[EVPN]Fixing race condition when EVPN NVO add comes late (sonic-net#2756)

* Sometime during config reload, EVPN NVO table arrives later than remote VNI table entries. In such scenarios, remote vni entries are ignored and this leads to traffic loss. Added fix to retry instead of ignoring.

Author: dgsudharsan

orchagent/vxlanorch.cpp

@@ -2353,7 +2353,7 @@ bool EvpnRemoteVnip2pOrch::addOperation(const Request& request)
     {
         SWSS_LOG_WARN("Remote VNI add: Source VTEP not found. remote=%s vid=%d",
                       remote_vtep.c_str(), vlan_id);
-        return true;
+        return false;
     }
 
     VxlanTunnelOrch* tunnel_orch = gDirectory.get<VxlanTunnelOrch*>();
@@ -2522,7 +2522,7 @@ bool EvpnRemoteVnip2mpOrch::addOperation(const Request& request)
     {
         SWSS_LOG_WARN("Remote VNI add: Source VTEP not found. remote=%s vid=%d",
                       end_point_ip.c_str(),vlan_id);
-        return true;
+        return false;
     }
 
     if (!gPortsOrch->getVlanByVlanId(vlan_id, vlanPort))
@@ -2602,7 +2602,7 @@ bool EvpnRemoteVnip2mpOrch::delOperation(const Request& request)
     auto vtep_ptr = evpn_orch->getEVPNVtep();
     if (!vtep_ptr)
     {
-        SWSS_LOG_WARN("Remote VNI add: VTEP not found. remote=%s vid=%d",
+        SWSS_LOG_WARN("Remote VNI del: VTEP not found. remote=%s vid=%d",
                       end_point_ip.c_str(), vlan_id);
         return true;
     }

tests/test_evpn_tunnel.py

@@ -213,6 +213,51 @@ def test_delayed_vlan_vni_map(self, dvs, testlog):
         vxlan_obj.remove_vlan(dvs, "100")
         vxlan_obj.remove_vlan(dvs, "101")
 
+    def test_delayed_evpn_nvo(self, dvs, testlog):
+        vxlan_obj = self.get_vxlan_obj()
+
+        tunnel_name = 'tunnel_2'
+        map_name = 'map_1000_100'
+        map_name_1 = 'map_1001_101'
+        vlanlist = ['100']
+        vnilist = ['1000']
+
+        vxlan_obj.fetch_exist_entries(dvs)
+        vxlan_obj.create_vlan1(dvs,"Vlan100")
+        vxlan_obj.create_vlan1(dvs,"Vlan101")
+
+        vxlan_obj.create_vxlan_tunnel(dvs, tunnel_name, '6.6.6.6')
+        vxlan_obj.create_vxlan_tunnel_map(dvs, tunnel_name, map_name_1, '1001', 'Vlan101')
+        vxlan_obj.create_vxlan_tunnel_map(dvs, tunnel_name, map_name, '1000', 'Vlan100')
+
+        vxlan_obj.check_vxlan_sip_tunnel(dvs, tunnel_name, '6.6.6.6', vlanlist, vnilist, tunnel_map_entry_count = 2)
+        vxlan_obj.check_vxlan_tunnel_map_entry(dvs, tunnel_name, vlanlist, vnilist)
+
+
+        vxlan_obj.create_evpn_remote_vni(dvs, 'Vlan101', '7.7.7.7', '1001')
+        vxlan_obj.check_vxlan_dip_tunnel_not_created(dvs, tunnel_name, '6.6.6.6', '7.7.7.7')
+        vxlan_obj.create_evpn_nvo(dvs, 'nvo1', tunnel_name)
+
+        print("Testing VLAN 101 extension")
+        vxlan_obj.check_vxlan_dip_tunnel(dvs, tunnel_name, '6.6.6.6', '7.7.7.7')
+        vxlan_obj.check_vlan_extension(dvs, '101', '7.7.7.7')
+
+        print("Testing Vlan Extension removal")
+        vxlan_obj.remove_evpn_remote_vni(dvs, 'Vlan101', '7.7.7.7')
+        vxlan_obj.check_vlan_extension_delete(dvs, '101', '7.7.7.7')
+        vxlan_obj.check_vxlan_dip_tunnel_delete(dvs, '7.7.7.7')
+
+        vxlan_obj.remove_vxlan_tunnel_map(dvs, tunnel_name, map_name, '1000', 'Vlan100')
+        vxlan_obj.remove_vxlan_tunnel_map(dvs, tunnel_name, map_name_1, '1001', 'Vlan101')
+        vxlan_obj.check_vxlan_tunnel_map_entry_delete(dvs, tunnel_name, vlanlist, vnilist)
+
+        print("Testing SIP Tunnel Deletion")
+        vxlan_obj.remove_evpn_nvo(dvs, 'nvo1')
+        vxlan_obj.remove_vxlan_tunnel(dvs, tunnel_name)
+        vxlan_obj.check_vxlan_sip_tunnel_delete(dvs, tunnel_name, '6.6.6.6')
+        vxlan_obj.remove_vlan(dvs, "100")
+        vxlan_obj.remove_vlan(dvs, "101")
+
     def test_invalid_vlan_extension(self, dvs, testlog):
         vxlan_obj = self.get_vxlan_obj()
 

