diff --git a/bgpd/bgp_zebra.c b/bgpd/bgp_zebra.c
index a1312b97b6..11e872f6b4 100644
--- a/bgpd/bgp_zebra.c
+++ b/bgpd/bgp_zebra.c
@@ -1144,7 +1144,8 @@ static bool update_ipv6nh_for_route_install(int nh_othervrf, struct bgp *nh_bgp,
 			api_nh->ifindex = 0;
 		}
 	}
-	api_nh->gate.ipv6 = *nexthop;
+	if (nexthop)
+		api_nh->gate.ipv6 = *nexthop;
 
 	return true;
 }
diff --git a/doc/user/bgp.rst b/doc/user/bgp.rst
index 427349c037..9983911581 100644
--- a/doc/user/bgp.rst
+++ b/doc/user/bgp.rst
@@ -2540,6 +2540,26 @@ the same behavior of using same next-hop and RMAC values.
 Enables or disables advertise-pip feature, specifiy system-IP and/or system-MAC
 parameters.
 
++Support with VRF network namespace backend
++^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+It is possible to separate overlay networks contained in VXLAN interfaces from
+underlay networks by using VRFs. VRF-lite and VRF-netns backends can be used for
+that. In the latter case, it is necessary to set both bridge and vxlan interface
+in the same network namespace, as below example illustrates:
+
+.. code-block:: shell
+
+   # linux shell
+   ip netns add vrf1
+   ip link add name vxlan101 type vxlan id 101 dstport 4789 dev eth0 local 10.1.1.1
+   ip link set dev vxlan101 netns vrf1
+   ip netns exec vrf1 ip link set dev lo up
+   ip netns exec vrf1 brctl addbr bridge101
+   ip netns exec vrf1 brctl addif bridge101 vxlan101
+
+This makes it possible to separate not only layer 3 networks like VRF-lite networks.
+Also, VRF netns based make possible to separate layer 2 networks on separate VRF
+instances.
 
 .. _bgp-debugging:
 
diff --git a/include/linux/net_namespace.h b/include/linux/net_namespace.h
index 0187c74d88..0ed9dd61d3 100644
--- a/include/linux/net_namespace.h
+++ b/include/linux/net_namespace.h
@@ -16,6 +16,7 @@ enum {
 	NETNSA_NSID,
 	NETNSA_PID,
 	NETNSA_FD,
+	NETNSA_TARGET_NSID,
 	__NETNSA_MAX,
 };
 
diff --git a/lib/netns_linux.c b/lib/netns_linux.c
index 98f359401e..e8d549b4e0 100644
--- a/lib/netns_linux.c
+++ b/lib/netns_linux.c
@@ -379,12 +379,20 @@ struct ns *ns_lookup(ns_id_t ns_id)
 	return ns_lookup_internal(ns_id);
 }
 
-void ns_walk_func(int (*func)(struct ns *))
+void ns_walk_func(int (*func)(struct ns *,
+			      void *param_in,
+			      void **param_out),
+		  void *param_in,
+		  void **param_out)
 {
 	struct ns *ns = NULL;
+	int ret;
 
-	RB_FOREACH (ns, ns_head, &ns_tree)
-		func(ns);
+	RB_FOREACH (ns, ns_head, &ns_tree) {
+		ret = func(ns, param_in, param_out);
+		if (ret == NS_WALK_STOP)
+			return;
+	}
 }
 
 const char *ns_get_name(struct ns *ns)
@@ -584,9 +592,35 @@ int ns_socket(int domain, int type, int protocol, ns_id_t ns_id)
 	return ret;
 }
 
+/* if relative link_nsid matches default netns,
+ * then return default absolute netns value
+ * otherwise, return NS_UNKNOWN
+ */
+ns_id_t ns_id_get_absolute(ns_id_t ns_id_reference, ns_id_t link_nsid)
+{
+	struct ns *ns;
+
+	ns = ns_lookup(ns_id_reference);
+	if (!ns)
+		return NS_UNKNOWN;
+
+	if (ns->relative_default_ns != link_nsid)
+		return NS_UNKNOWN;
+
+	ns = ns_get_default();
+	assert(ns);
+	return ns->ns_id;
+}
+
 ns_id_t ns_get_default_id(void)
 {
 	if (default_ns)
 		return default_ns->ns_id;
 	return NS_DEFAULT_INTERNAL;
 }
+
+struct ns *ns_get_default(void)
+{
+	return default_ns;
+}
+
diff --git a/lib/ns.h b/lib/ns.h
index 20e0a38e3b..286ff5b295 100644
--- a/lib/ns.h
+++ b/lib/ns.h
@@ -53,6 +53,11 @@ struct ns {
 	/* Identifier, mapped on the NSID value */
 	ns_id_t internal_ns_id;
 
+	/* Identifier, value of NSID of default netns,
+	 * relative value in that local netns
+	 */
+	ns_id_t relative_default_ns;
+
 	/* Name */
 	char *name;
 
@@ -120,7 +125,14 @@ int ns_socket(int domain, int type, int protocol, ns_id_t ns_id);
 extern char *ns_netns_pathname(struct vty *vty, const char *name);
 
 /* Parse and execute a function on all the NETNS */
-extern void ns_walk_func(int (*func)(struct ns *));
+#define NS_WALK_CONTINUE 0
+#define NS_WALK_STOP 1
+
+extern void ns_walk_func(int (*func)(struct ns *,
+				     void *,
+				     void **),
+			 void *param_in,
+			 void **param_out);
 
 /* API to get the NETNS name, from the ns pointer */
 extern const char *ns_get_name(struct ns *ns);
@@ -174,7 +186,9 @@ extern struct ns *ns_lookup_name(const char *name);
  */
 extern int ns_enable(struct ns *ns, void (*func)(ns_id_t, void *));
 extern struct ns *ns_get_created(struct ns *ns, char *name, ns_id_t ns_id);
+extern ns_id_t ns_id_get_absolute(ns_id_t ns_id_reference, ns_id_t link_nsid);
 extern void ns_disable(struct ns *ns);
+extern struct ns *ns_get_default(void);
 
 #ifdef __cplusplus
 }
diff --git a/lib/vrf.c b/lib/vrf.c
index 6e4c983c66..cc7445558c 100644
--- a/lib/vrf.c
+++ b/lib/vrf.c
@@ -659,7 +659,8 @@ int vrf_handler_create(struct vty *vty, const char *vrfname,
 }
 
 int vrf_netns_handler_create(struct vty *vty, struct vrf *vrf, char *pathname,
-			     ns_id_t ns_id, ns_id_t internal_ns_id)
+			     ns_id_t ns_id, ns_id_t internal_ns_id,
+			     ns_id_t rel_def_ns_id)
 {
 	struct ns *ns = NULL;
 
