Monitoring and alerts
Watch pipeline health and catch failures early.
A pipeline you can't see is a pipeline you can't trust. Databasin gives you a few places to look — pick whichever fits the moment.
Where to look, at a glance
| You want… | Open this |
|---|---|
| The pulse of everything in a project, right now | The project Command Center (home) |
| One pipeline in detail — status, progress, artifacts | The pipeline's detail page |
| The logs from a specific run | The container logs (Refresh to pull) |
| A heads-up that something happened | The notification bell in the top bar |
The Command Center
Open a project and the Command Center is the home screen. Its run-activity panel is the fastest read on what's happening:
- Currently running — a live list of pipelines and automations executing right now.
- Show me shortcuts — one-click Running / Successful / Failed filters, each with a count, spanning both pipelines and automations. Clicking one drops you into the Integrations list filtered to exactly that set.
- A 7-day health ratio — successful runs over all completed runs in the last seven days — for an at-a-glance sense of whether things are trending healthy.
It's the right view for "what's going on across all my work?"
The pipeline detail page
Every pipeline has its own page. The Current Status panel up top shows a status badge plus four metrics for the latest run:
- Completed — artifacts that finished successfully.
- Remaining — artifacts still to process.
- Failed — artifacts that errored.
- Duration — how long the run took.
The status badge uses the real run labels: Currently Running, Success (also shown as Completed), Failed, Stopped, Pending, and Not Started. Below that sits an artifact-progress bar, a history chart of recent runs over the last 30 days (green for all-successful, red for any failures, gray for nothing processed), and a card for each artifact so you can see and manage tables individually.
The page refreshes its status on its own every 30 seconds, and there's a Refresh button if you want to pull the latest immediately.
Run logs are refreshable, not streaming
Open a pipeline's container logs and you'll get the run's log output with a Refresh button. It's a pull, not a live tail — click Refresh to fetch the newest entries. For a finished run, that's everything; for an active run, refresh as it progresses.
Email notifications (built in)
Pipelines have their own email notifications — you don't need to wrap a pipeline in an automation to get them. In the pipeline's configuration there's a Notifications section where you add email addresses; those addresses receive run notifications for that pipeline.
The same Notifications, Schedule, Cluster, and Timeout settings are available both when you build a pipeline and when you edit it — so you can turn on emails at any time.
The notification bell
The bell in the top bar surfaces system notifications as things happen across the app. Click it to see the list and jump to the relevant place. What triggers a given notification is driven by the backend, so treat the bell as a convenient nudge rather than the system of record — the Command Center and the pipeline detail page are where you confirm status.
Debugging a failed run
When a run fails, work the detail page:
- Read the Current Status panel — which artifacts failed, and how far the run got.
- Open the container logs and Refresh to see the error coming from the source, target, or run itself.
- Re-run just what broke — the run menu lets you retry Failed Artifacts without reprocessing everything.
If the logs alone aren't enough, open a support ticket — see Getting support for what to include.
Healthy baselines (rules of thumb)
Not product features, just guidance for running a pipeline practice:
- Watch duration variance. A daily pipeline that usually takes 5 minutes and suddenly takes 90 has almost certainly hit a shape change in the source.
- Watch row-count variance. Order-of-magnitude jumps usually mean a new data source snuck in or a watermark got reset.
- Treat repeat failures as real. A pipeline that fails twice in a row is almost never just "flaky" — open the logs.