ovn: Don't fail db sync if new IP allocation fails for metadata

There is a scenario where IP allocation pool is depleted but OVN
metadata port got removed its IP manually. The DB sync script will
attempt to allocate a new IP address if DHCP is enabled in the subnet.
Since the pool has no available IP addresses an exception is raised and
the whole db sync stops.

This patch simply catches the exception, logs and error and continues
syncing other resources.

Closes-bug: #1987135
Signed-off-by: Jakub Libosvar <[email protected]>
Change-Id: Iaa7b0d7ceb244a38fddd7676066683bf2ca72341
(cherry picked from commit d237a2a)

Author: cubeek

neutron/plugins/ml2/drivers/ovn/mech_driver/ovsdb/ovn_db_sync.py

@@ -948,8 +948,13 @@ def _sync_metadata_ports(self, ctx, db_ports):
                     db_ports.pop(port['id'])
 
             if self.mode == SYNC_MODE_REPAIR:
-                # Make sure that this port has an IP address in all the subnets
-                self._ovn_client.update_metadata_port(ctx, net['id'])
+                try:
+                    # Make sure that this port has an IP address in all the
+                    # subnets
+                    self._ovn_client.update_metadata_port(ctx, net['id'])
+                except n_exc.IpAddressGenerationFailure:
+                    LOG.error('Could not allocate IP addresses for '
+                              'metadata port in network %s', net['id'])
         LOG.debug('OVN sync metadata ports finished')
 
     def sync_networks_ports_and_dhcp_opts(self, ctx):