https://rancher.com/ logo
Join Slack
Powered by
Greetings community, i'm researching power cut sce...
# longhorn-storage
r
Greetings community, i'm researching power cut scenario for 
K3s
cluster running in an 
on premises
infrastructure. Cluster components: Storage: Longhorn controller CNI: Cilium DNS: core-dns Scenario description: After power cut incident, when powr is back - cluster struggles to be back to the stable state. Specifically lot of PersistentVolumes are stuck in 
Detached
state preventing different cluster components to get back to up & running state. Question: i would very much appreciate sharing knowledge for such scenarios/use cases, i'm keen to understand best practices for Longhorn controller configuration for 
on premise
infrastructure environment with non-stable electricity.
s
Add a UPS?
r
Thank you for thought. Creating an artificial environment with no power cuts, is not a solution.
s
Just out of curiosity, why not? And why is it "artificial"?
a
I pmuch moved away from longhorn for this reason alot of issues when losing a k8s node with Rwx volumes getting stuck
r
@aloof-salesclerk-86781 wondering, moved to what solution as for Storage backend?
Just out of curiosity, why not? And why is it "artificial"?
@sparse-fireman-14239 bcs having 100% electricity all the time is an artificial environment. Nowadays, cluster should have capabilities to survive and get back to operational state in scenario after "an atomic bomb" hits the cluster. And this is where Platform engineer should get hands dirty 🙂
a
I moved to rook ceph
✅ 1
Open in Slack
PreviousNext
Powered by