Pipelines that have been running for hours

A card's pipeline dot shows what your provider says the pipeline is doing. If GitLab says it's running, the dot says running. That's what it should do. But a pipeline can sit in "running" for days, waiting on a job that will never start, and a dot that has said "running" since Monday looks like the board is broken. The board was right. It just wasn't telling you enough.

What changed

A clock on the dot. When a pipeline has been unfinished for three hours or more (running, queued, waiting for a runner and so on), its dot gets a small clock. The dot keeps its colour, and the card stays in its column. GitGumbo still shows your provider's verdict and doesn't replace it.

The tooltip says how long, and why. Hover the dot, or focus the merge request link, and you'll read something like:

Pipeline running for 2 days — it may be stuck. Already failed: lint. Still waiting on: deploy-review (not started after 2 days).

It names the jobs that have already failed and the ones still waiting, oldest first, so you know whether to retry a job, cancel the pipeline or go and look at your runners.

GitLab and GitHub both. On GitLab the clock starts when the pipeline was created. On GitHub it starts when the earliest unfinished check began.

What it doesn't change

The flag is information only. It doesn't move cards, change the dot's colour, or count towards anything in Stats. Three hours is a fixed threshold for now, long enough that an ordinary build never trips it.