@@ -706,6 +707,7 @@ int vrf_netns_handler_create(struct vty *vty, struct vrf *vrf, char *pathname,
 	}
 	ns = ns_get_created(ns, pathname, ns_id);
 	ns->internal_ns_id = internal_ns_id;
+	ns->relative_default_ns = rel_def_ns_id;
 	ns->vrf_ctxt = (void *)vrf;
 	vrf->ns_ctxt = (void *)ns;
 	/* update VRF netns NAME */
@@ -801,7 +803,9 @@ DEFUN_NOSH (vrf_netns,
 
 	frr_with_privs(vrf_daemon_privs) {
 		ret = vrf_netns_handler_create(vty, vrf, pathname,
-					       NS_UNKNOWN, NS_UNKNOWN);
+					       NS_UNKNOWN,
+					       NS_UNKNOWN,
+					       NS_UNKNOWN);
 	}
 	return ret;
 }
diff --git a/lib/vrf.h b/lib/vrf.h
index 83ed16b48e..a8514d74ed 100644
--- a/lib/vrf.h
+++ b/lib/vrf.h
@@ -315,7 +315,7 @@ extern int vrf_handler_create(struct vty *vty, const char *name,
  */
 extern int vrf_netns_handler_create(struct vty *vty, struct vrf *vrf,
 				    char *pathname, ns_id_t ext_ns_id,
-				    ns_id_t ns_id);
+				    ns_id_t ns_id, ns_id_t rel_def_ns_id);
 
