Created attachment 1747088 [details]
State to set
Description of problem:
At a OCP 4.7 cluster using nmstate 0.3.4 to create a linux-bridge in top of primary nic "ens3" and rolling it back fails at rolling it back and make the node connectivity being lost.
The OCP cluster is using OVN Kubernetes and the ens3 is already a slave under a ovs bridge "br-ext", issue is that rollback is not able to recover from the issue.
Version-Release number of selected component (if applicable):
ssh into a OCP 4.7 worker and set a password for root since we need to access to it using libvirt console or openstack console.
Then access to the worker using libvirt/openstack commands since connectivity is going to be lost ssh connection is broken.
Steps to Reproduce:
1. ping www.redhat.com
2. ./nmstatectl.sh set --no-commit --timeout 120 /quique/state.yaml
3. ping www.redhat.com (Connectivity is lost)
4. ./nmstatectl.sh rollback
5. ping www.redhat.com (Connectivity still lost)
nmstate is not able to rollback this configuration
nmstate restoring network configuration so we don't lose connectivity.
Created attachment 1747089 [details]
NetworkManager debug logs
Created attachment 1747090 [details]
show state before set
Created attachment 1747091 [details]
show state after set
Created attachment 1747092 [details]
show state after rollback
Gris, would you please prioritize it? Would it make it to 1.0?
If we are expecting fix in nmstate 1.0, could you try it on nmstate-1.0(RHEL 8.4)?
In my VM, it seems works well.
(In reply to Gris Ge from comment #6)
> Hi Quique,
> If we are expecting fix in nmstate 1.0, could you try it on nmstate-1.0(RHEL
> In my VM, it seems works well.
Please, let us know if this works on nmstate 1.0 or greater. It works well for me. Thanks!
@ferferna now it's easier with the nmstate/automation we have for k8s I will give it a try.
I have test it with CNV 4.8 and at a OVNKubernetes openshift cluster and now the rollback works fine creating a linux bridge with one of the ports matcing the br-ex primary nic port.
root@zeus12 quique]# oc exec -n openshift-cnv nmstate-handler-jrpdd -- rpm -qa |grep nmstate
(In reply to Quique Llorente from comment #9)
> I have test it with CNV 4.8 and at a OVNKubernetes openshift cluster and now
> the rollback works fine creating a linux bridge with one of the ports
> matcing the br-ex primary nic port.
> root@zeus12 quique]# oc exec -n openshift-cnv nmstate-handler-jrpdd -- rpm
> -qa |grep nmstate
If it works, what are your expectations to happen here?
Reporter(Quique Llorente) agree on bug closure.
But let's keep this bug for QE or devel effort to create auto test case for this.
As it works on nmstate-1.0 and above, mark as verified
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 (nmstate bug fix and enhancement 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.