SREs / Incident lab
Checking system

Investigate observed behavior

Start the same evidence workflow without introducing a fault.

The workload sends real HTTP requests. Agents receive logs, metrics, runtime state, database activity, and release history; the injection record is kept outside model context.

Redis outage

Cache dependency · Stop the lab Redis container.

Dependent applications make real Redis calls. An approved action starts the same observed container.

Database blocker

PostgreSQL query path · Hold an exclusive lock on the users table.

Application queries wait on a real PostgreSQL session. Recovery terminates the observed blocker.

Release regression

Sample API image · Replace the API with a prepared image containing a product-handler defect.

The image identity changes. Recovery deploys the previously observed image.

Environment health

Updated every eight seconds

Checking services…