Description of problem: OCP 4.7, and 4.9 Trying to disable the use of SLAAC within a container consuming an SRIOV vf using the net.ipv6.conf.all.accept_ra": 0 does not appear to disable it on the secondary interface in the pod we set the following sysctl net.ipv6.conf.all.accept_ra = 0 and looking at the secondary interface (nabeel1) see: net.ipv6.conf.nabeel1.accept_ra = 1 We are still working to gather some more information and data but we were alerted to the following and not sure if that is what is affecting us: https://github.com/containernetworking/plugins/pull/669 Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
This seems like an OCP issue to me, assigning to Peter to take a look . @dornelas if you have thoughts on other landing spots, please let me know.
Looks like a fix to the container networking plugins. Assigning to me.
Looks like the advice is to use the `.IFNAME.` syntax as provided in the verification example below. In order to accomplish this, it requires a 4.9.z backport from upstream. Backport PR @ https://github.com/openshift/containernetworking-plugins/pull/64 Verification procedure follows ----------------------------- For verification, use this YAML: ``` --- apiVersion: "k8s.cni.cncf.io/v1" kind: NetworkAttachmentDefinition metadata: name: example-tuning-conf spec: config: > { "cniVersion": "0.4.0", "name": "example-tuning-conf", "plugins": [{ "type": "bridge", "bridge": "test1234", "ipam": { "type": "host-local", "subnet": "192.168.1.0/24", "rangeStart": "192.168.1.200", "rangeEnd": "192.168.1.216", "routes": [ { "dst": "0.0.0.0/0" } ], "gateway": "192.168.1.1" } }, { "type": "tuning", "sysctl": { "net.ipv6.conf.IFNAME.accept_ra": "0" } } ] } --- apiVersion: v1 kind: Pod metadata: name: samplepod annotations: k8s.v1.cni.cncf.io/networks: example-tuning-conf spec: containers: - name: samplepod command: ["/bin/ash", "-c", "trap : TERM INT; sleep infinity & wait"] image: alpine ``` The resulting pod should have accept_ra set to zero, such as: ``` $ oc exec -it samplepod -- cat /proc/sys/net/ipv6/conf/net1/accept_ra 0 ```
Tested and verified in 4.9.0-0.nightly-2022-07-21-221241 [weliang@weliang Bug]$ oc get pod NAME READY STATUS RESTARTS AGE samplepod 1/1 Running 0 24s [weliang@weliang Bug]$ oc exec -it samplepod -- cat /proc/sys/net/ipv6/conf/net1/accept_ra 0 [weliang@weliang Bug]$ oc get clusterversion NAME VERSION AVAILABLE PROGRESSING SINCE STATUS version 4.9.0-0.nightly-2022-07-21-221241 True False 55m Cluster version is 4.9.0-0.nightly-2022-07-21-221241 [weliang@weliang Bug]$
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 Container Platform 4.9.45 bug fix and security 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-2022:5879