Open grzesuav opened 2 weeks ago
@grzesuav are there any exceptions that appear in the Cryostat container logs at the time (or within some seconds after) you scale down or delete one of these deployments?
And could you paste the output from:
$ kubectl get -o yaml endpoints
❯ k get endpoints registry -o yaml
apiVersion: v1
kind: Endpoints
metadata:
annotations:
endpoints.kubernetes.io/last-change-trigger-time: "2024-09-02T15:26:47Z
name: registry
namespace: registry
subsets:
- addresses:
- ip: 10.184.uuu.xxx
nodeName: aks-nodepool0609-redacted
targetRef:
kind: Pod
name: registry-8b68c85b8-mjt7n
namespace: registry
uid: 852fcb5b-redacted
- ip: 10.184.fff.rrr
nodeName: aks-nodepool0609-redacted
targetRef:
kind: Pod
name: registry-8b68c85b8-sqktd
namespace: registry
uid: 03b7ba07-redacted
notReadyAddresses:
- ip: 10.184.yyy.xxx
nodeName: aks-nodepool0609-redacted
targetRef:
kind: Pod
name: registry-8b68c85b8-2zj9n
namespace: registry
uid: 66499772-redacted
ports:
- name: http
port: 9000
protocol: TCP
- name: jfr-jmx
port: 9091
protocol: TCP
- name: http-prometheus
port: 9090
protocol: TCP
k get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
registry-8b68c85b8-2zj9n 2/2 Running 0 38s 10.184. aks-nodepool0609-redacted <none> <none>
registry-8b68c85b8-mjt7n 2/2 Running 0 6m22s 10.184. aks-nodepool0609-redacted <none> <none>
registry-8b68c85b8-sqktd 2/2 Running 0 6m43s 10.184. aks-nodepool0609-redacted <none> <none>
I see some various errors in cryostat logs, will continue tomorrow to provide more details
So actually today I see the same targets as in https://github.com/cryostatio/cryostat/issues/634#issuecomment-2324993814 where the pods aren't ther for many hours. Explore-logs-2024-09-03 12_26_01.txt
attaching logs which apper now when I am trying to connect.
@andrewazores what is the name of the kubernetes discovery logger ? Maybe I can filter logs related to that to find something intertesting ?
I think the Logger's name should be io.cryostat.discovery.KubeApiDiscovery
.
Possibly related: https://github.com/cryostatio/cryostat/pull/353 , https://github.com/cryostatio/cryostat/pull/396
Current Behavior
Currently in the topology/target selection I can still see old/nonexisting targets in the vew, even like 5 minutes after pods stopped
Expected Behavior
Non-running pods are removed
Steps To Reproduce
as you can see, replicaset
f459568bf
is quite old, and it does not currently any running podEnvironment
Anything else?
No response