Hi all, since updating to rancher 2.12 the UI sometimes hangs on "Loading" forever when selecting "pods" or other resources. Other times it does work. kubectl works as usual. Now it seems to have something to do with calico's tigera-operator as there are 1000+ tigera-operator pods in Evicted state. Also the deployment often says "deployment does not have minimum availabilty". Has anyone experienced the same and found a solution?

There's certainly IP addresses to consider if you're using a /24 (although,if you were, you'd have come across the problem long ago), but what about maximum pods per worker? Could that be an issue for you?

apiVersion: crd.projectcalico.org/v1 kind: IPPool metadata: creationTimestamp: "2025-05-20T123958Z" generation: 1 labels: app.kubernetes.io/managed-by: tigera-operator name: default-ipv4-ippool resourceVersion: "1109" uid: e6b64a88-c974-4dcf-8a8b-a0830ad221c6 spec: allowedUses: - Workload - Tunnel blockSize: 26 cidr: 10.42.0.0/16 ipipMode: Never natOutgoing: true nodeSelector: all() vxlanMode: Always

> as there are 1000+ tigera-operator pods in Evicted state

maybe calico runs out of IP's to hand out with so many evicted pods?

tigera-operator runs host-networked (because it is the thing that installs Calico). So it won't be running out of pod IPs, because its not using them 🙂 That said, it shouldn't be being Evicted. Eviction is something the scheduler does when a node doesn't have enough resources. What is the state of your nodes? What is causing the scheduler to evict it?

big thanks, will share here after find ing the cause

Would you be able to remember how many pods in total were that cluster (pre fix)? also how many there are now (with fix) and if the Pods list in the ui correctly loads and is usable?

with the Evicted tigera-operator there were over a thousand but dont remember how many were left after deleting those. right now there's 85 pods total with

k get po -A

thanks, and the Pods list is operational again?

yeah and everything lists pretty quick in the UI again

great news. In Rancher 2.12 there were some big updates to improve performance as resources scale. I'm wondering if this time around the UI failed to load lists due to a general node performance issue impacting rancher performance. If you do hit a similar scenario where the UI fails to load lists (or is slow to) would you be able to confirm if the node is under pressure, and if not see if there are any network errors in the browser or suspicious log lines in the rancher pod logs? That would help us greatly to make sure those improvements really to help