Your dashboard shows a latency spike. To investigate it, you copy the query, open logs in a new tab, paste and modify the query, lose your dashboard filters, and repeat for traces. By the time you find the issue, you have 15 tabs open.
Starting today, you can click any panel and investigate right there. All your filters and variables carry over. No more tab juggling.
What We Built
Every dashboard panel now has a right-click menu with actual debugging tools. Here's what you can do:
1. Jump between signals without losing context
You're looking at a metrics panel showing high latency. Right-click it and you'll see options to "View in Logs" or "View in Traces". Click either one - the explorer opens with the same filters, same time range, same everything. You immediately see the logs or traces related to that latency spike.
This works from any panel type. Whether you're looking at a metrics panel, a logs panel, or a traces panel, you can always view the related data in:
- Logs explorer (View in Logs)
- Traces explorer (View in Traces)
All your dashboard context - time range, filters, variables - automatically carries over to the new view.
2. See and edit queries directly
Remember having to switch to edit mode just to see what query powers a panel? Now you can see it right in view mode:
histogram_quantile(0.99, sum(rate(http_request_duration_seconds_bucket[5m])) by (le, service))
Want to check P95 instead of P99? Change it right there. Add a filter for production only? Do it. The query updates live. If you like the changes and have edit access, save it as a new panel.
3. Filter tables by clicking cells
Got a table showing errors by service? Click "payment-service" - now the table only shows payment service data. Click "500" in the status column - now it's only 500 errors. Click "response_time > 1000" - now it's only slow 500 errors from payment service.
The query builds as you click:
WHERE service='payment-service'
AND status_code=500
AND response_time > 1000
4. Break out data to find what's actually broken
Here's the killer feature. Your dashboard shows 500 errors per minute. But which service? Which endpoint? Which region?
Click the error spike and select "Break out by service":
- payment-service: 450 errors
- auth-service: 40 errors
- cart-service: 10 errors
Payment service is the problem. Click it, break out by endpoint:
- /api/process-payment: 440 errors
- /api/validate-card: 10 errors
Click /api/process-payment, break out by error_type:
- connection_timeout: 440 errors
Found it. The payment service can't connect to something. Three clicks, 30 seconds.
5. Link to your actual tools
Dashboards don't exist in a vacuum. You need to check Jenkins, look at GitHub, create Jira tickets. Context Links make this one click.
External tools
https://github.com/company/repo/commit/{{commit_hash}} https://jenkins.company.com/job/{{deployment_id}} https://app.stripe.com/payments/{{payment_id}}
Internal navigation
/trace/{{_trace_id}} /dashboard/service-metrics?service={{service_name}} /logs?query=correlation_id:{{correlation_id}}
See a trace ID in your dashboard? Click it to open trace details. See a deployment ID? Click it, Jenkins opens. Configure whatever links make sense for your workflow.
6. Update dashboard variables from anywhere
Your dashboard has a $region
variable. You see problems in eu-west-1. Instead of scrolling up to the dropdown, just click "eu-west-1" in any panel and select "Set $region". The whole dashboard updates.
Need a new variable? Click a value, create the variable on the spot. No config needed.
Example of a debugging session
Here's an actual production issue debugged with Interactive Dashboards:
- Alert fires: Checkout latency > 2s
- Open dashboard, see the spike, right-click it
- Break out by endpoint
- /api/payment is slow, everything else is fine
- Click /api/payment
- View traces
- All showing timeouts to payment-provider
- Click a timeout trace
- View logs
- "connection pool exhausted" everywhere
- Click the error
- Open runbook (context link)
- Instructions for scaling connection pool
Two minutes from alert to root cause. No tab switching. No query copying. Just clicking through the investigation.
Works on all your dashboards right now
Every dashboard you have is already interactive. No migration, no setup. Just start clicking panels.
Note: Only panels built with query builder are supported. Panels using raw ClickHouse queries aren't interactive yet.
Want to get more out of it?
- Add context links to your runbooks and tools
- Name dashboard variables consistently
- Use GROUP BY in your queries so breakout analysis works
Who benefits
If you're on-call: Debug without leaving the dashboard. Everything is one click away.
If you manage a team: Non-engineers can now investigate issues. "Check what's wrong with payments" becomes a task anyone can do.
If you run platform: Your dashboards now connect directly to your CD pipeline, your runbooks, your everything.
What's coming
We're actively working on expanding Interactive Dashboards with more capabilities. Stay tuned for updates on future enhancements.
Interactive Dashboards are live now. Next time you see something weird in a panel, just click it.