Honeycomb is integrated with Cased. When something breaks, the agent can pull your observability data to figure out what happened.
What this looks like
During or after an incident, ask questions and get real answers from your Honeycomb data:
- “Show me error rates for the API in the last two hours”
- “What’s the P95 latency for /checkout broken down by region?”
- “How many requests are hitting the slow database query?”
The agent runs the queries and analyzes the results, no need for dashboard hunting. Or use the Honeycomb integration as part of a Cased workflow, to regularly check that your systems are operating well.
Deployment correlation
Something that’s very useful: Cased can check if a deployment happened around the time things started breaking.
Honeycomb markers tell us when code went out. The agent lists them to see what changed and when. Sometimes the first answer you need is just “we deployed 5 minutes before errors spiked”.
Alert coverage
Cased also have metadata support for Honeycomb, so you can ask about monitoring:
- What alerts are configured and what they watch
- Who gets paged when something fires (PagerDuty, Slack, email)
- Whether you’re actually monitoring the thing that just broke
Comes up in retros when you’re figuring out why nobody got woken up.