 /* used internally to enable or disable VRF.
  * Notify a change in the VRF ID of the VRF
diff --git a/tests/topotests/bgp_evpn_rt5/__init__.py b/tests/topotests/bgp_evpn_rt5/__init__.py
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/tests/topotests/bgp_evpn_rt5/r1/bgpd.conf b/tests/topotests/bgp_evpn_rt5/r1/bgpd.conf
new file mode 100644
index 0000000000..9237682067
--- /dev/null
+++ b/tests/topotests/bgp_evpn_rt5/r1/bgpd.conf
@@ -0,0 +1,26 @@
+debug bgp neighbor-events
+debug bgp updates
+debug bgp zebra
+router bgp 65000
+ bgp router-id 192.168.100.21
+ bgp log-neighbor-changes
+ no bgp default ipv4-unicast
+ neighbor 192.168.100.41 remote-as 65000
+ neighbor 192.168.100.41 capability extended-nexthop
+ !
+ address-family l2vpn evpn
+  neighbor 192.168.100.41 activate
+  advertise-all-vni
+ exit-address-family
+!
+router bgp 65000 vrf r1-vrf-101
+ bgp router-id 192.168.102.21
+ bgp log-neighbor-changes
+ no bgp network import-check
+ address-family ipv4 unicast
+  network 192.168.102.21/32
+ exit-address-family
+ address-family l2vpn evpn
+  advertise ipv4 unicast
+ exit-address-family
+ !
diff --git a/tests/topotests/bgp_evpn_rt5/r1/zebra.conf b/tests/topotests/bgp_evpn_rt5/r1/zebra.conf
new file mode 100644
index 0000000000..f5eaab1953
--- /dev/null
+++ b/tests/topotests/bgp_evpn_rt5/r1/zebra.conf
@@ -0,0 +1,22 @@
+log stdout
+
+hostname r1
+password zebra
+
+debug zebra vxlan
+debug zebra kernel
+debug zebra dplane
+debug zebra rib
+log stdout
+vrf r1-vrf-101
+ vni 101
+ exit-vrf
+!
+interface r1-eth0
+ ip address 192.168.100.21/24
+!
+interface loop101 vrf r1-vrf-101
+ ip address 192.168.102.21/32
+!
+
+
diff --git a/tests/topotests/bgp_evpn_rt5/r2/bgpd.conf b/tests/topotests/bgp_evpn_rt5/r2/bgpd.conf
new file mode 100644
index 0000000000..6dcacd288d
--- /dev/null
+++ b/tests/topotests/bgp_evpn_rt5/r2/bgpd.conf
@@ -0,0 +1,27 @@
+debug bgp neighbor-events
+debug bgp updates
+debug bgp zebra
+router bgp 65000
+ bgp router-id 192.168.100.41
+ bgp log-neighbor-changes
+ no bgp default ipv4-unicast
+ neighbor 192.168.100.21 peer-group
+ neighbor 192.168.100.21 remote-as 65000
+ neighbor 192.168.100.21 capability extended-nexthop
+ !
+ address-family l2vpn evpn
+  neighbor 192.168.100.21 activate
+  advertise-all-vni
+ exit-address-family
+!
+router bgp 65000 vrf r2-vrf-101
+ bgp router-id 192.168.101.41
+ bgp log-neighbor-changes
+ no bgp network import-check
+ address-family ipv4 unicast
+  network 192.168.101.41/32
+ exit-address-family
+ address-family l2vpn evpn
+  advertise ipv4 unicast
+ exit-address-family
+ !
diff --git a/tests/topotests/bgp_evpn_rt5/r2/zebra.conf b/tests/topotests/bgp_evpn_rt5/r2/zebra.conf
new file mode 100644
index 0000000000..e5f962d254
--- /dev/null
+++ b/tests/topotests/bgp_evpn_rt5/r2/zebra.conf
@@ -0,0 +1,18 @@
+log stdout
+
+hostname r2
+password zebra
+
+debug zebra vxlan
+
+vrf r2-vrf-101
+ vni 101
+ exit-vrf
+!
+interface loop101 vrf r2-vrf-101
+ ip address 192.168.101.41/32
+!
+interface r2-eth0
+ ip address 192.168.100.41/24
+!
+
diff --git a/tests/topotests/bgp_evpn_rt5/test_bgp_evpn.py b/tests/topotests/bgp_evpn_rt5/test_bgp_evpn.py
new file mode 100755
index 0000000000..69ef7e9fae
--- /dev/null
+++ b/tests/topotests/bgp_evpn_rt5/test_bgp_evpn.py
@@ -0,0 +1,234 @@
+#!/usr/bin/env python
+
+#
+# test_bgp_evpn.py
+# Part of NetDEF Topology Tests
+#
+# Copyright (c) 2019 by 6WIND
+#
+# Permission to use, copy, modify, and/or distribute this software
+# for any purpose with or without fee is hereby granted, provided
+# that the above copyright notice and this permission notice appear
+# in all copies.
+#
+# THE SOFTWARE IS PROVIDED "AS IS" AND NETDEF DISCLAIMS ALL WARRANTIES
+# WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+# MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL NETDEF BE LIABLE FOR
+# ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY
+# DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS,
+# WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS
+# ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
+# OF THIS SOFTWARE.
+#
+
+"""
+ test_bgp_evpn.py: Test the FRR/Quagga BGP daemon with BGP IPv6 interface
+ with route advertisements on a separate netns.
+"""
+
+import os
+import sys
+import json
+from functools import partial
+import pytest
+import platform
+
+# Save the Current Working Directory to find configuration files.
+CWD = os.path.dirname(os.path.realpath(__file__))
+sys.path.append(os.path.join(CWD, '../'))
+
+# pylint: disable=C0413
+# Import topogen and topotest helpers
+from lib import topotest
+from lib.topogen import Topogen, TopoRouter, get_topogen
+from lib.topolog import logger
+
+# Required to instantiate the topology builder class.
+from mininet.topo import Topo
+
+l3mdev_accept = 0
+krel = ''
+
+class BGPEVPNTopo(Topo):
+    "Test topology builder"
+    def build(self, *_args, **_opts):
+        "Build function"
+        tgen = get_topogen(self)
+
+        tgen.add_router('r1')
+        tgen.add_router('r2')
+
+        switch = tgen.add_switch('s1')
+        switch.add_link(tgen.gears['r1'])
+        switch.add_link(tgen.gears['r2'])
+
+        switch = tgen.add_switch('s2')
+        switch.add_link(tgen.gears['r1'])
+
+        switch = tgen.add_switch('s3')
+        switch.add_link(tgen.gears['r2'])
+        
+def setup_module(mod):
+    "Sets up the pytest environment"
+    global l3mdev_accept
+    global krel
+
+    tgen = Topogen(BGPEVPNTopo, mod.__name__)
+    tgen.start_topology()
+
+    router_list = tgen.routers()
+
+    krel = platform.release()
+    if topotest.version_cmp(krel, '4.18') < 0:
+        logger.info('BGP EVPN RT5 NETNS tests will not run (have kernel "{}", but it requires 4.18)'.format(krel))
+        return pytest.skip('Skipping BGP EVPN RT5 NETNS Test. Kernel not supported')
+
+    l3mdev_accept = 1
+    logger.info('setting net.ipv4.tcp_l3mdev_accept={}'.format(l3mdev_accept))
+
+    # create VRF vrf-101 on R1 and R2
+    # create loop101
+    cmds_vrflite = ['sysctl -w net.ipv4.tcp_l3mdev_accept={}'.format(l3mdev_accept),
+                    'ip link add {}-vrf-101 type vrf table 101',
+                    'ip ru add oif {}-vrf-101 table 101',
+                    'ip ru add iif {}-vrf-101 table 101',
+                    'ip link set dev {}-vrf-101 up',
+                    'sysctl -w net.ipv4.tcp_l3mdev_accept={}'.format(l3mdev_accept),
+                    'ip link add loop101 type dummy',
+                    'ip link set dev loop101 master {}-vrf-101',
+                    'ip link set dev loop101 up']
+    cmds_netns = ['ip netns add {}-vrf-101',
+                  'ip link add loop101 type dummy',
+                  'ip link set dev loop101 netns {}-vrf-101',
+                  'ip netns exec {}-vrf-101 ip link set dev loop101 up']
+
+    cmds_r2 = [ # config routing 101
+               'ip link add name bridge-101 up type bridge stp_state 0',
+               'ip link set bridge-101 master {}-vrf-101',
+               'ip link set dev bridge-101 up',
+               'ip link add name vxlan-101 type vxlan id 101 dstport 4789 dev r2-eth0 local 192.168.100.41',
+               'ip link set dev vxlan-101 master bridge-101',
+               'ip link set vxlan-101 up type bridge_slave learning off flood off mcast_flood off']
+
+    cmds_r1_netns_method3 = ['ip link add name vxlan-{1} type vxlan id {1} dstport 4789 dev {0}-eth0 local 192.168.100.21',
+                             'ip link set dev vxlan-{1} netns {0}-vrf-{1}',
+                             'ip netns exec {0}-vrf-{1} ip li set dev lo up',
+                             'ip netns exec {0}-vrf-{1} ip link add name bridge-{1} up type bridge stp_state 0',
+                             'ip netns exec {0}-vrf-{1} ip link set dev vxlan-{1} master bridge-{1}',
+                             'ip netns exec {0}-vrf-{1} ip link set bridge-{1} up',
+                             'ip netns exec {0}-vrf-{1} ip link set vxlan-{1} up']
+
+    router = tgen.gears['r1']
+    for cmd in cmds_netns:
+        logger.info('cmd to r1: '+cmd);
+        output = router.run(cmd.format('r1'))
+        logger.info('result: '+output);
+
+    router = tgen.gears['r2']
+    for cmd in cmds_vrflite:
+        logger.info('cmd to r2: '+cmd.format('r2'));
+        output = router.run(cmd.format('r2'))
+        logger.info('result: '+output);
+
+    for cmd in cmds_r2:
+        logger.info('cmd to r2: '+cmd.format('r2'));
+        output = router.run(cmd.format('r2'))
+        logger.info('result: '+output);
+
+    router = tgen.gears['r1']
+    bridge_id = '101'
+    for cmd in cmds_r1_netns_method3:
+        logger.info('cmd to r1: '+cmd.format('r1', bridge_id));
+        output = router.run(cmd.format('r1', bridge_id))
+        logger.info('result: '+output);
+    router = tgen.gears['r1']
+
+    for rname, router in router_list.iteritems():
+        if rname == 'r1':
+            router.load_config(
+                TopoRouter.RD_ZEBRA,
+                os.path.join(CWD, '{}/zebra.conf'.format(rname)),
+                '--vrfwnetns -o vrf0'
+            )
+        else:
+            router.load_config(
+                TopoRouter.RD_ZEBRA,
+                os.path.join(CWD, '{}/zebra.conf'.format(rname))
+            )
+        router.load_config(
+            TopoRouter.RD_BGP,
+            os.path.join(CWD, '{}/bgpd.conf'.format(rname))
+        )
+
+    # Initialize all routers.
+    tgen.start_router()
+
+def teardown_module(_mod):
+    "Teardown the pytest environment"
+    tgen = get_topogen()
+    cmds_rx_netns = ['ip netns del {}-vrf-101']
+    
+    router = tgen.gears['r1']
+    for cmd in cmds_rx_netns:
+        logger.info('cmd to r1: '+cmd.format('r1'));
+        output = router.run(cmd.format('r1'))
+    tgen.stop_topology()
+
+
+def test_protocols_convergence():
+    """
+    Assert that all protocols have converged
+    statuses as they depend on it.
+    """
+    tgen = get_topogen()
+    if tgen.routers_have_failure():
+        pytest.skip(tgen.errors)
+    topotest.sleep(4, 'waiting 4 seconds for bgp convergence')
+    # Check IPv4/IPv6 routing tables.
+    output = tgen.gears['r1'].vtysh_cmd('show bgp l2vpn evpn', isjson=False)
+    logger.info('==== result from show bgp l2vpn evpn')
+    logger.info(output)
+    output = tgen.gears['r1'].vtysh_cmd('show bgp l2vpn evpn route detail', isjson=False)
+    logger.info('==== result from show bgp l2vpn evpn route detail')
+    logger.info(output)
+    output = tgen.gears['r1'].vtysh_cmd('show bgp vrf r1-vrf-101 ipv4', isjson=False)
+    logger.info('==== result from show bgp vrf r1-vrf-101 ipv4')
+    logger.info(output)
+    output = tgen.gears['r1'].vtysh_cmd('show bgp vrf r1-vrf-101', isjson=False)
+    logger.info('==== result from show bgp vrf r1-vrf-101 ')
+    logger.info(output)
+    output = tgen.gears['r1'].vtysh_cmd('show ip route vrf r1-vrf-101', isjson=False)
+    logger.info('==== result from show ip route vrf r1-vrf-101')
+    logger.info(output)
+    output = tgen.gears['r1'].vtysh_cmd('show evpn vni detail', isjson=False)
+    logger.info('==== result from show evpn vni detail')
+    logger.info(output)
+    output = tgen.gears['r1'].vtysh_cmd('show evpn next-hops vni all', isjson=False)
+    logger.info('==== result from show evpn next-hops vni all')
+    logger.info(output)
+    output = tgen.gears['r1'].vtysh_cmd('show evpn rmac vni all', isjson=False)
+    logger.info('==== result from show evpn next-hops vni all')
+    logger.info(output)
+    # Check IPv4 and IPv6 connectivity between r1 and r2 ( routing vxlan evpn)
+    pingrouter = tgen.gears['r1']
+    logger.info('Check Ping IPv4 from  R1(r1-vrf-101) to R2(r2-vrf-101 = 192.168.101.41)')
+    output = pingrouter.run('ip netns exec r1-vrf-101 ping 192.168.101.41 -f -c 1000')
+    logger.info(output)
+    if '1000 packets transmitted, 1000 received' not in output:
+        assertmsg = 'expected ping IPv4 from R1(r1-vrf-101) to R2(192.168.101.41) should be ok'
+        assert 0, assertmsg
+    else:
+        logger.info('Check Ping IPv4 from R1(r1-vrf-101) to R2(192.168.101.41) OK')
+
+def test_memory_leak():
+    "Run the memory leak test and report results."
+    tgen = get_topogen()
+    if not tgen.is_memleak_enabled():
+        pytest.skip('Memory leak test/report is disabled')
+
+    tgen.report_memory_leaks()
+
+
+if __name__ == '__main__':
+    args = ["-s"] + sys.argv[1:]
+    sys.exit(pytest.main(args))
diff --git a/zebra/if_netlink.c b/zebra/if_netlink.c
index 5cd3e69299..90a08bbd6c 100644
--- a/zebra/if_netlink.c
+++ b/zebra/if_netlink.c
@@ -798,8 +798,10 @@ static int netlink_interface(struct nlmsghdr *h, ns_id_t ns_id, int startup)
 	if (tb[IFLA_LINK])
 		link_ifindex = *(ifindex_t *)RTA_DATA(tb[IFLA_LINK]);
 
