-
Bug
-
Resolution: Done
-
Normal
-
None
-
4.10.z
-
None
-
Important
-
None
-
Plumbing Sprint 227, Plumbing Sprint 228, Plumbing Sprint 229, Plumbing Sprint 230
-
4
-
Rejected
-
False
-
Description of problem:
When creating a pod with an additional network that contains a `spec.config.ipam.exclude` range, any address within the excluded range is still iterated while searching for a suitable IP candidate. As a result, pod creation times out when large exclude ranges are used.
Version-Release number of selected component (if applicable):
How reproducible:
with big exclude ranges, 100%
Steps to Reproduce:
1. create network-attachment-definition with a large range: $ cat <<EOF| oc apply -f - apiVersion: k8s.cni.cncf.io/v1 kind: NetworkAttachmentDefinition metadata: name: nad-w-excludes spec: config: |- { "cniVersion": "0.3.1", "name": "macvlan-net", "type": "macvlan", "master": "ens3", "mode": "bridge", "ipam": { "type": "whereabouts", "range": "fd43:01f1:3daa:0baa::/64", "exclude": [ "fd43:01f1:3daa:0baa::/100" ], "log_file": "/tmp/whereabouts.log", "log_level" : "debug" } } EOF 2. create a pod with the network attached: $ cat <<EOF|oc apply -f - apiVersion: v1 kind: Pod metadata: name: pod-with-exclude-range annotations: k8s.v1.cni.cncf.io/networks: nad-w-excludes spec: containers: - name: pod-1 image: openshift/hello-openshift EOF 3. check pod status, event log and whereabouts logs after a while: $ oc get pods NAME READY STATUS RESTARTS AGE pod-with-exclude-range 0/1 ContainerCreating 0 2m23s $ oc get events <...> 6m39s Normal Scheduled pod/pod-with-exclude-range Successfully assigned default/pod-with-exclude-range to <worker-node> 6m37s Normal AddedInterface pod/pod-with-exclude-range Add eth0 [10.129.2.49/23] from openshift-sdn 2m39s Warning FailedCreatePodSandBox pod/pod-with-exclude-range Failed to create pod sandbox: rpc error: code = DeadlineExceeded desc = context deadline exceeded $ oc debug node/<worker-node> - tail /host/tmp/whereabouts.log Starting pod/<worker-node>-debug ... To use host binaries, run `chroot /host` 2022-10-27T14:14:50Z [debug] Finished leader election 2022-10-27T14:14:50Z [debug] IPManagement: {fd43:1f1:3daa:baa::1 ffffffffffffffff0000000000000000} , <nil> 2022-10-27T14:14:59Z [debug] Used defaults from parsed flat file config @ /etc/kubernetes/cni/net.d/whereabouts.d/whereabouts.conf 2022-10-27T14:14:59Z [debug] ADD - IPAM configuration successfully read: {Name:macvlan-net Type:whereabouts Routes:[] Datastore:kubernetes Addresses:[] OmitRanges:[fd43:01f1:3daa:0baa::/80] DNS: {Nameservers:[] Domain: Search:[] Options:[]} Range:fd43:1f1:3daa:baa::/64 RangeStart:fd43:1f1:3daa:baa:: RangeEnd:<nil> GatewayStr: EtcdHost: EtcdUsername: EtcdPassword:********* EtcdKeyFile: EtcdCertFile: EtcdCACertFile: LeaderLeaseDuration:1500 LeaderRenewDeadline:1000 LeaderRetryPeriod:500 LogFile:/tmp/whereabouts.log LogLevel:debug OverlappingRanges:true SleepForRace:0 Gateway:<nil> Kubernetes: {KubeConfigPath:/etc/kubernetes/cni/net.d/whereabouts.d/whereabouts.kubeconfig K8sAPIRoot:} ConfigurationPath:PodName:pod-with-exclude-range PodNamespace:default} 2022-10-27T14:14:59Z [debug] Beginning IPAM for ContainerID: f4ffd0e07d6c1a2b6ffb0fa29910c795258792bb1a1710ff66f6b48fab37af82 2022-10-27T14:14:59Z [debug] Started leader election 2022-10-27T14:14:59Z [debug] OnStartedLeading() called 2022-10-27T14:14:59Z [debug] Elected as leader, do processing 2022-10-27T14:14:59Z [debug] IPManagement - mode: 0 / containerID:f4ffd0e07d6c1a2b6ffb0fa29910c795258792bb1a1710ff66f6b48fab37af82 / podRef: default/pod-with-exclude-range 2022-10-27T14:14:59Z [debug] IterateForAssignment input >> ip: fd43:1f1:3daa:baa:: | ipnet: {fd43:1f1:3daa:baa:: ffffffffffffffff0000000000000000} | first IP: fd43:1f1:3daa:baa::1 | last IP: fd43:1f1:3daa:baa:ffff:ffff:ffff:ffff
Actual results:
Failed to create pod sandbox: rpc error: code = DeadlineExceeded desc = context deadline exceeded
Expected results:
additional network gets attached to the pod
Additional info:
- is cloned by
-
OCPBUGS-3941 Whereabouts CNI timesout while iterating exclude range [backport 4.12]
- Closed
- is depended on by
-
OCPBUGS-3941 Whereabouts CNI timesout while iterating exclude range [backport 4.12]
- Closed
- links to