kube-ops-view icon indicating copy to clipboard operation
kube-ops-view copied to clipboard

Ops view loses connection to cluster on EKS

Open dmarkey opened this issue 6 years ago • 4 comments

It works for about 10 minutes then it "flashes" and then it seems to not work anymore. Pity I really like it!

EKS deployment using supplied manifests

dmarkey avatar Dec 19 '18 22:12 dmarkey

Seems to work OK with not using redis

dmarkey avatar Dec 19 '18 23:12 dmarkey

@dmarkey can you check where the problem might be? The frontend assumes that "no data" for some time means the cluster went away, I guess the backend is too slow refreshing in this time frame. Logs with timing would help.

hjacobs avatar Dec 20 '18 08:12 hjacobs

Pretty sure this is related to https://github.com/hjacobs/kube-ops-view/issues/193

The failure starts happening for me about the time redis starts failing to offload data to disk.

I change the security context in the redis deployment to run as UID 100 instead of 1000, and changed the query rate to 10s and it stays running forever.

MarkDeckert avatar Jan 04 '19 18:01 MarkDeckert

@dmarkey I changed the UID to 100 in the deployment file (#193) and it works fine, can you check?

hjacobs avatar Jan 11 '19 22:01 hjacobs