-	if (tb[IFLA_LINK_NETNSID])
+	if (tb[IFLA_LINK_NETNSID]) {
 		link_nsid = *(ns_id_t *)RTA_DATA(tb[IFLA_LINK_NETNSID]);
+		link_nsid = ns_id_get_absolute(ns_id, link_nsid);
+	}
 
 	/* Add interface.
 	 * We add by index first because in some cases such as the master
@@ -848,7 +850,7 @@ static int netlink_interface(struct nlmsghdr *h, ns_id_t ns_id, int startup)
 	netlink_interface_update_l2info(ifp, linkinfo[IFLA_INFO_DATA],
 					1, link_nsid);
 	if (IS_ZEBRA_IF_BRIDGE_SLAVE(ifp))
-		zebra_l2if_update_bridge_slave(ifp, bridge_ifindex);
+		zebra_l2if_update_bridge_slave(ifp, bridge_ifindex, ns_id);
 	else if (IS_ZEBRA_IF_BOND_SLAVE(ifp))
 		zebra_l2if_update_bond_slave(ifp, bond_ifindex);
 
@@ -1349,9 +1351,10 @@ int netlink_link_change(struct nlmsghdr *h, ns_id_t ns_id, int startup)
 	if (tb[IFLA_LINK])
 		link_ifindex = *(ifindex_t *)RTA_DATA(tb[IFLA_LINK]);
 
-	if (tb[IFLA_LINK_NETNSID])
+	if (tb[IFLA_LINK_NETNSID]) {
 		link_nsid = *(ns_id_t *)RTA_DATA(tb[IFLA_LINK_NETNSID]);
-
+		link_nsid = ns_id_get_absolute(ns_id, link_nsid);
+	}
 	if (tb[IFLA_IFALIAS]) {
 		desc = (char *)RTA_DATA(tb[IFLA_IFALIAS]);
 	}
@@ -1439,7 +1442,8 @@ int netlink_link_change(struct nlmsghdr *h, ns_id_t ns_id, int startup)
 				1, link_nsid);
 			if (IS_ZEBRA_IF_BRIDGE_SLAVE(ifp))
 				zebra_l2if_update_bridge_slave(ifp,
-							       bridge_ifindex);
+							       bridge_ifindex,
+							       ns_id);
 			else if (IS_ZEBRA_IF_BOND_SLAVE(ifp))
 				zebra_l2if_update_bond_slave(ifp, bond_ifindex);
 		} else if (ifp->vrf_id != vrf_id) {
@@ -1540,7 +1544,8 @@ int netlink_link_change(struct nlmsghdr *h, ns_id_t ns_id, int startup)
 				0, link_nsid);
 			if (IS_ZEBRA_IF_BRIDGE_SLAVE(ifp) || was_bridge_slave)
 				zebra_l2if_update_bridge_slave(ifp,
-							       bridge_ifindex);
+							       bridge_ifindex,
+							       ns_id);
 			else if (IS_ZEBRA_IF_BOND_SLAVE(ifp) || was_bond_slave)
 				zebra_l2if_update_bond_slave(ifp, bond_ifindex);
 		}
diff --git a/zebra/interface.c b/zebra/interface.c
index b824e313ec..d29f61450e 100644
--- a/zebra/interface.c
+++ b/zebra/interface.c
@@ -1242,6 +1242,23 @@ static void nbr_connected_dump_vty(struct vty *vty,
 	vty_out(vty, "\n");
 }
 
