Repository navigation
Monitoring tab for the dashboard: would an outside contribution be welcome? #5370
sarangtagad-git
started this conversation in
Forum
Replies: 1 comment
|
Hey @sarangtagad-git, thank you for sharing your detailed proposal. We like the idea, in fact, this is something that we have been thinking about implementing ourselves. However, this is a relatively big feature. It would require communication between the backend-, frontend- and design-teams. On top of that, our dashboard is implemented in a private repository, which makes it impossible for us to accept external contributions. We still appreciate your initiative and will keep your proposal in mind as we plan it in the future. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
I'd like to contribute a Monitoring tab to the dashboard. Before writing any code I want to check whether that is something you'd accept from an outside contributor, and how you'd like it scoped.
The gap
The dashboard shows resources one at a time, but there is no place that answers "how is this server doing right now?" at a glance: how many runs succeeded in the last 24 hours, how long they take, which runs are still in flight, which service-account keys have never been used. Today that means clicking through several lists or calling the API by hand.
The idea
One Monitoring page with two tabs. Screenshots of a mockup (sample data, not a finished design):
Overview Tab - Scope set to All Runs
Overview Tab - Scope set to Batch
Live Activity Tab
What already exists
POST /runs/statistics(Run statistics endpoint #4860) already does the group-by and aggregation work (status, pipeline, stack, trigger, user, time buckets; count, duration, step count). From a read of the dashboard source (zenml-io/zenml-dashboard), nothing calls it today. The only statistics the UI uses are the per-project ones.Page.total, so?size=1gives a cheap exact count for resources like stacks, components, and connectors.triggered_by_deploymentfilter so the statistics can separate deployment invocations from batch runs. Without it, an average duration blends two very different workloads.Proposed order of work
I would not send this as one large PR. Each step below would be its own PR, and I'd only start the next one after feedback:
POST /runs/statisticscalls. No backend change.I'm deliberately leaving out anything that needs new backend work, such as percentile durations or stack health, unless you'd want it.
Questions
zenml-dashboardREADME says contributions are licensed under Apache 2.0, but I couldn't find aLICENSEorCONTRIBUTING.mdfile in the repo. Is the CLA used for the main repo also what applies there?I've already contributed #4854 (OTel MetricStore) and #5368 (the filter above), so I'm familiar with the contribution flow on the main repo. Happy to adjust the plan based on your feedback, or to drop the idea if it doesn't fit the roadmap.
All reactions