Created attachment 1962335 [details] DPDK checkup manifests Description of problem: When running a DPDK checkup job, attempting to run the VM and the traffic generator on specific nodes is ignored. Version-Release number of selected component (if applicable): CNV 4.13.0 container-native-virtualization-kubevirt-dpdk-checkup-rhel9:v4.13.0-37 How reproducible: 100% Steps to Reproduce: 1. Create namespace for the job, and change context to the new namespace. $ oc create ns dpdk-checkup-ns $ oc project dpdk-checkup-ns 2. Label the worker nodes with "worker-dpdk" label. 3. Apply the resources manifests in the attached file in their numeric order: $ oc apply -f 1-dpdk-checkup-resources.yaml $ oc apply -f 2-dpdk-checkup-scc.yaml ... change the resources according to your cluster. In the ConfigMap manifest - note these 2 params: spec.param.trafficGeneratorNodeSelector: "cnv-qe-infra-12.cnvqe2.lab.eng.rdu2.redhat.com" spec.param.DPDKLabelSelector: "cnv-qe-infra-12.cnvqe2.lab.eng.rdu2.redhat.com" 4. After applying the job itself (8-dpdk-checkup-job.yaml) - follow the pods that are created and the nodes they are created on (or wait for the job to finish, as this info will also appear in the result ConfigMap). $ oc get cm dpdk-checkup-config -o yaml | grep "status.result" | grep Node status.result.DPDKVMNode: cnv-qe-infra-13.cnvqe2.lab.eng.rdu2.redhat.com status.result.trafficGeneratorNode: cnv-qe-infra-13.cnvqe2.lab.eng.rdu2.redhat.com <BUG> Although the setup ConfigMap is attempting to set cnv-qe-infra-12.cnvqe2.lab.eng.rdu2.redhat.com as the node for both the traffic generator and the VM, the actual used node is cnv-qe-infra-13.cnvqe2.lab.eng.rdu2.redhat.com. Additional info: Checking the log of the checkup job pod shows that these fields remain blank and not taken from the ConfigMap: 2023/05/04 14:05:11 "trafficGeneratorNodeLabelSelector": "" 2023/05/04 14:05:11 "trafficGeneratorPacketsPerSecond": "8m" 2023/05/04 14:05:11 "DPDKNodeLabelSelector": ""
*** Bug 2193235 has been marked as a duplicate of this bug. ***
Verified by running the same scenario as in the bug description. CNV 4.14.0 container-native-virtualization/kubevirt-dpdk-checkup-rhel9:v4.14.0-116 checking the ConfigMap after the job is done shows that the pods were scheduled on the target node I set: $ oc get cm dpdk-checkup-config -o yaml apiVersion: v1 data: ... spec.param.trafficGenTargetNodeName: cnv-qe-19.cnvqe.lab.eng.rdu2.redhat.com ... spec.param.vmUnderTestTargetNodeName: cnv-qe-19.cnvqe.lab.eng.rdu2.redhat.com ... status.result.trafficGenActualNodeName: cnv-qe-19.cnvqe.lab.eng.rdu2.redhat.com ... status.result.vmUnderTestActualNodeName: cnv-qe-19.cnvqe.lab.eng.rdu2.redhat.com ... Following the pods during the job run shows they are scheduled on the selected node: $ oc get pods -o wide -w NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES virt-launcher-dpdk-traffic-gen-6mtfv-vl955 2/2 Running 0 78s 10.130.0.124 cnv-qe-19.cnvqe.lab.eng.rdu2.redhat.com <none> 1/1 virt-launcher-vmi-under-test-6mtfv-gz8b2 2/2 Running 0 79s 10.130.0.121 cnv-qe-19.cnvqe.lab.eng.rdu2.redhat.com <none> 1/1
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Important: OpenShift Virtualization 4.14.0 Images security and bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2023:6817