+static const char *zebra_zifslavetype_2str(zebra_slave_iftype_t zif_slave_type)
+{
+	switch (zif_slave_type) {
+	case ZEBRA_IF_SLAVE_BRIDGE:
+		return "Bridge";
+	case ZEBRA_IF_SLAVE_VRF:
+		return "Vrf";
+	case ZEBRA_IF_SLAVE_BOND:
+		return "Bond";
+	case ZEBRA_IF_SLAVE_OTHER:
+		return "Other";
+	case ZEBRA_IF_SLAVE_NONE:
+		return "None";
+	}
+	return "None";
+}
+
 static const char *zebra_ziftype_2str(zebra_iftype_t zif_type)
 {
 	switch (zif_type) {
@@ -1469,6 +1486,9 @@ static void if_dump_vty(struct vty *vty, struct interface *ifp)
 
 	vty_out(vty, "  Interface Type %s\n",
 		zebra_ziftype_2str(zebra_if->zif_type));
+	vty_out(vty, "  Interface Slave Type %s\n",
+		zebra_zifslavetype_2str(zebra_if->zif_slave_type));
+
 	if (IS_ZEBRA_IF_BRIDGE(ifp)) {
 		struct zebra_l2info_bridge *bridge_info;
 
diff --git a/zebra/main.c b/zebra/main.c
index cfc45567d7..2afef46bb2 100644
--- a/zebra/main.c
+++ b/zebra/main.c
@@ -183,7 +183,7 @@ static void sigint(void)
 	vrf_terminate();
 	rtadv_terminate();
 
-	ns_walk_func(zebra_ns_early_shutdown);
+	ns_walk_func(zebra_ns_early_shutdown, NULL, NULL);
 	zebra_ns_notify_close();
 
 	access_list_reset();
@@ -214,7 +214,7 @@ int zebra_finalize(struct thread *dummy)
 	zlog_info("Zebra final shutdown");
 
 	/* Final shutdown of ns resources */
-	ns_walk_func(zebra_ns_final_shutdown);
+	ns_walk_func(zebra_ns_final_shutdown, NULL, NULL);
 
 	/* Stop dplane thread and finish any cleanup */
 	zebra_dplane_shutdown();
diff --git a/zebra/zebra_evpn.c b/zebra/zebra_evpn.c
index 653783f28c..56699c4e83 100644
--- a/zebra/zebra_evpn.c
+++ b/zebra/zebra_evpn.c
@@ -622,32 +622,31 @@ void zebra_evpn_svi_macip_del_for_evpn_hash(struct hash_bucket *bucket,
 	return;
 }
 
-/*
- * Map port or (port, VLAN) to an EVPN. This is invoked upon getting MAC
- * notifications, to see if they are of interest.
- */
-zebra_evpn_t *zebra_evpn_map_vlan(struct interface *ifp,
-				  struct interface *br_if, vlanid_t vid)
+static int zebra_evpn_map_vlan_ns(struct ns *ns,
+				  void *_in_param,
+				  void **_p_zevpn)
 {
-	struct zebra_ns *zns;
+	struct zebra_ns *zns = ns->info;
 	struct route_node *rn;
+	struct interface *br_if;
+	zebra_evpn_t **p_zevpn = (zebra_evpn_t **)_p_zevpn;
+	zebra_evpn_t *zevpn;
 	struct interface *tmp_if = NULL;
 	struct zebra_if *zif;
-	struct zebra_l2info_bridge *br;
 	struct zebra_l2info_vxlan *vxl = NULL;
-	uint8_t bridge_vlan_aware;
-	zebra_evpn_t *zevpn;
+	struct zebra_from_svi_param *in_param =
+		(struct zebra_from_svi_param *)_in_param;
 	int found = 0;
 
-	/* Determine if bridge is VLAN-aware or not */
-	zif = br_if->info;
+	if (!in_param)
+		return NS_WALK_STOP;
+	br_if = in_param->br_if;
+	zif = in_param->zif;
 	assert(zif);
-	br = &zif->l2info.br;
-	bridge_vlan_aware = br->vlan_aware;
+	assert(br_if);
 
 	/* See if this interface (or interface plus VLAN Id) maps to a VxLAN */
 	/* TODO: Optimize with a hash. */
-	zns = zebra_ns_lookup(NS_DEFAULT);
 	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
 		tmp_if = (struct interface *)rn->info;
 		if (!tmp_if)
@@ -662,17 +661,101 @@ zebra_evpn_t *zebra_evpn_map_vlan(struct interface *ifp,
 		if (zif->brslave_info.br_if != br_if)
 			continue;
 
-		if (!bridge_vlan_aware || vxl->access_vlan == vid) {
+		if (!in_param->bridge_vlan_aware
+		    || vxl->access_vlan == in_param->vid) {
+			found = 1;
+			break;
+		}
+	}
+	if (!found)
+		return NS_WALK_CONTINUE;
+
+	zevpn = zebra_evpn_lookup(vxl->vni);
+	if (p_zevpn)
+		*p_zevpn = zevpn;
+	return NS_WALK_STOP;
+}
+
+/*
+ * Map port or (port, VLAN) to an EVPN. This is invoked upon getting MAC
+ * notifications, to see if they are of interest.
+ */
+zebra_evpn_t *zebra_evpn_map_vlan(struct interface *ifp,
+				  struct interface *br_if, vlanid_t vid)
+{
+	struct zebra_if *zif;
+	struct zebra_l2info_bridge *br;
+	zebra_evpn_t **p_zevpn;
+	zebra_evpn_t *zevpn = NULL;
+	struct zebra_from_svi_param in_param;
+
+	/* Determine if bridge is VLAN-aware or not */
+	zif = br_if->info;
+	assert(zif);
+	br = &zif->l2info.br;
+	in_param.bridge_vlan_aware = br->vlan_aware;
+	in_param.vid = vid;
+	in_param.br_if = br_if;
+	in_param.zif = zif;
+	p_zevpn = &zevpn;
+
+	ns_walk_func(zebra_evpn_map_vlan_ns,
+		     (void *)&in_param,
+		     (void **)p_zevpn);
+	return zevpn;
+}
+
+static int zebra_evpn_from_svi_ns(struct ns *ns,
+				  void *_in_param,
+				  void **_p_zevpn)
+{
+	struct zebra_ns *zns = ns->info;
+	struct route_node *rn;
+	struct interface *br_if;
+	zebra_evpn_t **p_zevpn = (zebra_evpn_t **)_p_zevpn;
+	zebra_evpn_t *zevpn;
+	struct interface *tmp_if = NULL;
+	struct zebra_if *zif;
+	struct zebra_l2info_vxlan *vxl = NULL;
+	struct zebra_from_svi_param *in_param =
+		(struct zebra_from_svi_param *)_in_param;
+	int found = 0;
+
+	if (!in_param)
+		return NS_WALK_STOP;
+	br_if = in_param->br_if;
+	zif = in_param->zif;
+	assert(zif);
+
+	/* TODO: Optimize with a hash. */
+	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
+		tmp_if = (struct interface *)rn->info;
+		if (!tmp_if)
+			continue;
+		zif = tmp_if->info;
+		if (!zif || zif->zif_type != ZEBRA_IF_VXLAN)
+			continue;
+		if (!if_is_operative(tmp_if))
+			continue;
+		vxl = &zif->l2info.vxl;
+
+		if (zif->brslave_info.br_if != br_if)
+			continue;
+
+		if (!in_param->bridge_vlan_aware
+		    || vxl->access_vlan == in_param->vid) {
 			found = 1;
 			break;
 		}
 	}
 
