dashboard icon indicating copy to clipboard operation
dashboard copied to clipboard

Intermittent 502 in API polls causes UI to redirect

Open colearendt opened this issue 5 years ago • 7 comments

Environment
Installation method: kubernetes YAML file
Kubernetes version:
Dashboard version: 2.0.0-rc7
Operating system:
Node.js version ('node --version' output):
Go version ('go version' output):
Steps to reproduce

Put a proxy in front of the Kubernetes dashboard that intermittently responds (i.e. 1 out of every 10-20k) with a 502 to the polls for things like: https://myurl.com/api/v1/configmap/default?itemsPerPage=10&page=1&sortBy=d,creationTimestamp

Observed result

The dashboard redirects to #/error?namespace=default and presents an ugly UI:

image

Expected result

The dashboard UI should try again with some backoff interval, and should not surface the bad response of this polling to end users.

Comments

This is mostly a problem with my proxy, but wanted to mention it because it feels like an instability in the kubernetes dashboard as well. Handling networking blips more gracefully would definitely give the user a better experience!

colearendt avatar Apr 08 '20 17:04 colearendt

True, this should be improved. However, priority is rather low right now. Thanks for reporting.

maciaszczykm avatar Apr 09 '20 07:04 maciaszczykm

Hi, I have the exact same behavior in my setup. I have a k8s deployment with an HAproxy LoadBalancer in front of the API servers and I'm having an intermittent 502 at the Kubernetes Dashboard. Even if use a keepalive connection for about 60 seconds.

kubectl proxy --accept-hosts='.*' --keepalive=60s

4ng31 avatar May 07 '20 16:05 4ng31

Issues go stale after 90d of inactivity. Mark the issue as fresh with /remove-lifecycle stale. Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta. /lifecycle stale

fejta-bot avatar Aug 05 '20 17:08 fejta-bot

/remove-lifecycle stale

colearendt avatar Aug 10 '20 20:08 colearendt

Issues go stale after 90d of inactivity. Mark the issue as fresh with /remove-lifecycle stale. Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta. /lifecycle stale

fejta-bot avatar Nov 08 '20 21:11 fejta-bot

/freeze

floreks avatar Nov 08 '20 21:11 floreks

/lifecycle frozen

maciaszczykm avatar Nov 30 '20 08:11 maciaszczykm