tests/test_evpn_tunnel_p2mp.py

@@ -172,6 +172,49 @@ def test_delayed_vlan_vni_map(self, dvs, testlog):
         vxlan_obj.remove_vlan(dvs, "100")
         vxlan_obj.remove_vlan(dvs, "101")
 
+    def test_delayed_evpn_nvo(self, dvs, testlog):
+        vxlan_obj = self.get_vxlan_obj()
+
+        tunnel_name = 'tunnel_2'
+        map_name = 'map_1000_100'
+        map_name_1 = 'map_1001_101'
+        vlanlist = ['100']
+        vnilist = ['1000']
+
+        vxlan_obj.fetch_exist_entries(dvs)
+        vxlan_obj.create_vlan1(dvs,"Vlan100")
+        vxlan_obj.create_vlan1(dvs,"Vlan101")
+
+        vxlan_obj.create_vxlan_tunnel(dvs, tunnel_name, '6.6.6.6')
+        vxlan_obj.create_vxlan_tunnel_map(dvs, tunnel_name, map_name, '1000', 'Vlan100')
+        vxlan_obj.create_vxlan_tunnel_map(dvs, tunnel_name, map_name_1, '1001', 'Vlan101')
+        vxlan_obj.create_evpn_remote_vni(dvs, 'Vlan101', '7.7.7.7', '1001')
+
+        vxlan_obj.check_vxlan_sip_tunnel(dvs, tunnel_name, '6.6.6.6', vlanlist, vnilist, ignore_bp=False, tunnel_map_entry_count = 2)
+        vxlan_obj.check_vxlan_tunnel_map_entry(dvs, tunnel_name, vlanlist, vnilist)
+
+
+        vxlan_obj.check_vlan_extension_not_created_p2mp(dvs, '101', '6.6.6.6', '7.7.7.7')
+        vxlan_obj.create_evpn_nvo(dvs, 'nvo1', tunnel_name)
+
+        print("Testing VLAN 101 extension")
+        vxlan_obj.check_vlan_extension_p2mp(dvs, '101', '6.6.6.6', '7.7.7.7')
+
+        print("Testing Vlan Extension removal")
+        vxlan_obj.remove_evpn_remote_vni(dvs, 'Vlan101', '7.7.7.7')
+        vxlan_obj.check_vlan_extension_delete_p2mp(dvs, '101', '6.6.6.6', '7.7.7.7')
+
+        vxlan_obj.remove_vxlan_tunnel_map(dvs, tunnel_name, map_name, '1000', 'Vlan100')
+        vxlan_obj.remove_vxlan_tunnel_map(dvs, tunnel_name, map_name_1, '1001', 'Vlan101')
+        vxlan_obj.check_vxlan_tunnel_map_entry_delete(dvs, tunnel_name, vlanlist, vnilist)
+
+        print("Testing SIP Tunnel Deletion")
+        vxlan_obj.remove_evpn_nvo(dvs, 'nvo1')
+        vxlan_obj.remove_vxlan_tunnel(dvs, tunnel_name)
+        vxlan_obj.check_vxlan_sip_tunnel_delete(dvs, tunnel_name, '6.6.6.6', ignore_bp=False)
+        vxlan_obj.remove_vlan(dvs, "100")
+        vxlan_obj.remove_vlan(dvs, "101")
+
     def test_invalid_vlan_extension(self, dvs, testlog):
         vxlan_obj = self.get_vxlan_obj()