 	if (!found)
-		return NULL;
+		return NS_WALK_CONTINUE;
 
 	zevpn = zebra_evpn_lookup(vxl->vni);
-	return zevpn;
+	if (p_zevpn)
+		*p_zevpn = zevpn;
+	return NS_WALK_STOP;
 }
 
 /*
@@ -682,16 +765,11 @@ zebra_evpn_t *zebra_evpn_map_vlan(struct interface *ifp,
 zebra_evpn_t *zebra_evpn_from_svi(struct interface *ifp,
 				  struct interface *br_if)
 {
-	struct zebra_ns *zns;
-	struct route_node *rn;
-	struct interface *tmp_if = NULL;
-	struct zebra_if *zif;
 	struct zebra_l2info_bridge *br;
-	struct zebra_l2info_vxlan *vxl = NULL;
-	uint8_t bridge_vlan_aware;
-	vlanid_t vid = 0;
-	zebra_evpn_t *zevpn;
-	int found = 0;
+	zebra_evpn_t *zevpn = NULL;
+	zebra_evpn_t **p_zevpn;
+	struct zebra_if *zif;
+	struct zebra_from_svi_param in_param;
 
 	if (!br_if)
 		return NULL;
@@ -704,8 +782,10 @@ zebra_evpn_t *zebra_evpn_from_svi(struct interface *ifp,
 	zif = br_if->info;
 	assert(zif);
 	br = &zif->l2info.br;
-	bridge_vlan_aware = br->vlan_aware;
-	if (bridge_vlan_aware) {
+	in_param.bridge_vlan_aware = br->vlan_aware;
+	in_param.vid = 0;
+
+	if (in_param.bridge_vlan_aware) {
 		struct zebra_l2info_vlan *vl;
 
 		if (!IS_ZEBRA_IF_VLAN(ifp))
@@ -714,37 +794,52 @@ zebra_evpn_t *zebra_evpn_from_svi(struct interface *ifp,
 		zif = ifp->info;
 		assert(zif);
 		vl = &zif->l2info.vl;
-		vid = vl->vid;
+		in_param.vid = vl->vid;
 	}
 
+	in_param.br_if = br_if;
+	in_param.zif = zif;
+	p_zevpn = &zevpn;
 	/* See if this interface (or interface plus VLAN Id) maps to a VxLAN */
-	/* TODO: Optimize with a hash. */
-	zns = zebra_ns_lookup(NS_DEFAULT);
+	ns_walk_func(zebra_evpn_from_svi_ns, (void *)&in_param,
+		     (void **)p_zevpn);
+	return zevpn;
+}
+
+static int zvni_map_to_macvlan_ns(struct ns *ns,
+				  void *_in_param,
+				  void **_p_ifp)
+{
+	struct zebra_ns *zns = ns->info;
+	struct zebra_from_svi_param *in_param =
+		(struct zebra_from_svi_param *)_in_param;
+	struct interface **p_ifp = (struct interface **)_p_ifp;
+	struct route_node *rn;
+	struct interface *tmp_if = NULL;
+	struct zebra_if *zif;
+
+	if (!in_param)
+		return NS_WALK_STOP;
+
+	/* Identify corresponding VLAN interface. */
 	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
 		tmp_if = (struct interface *)rn->info;
-		if (!tmp_if)
+		/* Check oper status of the SVI. */
+		if (!tmp_if || !if_is_operative(tmp_if))
 			continue;
 		zif = tmp_if->info;
-		if (!zif || zif->zif_type != ZEBRA_IF_VXLAN)
-			continue;
-		if (!if_is_operative(tmp_if))
-			continue;
-		vxl = &zif->l2info.vxl;
 
-		if (zif->brslave_info.br_if != br_if)
+		if (!zif || zif->zif_type != ZEBRA_IF_MACVLAN)
 			continue;
 
-		if (!bridge_vlan_aware || vxl->access_vlan == vid) {
-			found = 1;
-			break;
+		if (zif->link == in_param->svi_if) {
+			if (p_ifp)
+				*p_ifp = tmp_if;
+			return NS_WALK_STOP;
 		}
 	}
 
-	if (!found)
-		return NULL;
-
-	zevpn = zebra_evpn_lookup(vxl->vni);
-	return zevpn;
+	return NS_WALK_CONTINUE;
 }
 
 /* Map to MAC-VLAN interface corresponding to specified SVI interface.
@@ -752,11 +847,10 @@ zebra_evpn_t *zebra_evpn_from_svi(struct interface *ifp,
 struct interface *zebra_evpn_map_to_macvlan(struct interface *br_if,
 					    struct interface *svi_if)
 {
-	struct zebra_ns *zns;
-	struct route_node *rn;
 	struct interface *tmp_if = NULL;
 	struct zebra_if *zif;
-	int found = 0;
+	struct interface **p_ifp;
+	struct zebra_from_svi_param in_param;
 
 	/* Defensive check, caller expected to invoke only with valid bridge. */
 	if (!br_if)
@@ -771,25 +865,17 @@ struct interface *zebra_evpn_map_to_macvlan(struct interface *br_if,
 	zif = br_if->info;
 	assert(zif);
 
