Avoid openvswitch restarts during re-run

For offload configuration on the first time, opevswitch
need to be restarted after the interface configured to
switchdev mode to enable VF-LAG. But once it is configured,
openvswitch should not be restarted, which will create
VF-LAG failures. In general, in the TripleO during
updates/upgrades, ovs restart will not be triggered
and it will be allowed to the user to restart or reboot
the node. Aligning with it, opevswitch restart has been
avoided on all os-net-config re-runs.
Closes-Bug: #1887957

Change-Id: I0a3b374b321ba039c8e0b1c126ac3eb59e6dc980
(cherry picked from commit 4c95f7d087)

os_net_config/cli.py

@@ -320,9 +320,15 @@ def main(argv=sys.argv):
         pf_files_changed = provider.apply(cleanup=opts.cleanup,
                                           activate=not opts.no_activate)
         if not opts.noop:
+            restart_ovs = bool(sriovpf_member_of_bond_ovs_port_list)
+            # Avoid ovs restart for os-net-config re-runs, which will
+            # dirupt the offload configuration
+            if os.path.exists(utils._SRIOV_CONFIG_SERVICE_FILE):
+                restart_ovs = False
+
             utils.configure_sriov_pfs(
                 execution_from_cli=True,
-                restart_openvswitch=bool(sriovpf_member_of_bond_ovs_port_list))
+                restart_openvswitch=restart_ovs)
 
     for iface_json in iface_array:
         # All objects other than the sriov_pf will be added here.