Closed JornShen closed 1 year ago
@JornShen what version of Calico are you running on this cluster?
@JornShen what version of Calico are you running on this cluster?
@caseydavenport we use 3.20.4 version
We made quite a few IPAM garbage collector improvements in Calico v3.23, including this one which might address your problem: https://github.com/projectcalico/calico/pull/5613
@JornShen have upgrading to newer calico fixed your problem? Closing for now, we can reopen if that issue persists.
our cluster use ipam check to check ipam result like this
we found some ip is vxlan ip
And I see the calico-kube-controller log, found calico-kube-controller collect these ip.
we can see the calico-kube-controller collect the ip in the last.
our cluser will add and remove same name node(name will be same) in a period of time. eg node 10.1.154.0.
And node 10.1.154.0 caico node log. the latest time it start is about at 2023-01-08 10:39:07.
AND I also see a strange appearance. in the head the log before:
we can see calico-kube-controller confirmed the ip, But refuse to collect it. But the code logic seems difficult to deduce
/cc @caseydavenport