Component logs
Big picture
View and collect Calico logs.
Value
It is useful to view logs to monitor component health and diagnose potential issues.
Concepts
calico/node logs
The calico/node logs contain log output from the following subcomponents:
- Per-node startup logic
- BGP agent
- Felix policy agent
Components log either to disk within /var/log/calico
, to stdout, or both.
For components that log to disk, files are automatically rotated, and by default 10 files of 1MB each are kept. The current log file is called current
and rotated files have @ followed by a timestamp detailing when the files was rotated in tai64n format.
How to
View logs for a calico/node instance
You can view logs for a node using the kubectl logs
command. This will show logs for all subcomponents of the given node.
For example:
kubectl logs -n calico-system calico-node-xxxx
View logs from the CNI plugin
CNI plugin logs are not available through kubectl and are instead logged both to the host machine’s disk as well as stderr.
By default, these logs can be found at /var/log/calico/cni/
on the host machine.
The container runtime may also display the CNI plugin logs within its own log output.
Configure BGP agent log level
BGP log level is configured via the BGPConfiguration API, and can be one of the following values:
Debug
: enables “debug all” logging for BIRD. The most verbose logging level.Info
: enables logging for protocol state changes. This is the default log level.Warning
: disables BIRD logging, emits warning level configuration logs only.Error
: disables BIRD logging, emits error level configuration logs only.Fatal
: disables BIRD logging, emits fatal level configuration logs only.
To modify the BGP log level:
Get the current bgpconfig settings.
kubectl get bgpconfiguration -o yaml > bgp.yaml
Modify logSeverityScreen to the desired value.
vim bgp.yaml
tip
For a global change set the name to “default”. For a node-specific change set the name to the node name prefixed with “node.”, e.g., “node.node-1”.
Replace the current bgpconfig settings.
kubectl replace -f bgp.yaml
Configure Felix log level
Felix log level is configured via the FelixConfiguration API, and can be one of the following values:
Debug
: The most verbose logging level - for development and debugging.Info
: The default log level. Shows important state changes.Warning
: Shows warnings only.Error
: Shows errors only.Fatal
: Shows fatal errors only.
To modify Felix’s log level:
Get the current felixconfig settings.
kubectl get felixconfiguration -o yaml > felixconfig.yaml
Modify logSeverityScreen to desired value.
vim felixconfig.yaml
tip
For a global change set the name to “default”. For a node-specific change set the name to the node name, e.g., “Calico-Node-1”.
Replace the current felixconfig settings.
kubectl replace -f felixconfig.yaml