Monitoring¶
Wodby reports server reachability and task results, but it does not provide a complete host-monitoring and alerting service. Install a monitoring agent from your cloud provider or a service such as New Relic Infrastructure, Datadog, or your existing observability platform.
At minimum, monitor:
- CPU utilization and load average;
- available memory and out-of-memory events;
- filesystem bytes and inodes;
- disk latency and throughput;
- network availability and latency;
- Docker, containerd, kubelet, and Kubernetes control-plane health; and
- restarts and readiness of application and Wodby system workloads.
Alert before the server exhausts disk or memory rather than relying only on a Wodby task failure.
Infrastructure 7¶
Use Kubernetes to inspect node and workload state:
kubectl get nodes -o wide
kubectl get pods --all-namespaces -o wide
kubectl -n wodby get pods -o wide
For one application instance:
kubectl get pods -n INSTANCE_UUID -o wide
kubectl describe pod -n INSTANCE_UUID POD_NAME
If your monitoring setup provides the Kubernetes Metrics API, current resource usage is available through:
kubectl top node
kubectl top pods -n INSTANCE_UUID --containers
kubectl top is not a historical monitoring system and will fail when no
Metrics API provider is installed. Use your monitoring platform for retention,
dashboards, and alerts.
Host-level snapshots remain useful during an incident:
free -h
df -hT
vmstat 1
docker stats --no-stream
See CLI commands for system workload logs and component checks.
Infrastructure 6 and older¶
Legacy single-server installations expose application containers directly
through Docker. docker stats provides a live resource snapshot. If you choose
to install a third-party terminal tool such as
ctop, follow its current release and checksum
instructions instead of using an old pinned binary from these docs.
These tools show current container usage only. Continue to use a host-monitoring agent for history and alerts, and do not apply Infrastructure 7 kubeadm or namespace assumptions to a legacy server.