@@ -76,7 +76,7 @@ For more information about the `TracingConfiguration` struct, see
76
76
77
77
# ## kubelet traces
78
78
79
- {{< feature-state for_k8s_version="v1.25 " state="alpha " >}}
79
+ {{< feature-state for_k8s_version="v1.27 " state="beta " >}}
80
80
81
81
The kubelet CRI interface and authenticated http servers are instrumented to generate
82
82
trace spans. As with the apiserver, the endpoint and sampling rate are configurable.
@@ -86,10 +86,7 @@ Enabled without a configured endpoint, the default OpenTelemetry Collector recei
86
86
87
87
# ### Enabling tracing in the kubelet
88
88
89
- To enable tracing, enable the `KubeletTracing`
90
- [feature gate](/docs/reference/command-line-tools-reference/feature-gates/)
91
- on the kubelet. Also, provide the kubelet with a
92
- [tracing configuration](https://github.com/kubernetes/component-base/blob/release-1.25/tracing/api/v1/types.go).
89
+ To enable tracing, apply the [tracing configuration](https://github.com/kubernetes/component-base/blob/release-1.27/tracing/api/v1/types.go).
93
90
This is an example snippet of a kubelet config that records spans for 1 in 10000 requests, and uses the default OpenTelemetry endpoint :
94
91
95
92
` ` ` yaml
@@ -103,6 +100,21 @@ tracing:
103
100
samplingRatePerMillion: 100
104
101
` ` `
105
102
103
+ If the `samplingRatePerMillion` is set to one million (`1000000`), then every
104
+ span will be sent to the exporter.
105
+
106
+ The kubelet in Kubernetes v{{< skew currentVersion >}} collects spans from
107
+ the garbage collection, pod synchronization routine as well as every gRPC
108
+ method. Connected container runtimes like CRI-O and containerd can link the
109
+ traces to their exported spans to provide additional context of information.
110
+
111
+ Please note that exporting spans always comes with a small performance overhead
112
+ on the networking and CPU side, depending on the overall configuration of the
113
+ system. If there is any issue like that in a cluster which is running with
114
+ tracing enabled, then mitigate the problem by either reducing the
115
+ ` samplingRatePerMillion` or disabling tracing completely by removing the
116
+ configuration.
117
+
106
118
# # Stability
107
119
108
120
Tracing instrumentation is still under active development, and may change
0 commit comments