You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request designed to solve a problem? If so describe the problem this feature should solve.:
Currently with a cluster with 50+ nodes, the cluster-autoscaler controller needs a couple of minutes to do scale ups/downs, because it continuously is looking up the nodegroup belonging to a node.
Describe the solution you'd like.:
node metadata is cached, so the main loop runs fast, and is done in a couple of seconds
Describe any alternative solutions you've considered.:
-
Additional context.:
In these logs you see how long the controller currently takes for each node:
I1109 23:12:08.429748 1 log.go:36] exoscale-provider: looking up node group for node ID 1c02381c-275a-42cc-b2ba-6247754992f4
I1109 23:12:08.698391 1 log.go:36] exoscale-provider: found node 1c02381c-275a-42cc-b2ba-6247754992f4 belonging to SKS Nodepool d83024e8-e362-4d54-b259-c3f25d249f51
I1109 23:12:09.222860 1 log.go:36] exoscale-provider: looking up node group for node ID 4f6e4a80-b7d3-4846-bdfe-40759caa208d
I1109 23:12:09.524159 1 log.go:36] exoscale-provider: found node 4f6e4a80-b7d3-4846-bdfe-40759caa208d belonging to SKS Nodepool d83024e8-e362-4d54-b259-c3f25d249f51
I1109 23:12:10.014878 1 log.go:36] exoscale-provider: looking up node group for node ID e6898ac6-d452-4451-9ab3-63a0963c2585
I1109 23:12:10.284943 1 log.go:36] exoscale-provider: found node e6898ac6-d452-4451-9ab3-63a0963c2585 belonging to SKS Nodepool d83024e8-e362-4d54-b259-c3f25d249f51
I1109 23:12:10.832619 1 log.go:36] exoscale-provider: looking up node group for node ID bbb6a12f-4bfa-4781-9bf5-2413fb31b5b0
I1109 23:12:11.117948 1 log.go:36] exoscale-provider: found node bbb6a12f-4bfa-4781-9bf5-2413fb31b5b0 belonging to SKS Nodepool d83024e8-e362-4d54-b259-c3f25d249f51
I1109 23:12:11.688147 1 log.go:36] exoscale-provider: looking up node group for node ID cc65c8a2-8938-44fc-81bc-59b07fef3e47
I1109 23:12:11.947046 1 log.go:36] exoscale-provider: found node cc65c8a2-8938-44fc-81bc-59b07fef3e47 belonging to SKS Nodepool d83024e8-e362-4d54-b259-c3f25d249f51
I1109 23:12:12.518103 1 log.go:36] exoscale-provider: looking up node group for node ID 0b4080da-67ec-4c18-9e35-f1e59de07aef
I1109 23:12:12.792736 1 log.go:36] exoscale-provider: found node 0b4080da-67ec-4c18-9e35-f1e59de07aef belonging to SKS Nodepool d83024e8-e362-4d54-b259-c3f25d249f51
The text was updated successfully, but these errors were encountered:
Which component are you using?:
cluster-autoscaler / cloudprovider / exoscale
Is your feature request designed to solve a problem? If so describe the problem this feature should solve.:
Currently with a cluster with 50+ nodes, the cluster-autoscaler controller needs a couple of minutes to do scale ups/downs, because it continuously is looking up the nodegroup belonging to a node.
Describe the solution you'd like.:
Describe any alternative solutions you've considered.:
-
Additional context.:
In these logs you see how long the controller currently takes for each node:
The text was updated successfully, but these errors were encountered: