Integrations
Grafana
Pull the logs and traces around a failure
The agent reads your Grafana metrics, logs, traces, alerts and dashboards in the same investigation as the robot's recording. Each answer links to its queries.
Setup: you connect Grafana with a read-only token.
What to use it for
- See what happened around a failure: the service that restarted, the network that dropped, the alert that fired.
- Pull the logs and metrics for one robot and one time window.
- Check whether a fault hits one robot or the whole fleet.
- List the alerts that fired during a mission.
Example
Fictional example.
Ask: "Tug BT-14 stopped on service road C at about 14:30 and needed a restart. What happened?"
Agent:
nav-serverexited after three map-tile timeouts from 14:31:40 (Loki). Map service p99 latency rose to 9.8 s for every tug at that time (Prometheus). It is a fleet-wide back-end slowdown, not a BT-14 fault: 23 timeouts on 6 tugs this week, all around 14:30. [Queries linked]
How to use it
- Connect Grafana in Alloy under Integrations. Use a service account token with the Viewer role.
- Ask in the Alloy app, in Slack, on a Jira ticket or in a recurring report.
- Give a robot name and a time, even an approximate one.
- The connection is read-only.