Re:K9s – Manage Your Kubernetes Clusters in Style
by throwanem on Saturday May 30, 2020 @08:59PM
(#23366275)
This might be faster:
kubectl logs deployment/<whatever> | less
kubectl exec -it deployment/whatever sh
Re:K9s – Manage Your Kubernetes Clusters in Style
by neurostimulant on Saturday May 30, 2020 @09:25PM
(#23366405)
The benefit of using tools like k9s and rancher is making it easy to explore your cluster. If you have hundreds of pods and services, using a gui tools to explore and manage your cluster is much easier than using kubectl.
Re:K9s – Manage Your Kubernetes Clusters in Style
by throwanem on Saturday May 30, 2020 @11:23PM
(#23366905)
I've found just the opposite, in clusters with usually about a hundred or so deployments per namespace and concomitant scale in pods, services, etc. GUI tools proved occasionally useful early on as I familiarized myself with k8s concepts and the system architecture at hand. Once I developed that basic level of familiarity, the 30,000' view largely ceased to be useful.
Re:K9s – Manage Your Kubernetes Clusters in Style
by neurostimulant on Sunday May 31, 2020 @05:08AM
(#23368262)
Yup, it's useful for exploration, but once you have a good mental model of the cluster's contents, the gui probably won't be as useful anymore, though personally I like to open the gui dashboard to watch live logs and graphana graph (it somehow calming to me).
If a production bug happens and I need to quickly debug, not much is faster than:
- "k9s" (start the CLI)
- "/myservice" (search for the pod)
- "l" (open the logs)
- "/Exception" (search the logs)
Maybe the bug was a bad config file I need to manually edit:
- "<escape>" (close the log view)
- "s" (exec into the container)
K9s has almost completely replaced "kubectl" for me as well. And the dev team behind k9s is great too. We get a new patch about every week and a new feature every 2 weeks or so on average: https://github.com/derailed/k9s/releases