-
Notifications
You must be signed in to change notification settings - Fork 115
Closed
Labels
component/dashboardGardener DashboardGardener Dashboardeffort/1dEffort for issue is around 1 dayEffort for issue is around 1 daykind/bugBugBug
Description
We have seen that the shoot list on production clusters sometimes does not show any clusters. The list appears empty, even though they can be fetched directly from the kube apiserver e.g. using kubectl.
Restarting the dashboard pod solves the problem temporarily.
We have seen that this issue starts to occur with mass disconnections of websockets, e.g. when the nginx ingress controller detects a configuration change and restarts it's backends.
I could reproduce the issue locally and websockets/ws#1617 (comment) is a pretty good description of what I saw in the debugger. The issue seems to be solved with ws
7.1.2.
There is already a PR on engine.io
to upgrade to ws
7.1.2
Metadata
Metadata
Assignees
Labels
component/dashboardGardener DashboardGardener Dashboardeffort/1dEffort for issue is around 1 dayEffort for issue is around 1 daykind/bugBugBug