+	in_param.vid = 0;
+	in_param.br_if = br_if;
+	in_param.zif = NULL;
+	in_param.svi_if = svi_if;
+	p_ifp = &tmp_if;
+
 	/* Identify corresponding VLAN interface. */
-	zns = zebra_ns_lookup(NS_DEFAULT);
-	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
-		tmp_if = (struct interface *)rn->info;
-		/* Check oper status of the SVI. */
-		if (!tmp_if || !if_is_operative(tmp_if))
-			continue;
-		zif = tmp_if->info;
-
-		if (!zif || zif->zif_type != ZEBRA_IF_MACVLAN)
-			continue;
-
-		if (zif->link == svi_if) {
-			found = 1;
-			break;
-		}
-	}
-
-	return found ? tmp_if : NULL;
+	ns_walk_func(zvni_map_to_macvlan_ns,
+		     (void *)&in_param,
+		     (void **)p_ifp);
+	return tmp_if;
 }
 
 /*
@@ -812,6 +898,7 @@ void zebra_evpn_install_mac_hash(struct hash_bucket *bucket, void *ctxt)
 void zebra_evpn_read_mac_neigh(zebra_evpn_t *zevpn, struct interface *ifp)
 {
 	struct zebra_ns *zns;
+	struct zebra_vrf *zvrf;
 	struct zebra_if *zif;
 	struct interface *vlan_if;
 	struct zebra_l2info_vxlan *vxl;
@@ -819,7 +906,10 @@ void zebra_evpn_read_mac_neigh(zebra_evpn_t *zevpn, struct interface *ifp)
 
 	zif = ifp->info;
 	vxl = &zif->l2info.vxl;
-	zns = zebra_ns_lookup(NS_DEFAULT);
+	zvrf = zebra_vrf_lookup_by_id(zevpn->vrf_id);
+	if (!zvrf || !zvrf->zns)
+		return;
+	zns = zvrf->zns;
 
 	if (IS_ZEBRA_DEBUG_VXLAN)
 		zlog_debug(
@@ -1315,7 +1405,7 @@ void process_remote_macip_add(vni_t vni, struct ethaddr *macaddr,
 		}
 	}
 
-	zvrf = vrf_info_lookup(zevpn->vxlan_if->vrf_id);
+	zvrf = zebra_vrf_get_evpn();
 	if (!zvrf)
 		return;
 
diff --git a/zebra/zebra_evpn.h b/zebra/zebra_evpn.h
index 3b6a5b21e8..27392ec85c 100644
--- a/zebra/zebra_evpn.h
+++ b/zebra/zebra_evpn.h
@@ -123,6 +123,15 @@ struct zebra_evpn_t_ {
 	struct list *local_es_evi_list;
 };
 
+/* for parsing evpn and vni contexts */
+struct zebra_from_svi_param {
+	struct interface *br_if;
+	struct interface *svi_if;
+	struct zebra_if *zif;
+	uint8_t bridge_vlan_aware;
+	vlanid_t vid;
+};
+
 struct interface *zvni_map_to_svi(vlanid_t vid, struct interface *br_if);
 
 static inline struct interface *zevpn_map_to_svi(zebra_evpn_t *zevpn)
diff --git a/zebra/zebra_evpn_mac.c b/zebra/zebra_evpn_mac.c
index eb4e4f9a57..75031ddba6 100644
--- a/zebra/zebra_evpn_mac.c
+++ b/zebra/zebra_evpn_mac.c
@@ -249,7 +249,7 @@ static void zebra_evpn_mac_get_access_info(zebra_mac_t *mac,
 		struct zebra_ns *zns;
 
 		*vid = mac->fwd_info.local.vid;
-		zns = zebra_ns_lookup(NS_DEFAULT);
+		zns = zebra_ns_lookup(mac->fwd_info.local.ns_id);
 		*ifpP = if_lookup_by_index_per_ns(zns,
 						  mac->fwd_info.local.ifindex);
 	}
@@ -1610,6 +1610,12 @@ static bool zebra_evpn_local_mac_update_fwd_info(zebra_mac_t *mac,
 {
 	struct zebra_if *zif = ifp->info;
 	bool es_change;
+	ns_id_t local_ns_id = NS_DEFAULT;
+	struct zebra_vrf *zvrf;
+
+	zvrf = zebra_vrf_lookup_by_id(ifp->vrf_id);
+	if (zvrf && zvrf->zns)
+		local_ns_id = zvrf->zns->ns_id;
 
 	memset(&mac->fwd_info, 0, sizeof(mac->fwd_info));
 
@@ -1618,6 +1624,7 @@ static bool zebra_evpn_local_mac_update_fwd_info(zebra_mac_t *mac,
 	if (!mac->es) {
 		/* if es is set fwd_info is not-relevant/taped-out */
 		mac->fwd_info.local.ifindex = ifp->ifindex;
+		mac->fwd_info.local.ns_id = local_ns_id;
 		mac->fwd_info.local.vid = vid;
 	}
 
@@ -2206,6 +2213,12 @@ int zebra_evpn_mac_gw_macip_add(struct interface *ifp, zebra_evpn_t *zevpn,
 {
 	char buf[ETHER_ADDR_STRLEN];
 	zebra_mac_t *mac;
+	ns_id_t local_ns_id = NS_DEFAULT;
+	struct zebra_vrf *zvrf;
+
+	zvrf = zebra_vrf_lookup_by_id(ifp->vrf_id);
+	if (zvrf && zvrf->zns)
+		local_ns_id = zvrf->zns->ns_id;
 
 	mac = zebra_evpn_mac_lookup(zevpn, macaddr);
 	if (!mac) {
@@ -2225,6 +2238,7 @@ int zebra_evpn_mac_gw_macip_add(struct interface *ifp, zebra_evpn_t *zevpn,
 	SET_FLAG(mac->flags, ZEBRA_MAC_DEF_GW);
 	memset(&mac->fwd_info, 0, sizeof(mac->fwd_info));
 	mac->fwd_info.local.ifindex = ifp->ifindex;
+	mac->fwd_info.local.ns_id = local_ns_id;
 	mac->fwd_info.local.vid = vlan_id;
 
 	*macp = mac;
diff --git a/zebra/zebra_evpn_mac.h b/zebra/zebra_evpn_mac.h
index 39aaf1fb30..f9ca81445f 100644
--- a/zebra/zebra_evpn_mac.h
+++ b/zebra/zebra_evpn_mac.h
@@ -91,6 +91,7 @@ struct zebra_mac_t_ {
 	union {
 		struct {
 			ifindex_t ifindex;
+			ns_id_t ns_id;
 			vlanid_t vid;
 		} local;
 
diff --git a/zebra/zebra_l2.c b/zebra/zebra_l2.c
index 1758c8f96a..417056ecb0 100644
--- a/zebra/zebra_l2.c
+++ b/zebra/zebra_l2.c
@@ -54,7 +54,13 @@ static void map_slaves_to_bridge(struct interface *br_if, int link)
 {
 	struct vrf *vrf;
 	struct interface *ifp;
+	struct zebra_vrf *zvrf;
+	struct zebra_ns *zns;
 
