taking the longest and identify opportunities where caching, parallelization, and our new convenience images can help speed things up.
Optimize pipelines: Get insight into credit spend along with throughput, success rate, and mean time to recovery to get the most out of your CircleCI experience.
We always welcome your feedback here, so feel free to post a reply!
Hi @callafeucht! My current understanding is that the rollout is random so there’s no way to specifically request Insights, but we hope to have this rolled out to as many folks as possible in the next week or so.
Update: 100% of CircleCI users on our cloud offering have access to Insights now
If you’re an on-prem customer, you should be receiving emails with details about availability.
We want to hear from you about your experience with Insights. Leave feedback here and let us know what Insights is doing well and what needs improvement.
For a frontend to the insights API this is quite nice, but I’ve found the insights API lacking and have always had to go through one or two more API layers to get the information I wanted. Hopefully the existence of this frontend will provide an easy way to dig deeper into metrics:
The current time window of 90 days is pretty extreme. If there was a way to shorten that down, that would improve some metrics, mostly jobs, to something more usable. I’ve never cared about things that happened in CI more than a month ago, even during the holidays.
Blanket job failure rates aren’t very interesting in a sufficiently complex repository. I’d love to be able to see which tests or steps failed, and have a link to the failed jobs.
Related to the above, but the most interesting metric to me are the test flakes. Seeing a blanket percentage of failures per test would be much more useful than job failure rates.
We use a merge bot for managing PRs (very similar to bors-ng), and so there are only a few branches we actually care about. For those branches, it can make sense to group them together. Could there be a way to select multiple branches, or even match branches based on a regex?
For the moment, I’ll continue supporting my homebrewed ETL system for finding the metrics I listed above. Again, as a frontend to the existing insights API this is really nice, but as mentioned it’s lacking some functionality that would really drive it over the top. The real insights that come off the API are based in the ability to draw connections between the different levels of abstraction in the system, so I hope there’s some movement in this direction.
Hi @simon-swanson, thanks for the feedback! We plan to add support for more granular time windows as well as Test insights (including flaky tests) in the coming weeks and months. We are also looking into how we can enhance the branch filtering feature so keep an eye out for that as well. Once again, thanks for the suggestions here!
(I remember seeing some historical stats for building time etc.).