Latency

The Latency tab, on the Observability page, shows the current round-trip latency measured from four sources, so you can spot network or connectivity problems before they turn into outages.

Sources

• Agents: latency from each connected agent back to KubeWatch.
• Nodes: latency measured from your Kubernetes nodes (TCP to the kubelet port).
• Docker Containers: latency to any container with a published host port, measured on the loopback interface. Containers with no published port aren't reachable from the host network and don't appear here.
• Probes: latency to any targets you configure (for example an internal API or database endpoint).

Each row shows the latest measured latency and whether the target is currently reachable. A row greys out if its agent hasn't reported in over 5 minutes.

Getting data in

Agent, node, and container latency are collected automatically once the agent is connected, no configuration needed. To measure additional endpoints, configure probe targets with the agent's KUBEWATCH_LATENCY_TARGETS env var (comma-separated name=target pairs, each a host:port or http(s)://url). KubeWatch measures round-trip time to each and lists it here.