Logs & metrics
Open an app and you get its logs and metrics in the dashboard, without switching to the cluster or running kubectl logs.
There are two levels to this, and the difference matters because one of them needs something installed.
| Works immediately | Needs Observability | |
|---|---|---|
| Logs | Live tail from the running pods | Search, and history beyond what the pods still hold |
| Metrics | A current snapshot of CPU and memory per pod | Trends over time |
Installing Observability
Observability is a single entry in the services catalog, installed once per cluster. It installs a metrics store and a log store with lightweight, project-scoped collectors, and everything it collects stays on your own hardware, so there is no retention bill.
Install it from the cluster's Services catalog. Until you do, the Logs and Metrics tabs show a banner with a link to install it, and the live views keep working in the meantime.
Note that both stores need somewhere to write. On a cluster with no default StorageClass they will not start. See Cluster requirements.
Logs
Open any app or managed service and go to Logs. There are two tabs.
Live streams from the app's pods as lines are written. Filter to one pod to watch a single replica, or to one container to separate a sidecar from the main process. This works whether or not Observability is installed, and it only shows what the running pods still hold: restart a pod and its previous output is gone.
Historical queries the log store. This is the one you want when the pod that misbehaved no longer exists, or when you need to search rather than scroll.
Logs are not available for a resource that has been deleted or is being removed. The interface says so rather than showing an empty stream.
Metrics
Open an app and go to Metrics.
Without Observability you get a snapshot: a row per pod with current CPU and memory. If the app has resource requests or limits set, usage is shown against them, which is usually enough to tell whether something is starved or over-provisioned.
With Observability you also get those figures over time, so you can see whether a spike was a one-off or a trend.
Managed services show status, pods and logs, but no metrics tab.
Next steps
- Managed services — The catalog Observability is installed from.
- Cluster requirements — Why storage matters for both stores.
- Apps — Requests, limits, and health checks.