You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Analyze a stacklog using the interactive webserver:
63
63
@@ -77,7 +77,9 @@ To output a text summary to `out.txt`:
77
77
slowjam -html out.txt /path/to/stack.slog
78
78
```
79
79
80
-
## Real World Example
80
+
## Real World Examples
81
+
82
+
1. Integrating SlowJam with Go binary.
81
83
82
84
Here's an example PR to integrate SlowJam analysis into minikube: [minikube#8329](https://github.com/kubernetes/minikube/pull/8329).
83
85
@@ -95,4 +97,14 @@ What minikube contributors discovered with these results were:
95
97
* Functions which could obviously be run in parallel were executed in serial.
96
98
* Functions which we expected to be fast (<1s) were slow (10s). In many cases we were able to remove or rewrite these functions to do less work.
97
99
98
-
The net result was a 2.5X reduction in start-up latency: from ~66 seconds to ~26 seconds.
100
+
The net result was a 2.5X reduction in start-up latency: from ~66 seconds to ~26 seconds.
101
+
102
+
2. Analyzing an integrated Go binary in kubernetes cluster.
103
+
104
+
Here's an pod config to analyze a Go binary running on a kuberenets cluster: [Pod SlowJam Profile](https://github.com/GoogleContainerTools/kaniko/blob/master/examples/pod-build-profile.yaml)
105
+
106
+
In this Pod Config,
107
+
1. Set the environment variable `STACKLOG_PATH` to to generate slowjam profile.
108
+
2. Copy the generated stack samples to a accessible location in [Container Lifecyle hooks](https://kubernetes.io/docs/tasks/configure-pod-container/attach-handler-lifecycle-event/#define-poststart-and-prestop-handlers) `pre-stop`.
0 commit comments