+	zvrf = zebra_vrf_lookup_by_id(br_if->vrf_id);
+	assert(zvrf);
+	zns = zvrf->zns;
+	assert(zns);
 	RB_FOREACH (vrf, vrf_name_head, &vrfs_by_name) {
 		FOR_ALL_INTERFACES (vrf, ifp) {
 			struct zebra_if *zif;
@@ -73,7 +79,8 @@ static void map_slaves_to_bridge(struct interface *br_if, int link)
 			br_slave = &zif->brslave_info;
 
 			if (link) {
-				if (br_slave->bridge_ifindex == br_if->ifindex)
+				if (br_slave->bridge_ifindex == br_if->ifindex &&
+				    br_slave->ns_id == zns->ns_id)
 					br_slave->br_if = br_if;
 			} else {
 				if (br_slave->br_if == br_if)
@@ -84,12 +91,14 @@ static void map_slaves_to_bridge(struct interface *br_if, int link)
 }
 
 /* Public functions */
-void zebra_l2_map_slave_to_bridge(struct zebra_l2info_brslave *br_slave)
+void zebra_l2_map_slave_to_bridge(struct zebra_l2info_brslave *br_slave,
+				  struct zebra_ns *zns)
 {
 	struct interface *br_if;
 
 	/* TODO: Handle change of master */
-	br_if = if_lookup_by_index_per_ns(zebra_ns_lookup(NS_DEFAULT),
+	assert(zns);
+	br_if = if_lookup_by_index_per_ns(zebra_ns_lookup(zns->ns_id),
 					  br_slave->bridge_ifindex);
 	if (br_if)
 		br_slave->br_if = br_if;
@@ -248,23 +257,32 @@ void zebra_l2_vxlanif_del(struct interface *ifp)
  * from a bridge before it can be mapped to another bridge.
  */
 void zebra_l2if_update_bridge_slave(struct interface *ifp,
-				    ifindex_t bridge_ifindex)
+				    ifindex_t bridge_ifindex,
+				    ns_id_t ns_id)
 {
 	struct zebra_if *zif;
 	ifindex_t old_bridge_ifindex;
+	ns_id_t old_ns_id;
+	struct zebra_vrf *zvrf;
 
 	zif = ifp->info;
 	assert(zif);
 
-	old_bridge_ifindex = zif->brslave_info.bridge_ifindex;
-	if (old_bridge_ifindex == bridge_ifindex)
+	zvrf = zebra_vrf_lookup_by_id(ifp->vrf_id);
+	if (!zvrf)
 		return;
 
-	zif->brslave_info.bridge_ifindex = bridge_ifindex;
+	old_bridge_ifindex = zif->brslave_info.bridge_ifindex;
+	old_ns_id = zif->brslave_info.ns_id;
+	if (old_bridge_ifindex == bridge_ifindex &&
+	    old_ns_id == zif->brslave_info.ns_id)
+		return;
 
+	zif->brslave_info.ns_id = ns_id;
+	zif->brslave_info.bridge_ifindex = bridge_ifindex;
 	/* Set up or remove link with master */
 	if (bridge_ifindex != IFINDEX_INTERNAL) {
-		zebra_l2_map_slave_to_bridge(&zif->brslave_info);
+		zebra_l2_map_slave_to_bridge(&zif->brslave_info, zvrf->zns);
 		/* In the case of VxLAN, invoke the handler for EVPN. */
 		if (zif->zif_type == ZEBRA_IF_VXLAN)
 			zebra_vxlan_if_update(ifp, ZEBRA_VXLIF_MASTER_CHANGE);
diff --git a/zebra/zebra_l2.h b/zebra/zebra_l2.h
index 2735d915ec..f3b15c7770 100644
--- a/zebra/zebra_l2.h
+++ b/zebra/zebra_l2.h
@@ -37,6 +37,7 @@ extern "C" {
 struct zebra_l2info_brslave {
 	ifindex_t bridge_ifindex; /* Bridge Master */
 	struct interface *br_if;  /* Pointer to master */
+	ns_id_t ns_id; /* network namespace where bridge is */
 };
 
 /* zebra L2 interface information - bridge interface */
@@ -81,7 +82,8 @@ union zebra_l2if_info {
 
 #define IS_ZEBRA_IF_BRIDGE_VLAN_AWARE(zif) ((zif)->l2info.br.vlan_aware == 1)
 
-extern void zebra_l2_map_slave_to_bridge(struct zebra_l2info_brslave *br_slave);
+extern void zebra_l2_map_slave_to_bridge(struct zebra_l2info_brslave *br_slave,
+					 struct zebra_ns *zns);
 extern void
 zebra_l2_unmap_slave_from_bridge(struct zebra_l2info_brslave *br_slave);
 extern void
@@ -101,7 +103,8 @@ extern void zebra_l2_vxlanif_update_access_vlan(struct interface *ifp,
 						vlanid_t access_vlan);
 extern void zebra_l2_vxlanif_del(struct interface *ifp);
 extern void zebra_l2if_update_bridge_slave(struct interface *ifp,
-					   ifindex_t bridge_ifindex);
+					   ifindex_t bridge_ifindex,
+					   ns_id_t ns_id);
 
 extern void zebra_l2if_update_bond_slave(struct interface *ifp,
 					 ifindex_t bond_ifindex);
diff --git a/zebra/zebra_netns_id.c b/zebra/zebra_netns_id.c
index 8de4daf439..79121bb086 100644
--- a/zebra/zebra_netns_id.c
+++ b/zebra/zebra_netns_id.c
@@ -159,27 +159,34 @@ static ns_id_t extract_nsid(struct nlmsghdr *nlh, char *buf)
 	return ns_id;
 }
 
-ns_id_t zebra_ns_id_get(const char *netnspath)
+/* fd_param = -1 is ignored.
+ * netnspath set to null is ignored.
+ * one of the 2 params is mandatory. netnspath is looked in priority
+ */
+ns_id_t zebra_ns_id_get(const char *netnspath, int fd_param)
 {
 	int ns_id = -1;
 	struct sockaddr_nl snl;
-	int fd, sock, ret;
+	int fd = -1, sock, ret;
 	unsigned int seq;
 	ns_id_t return_nsid = NS_UNKNOWN;
 
 	/* netns path check */
-	if (!netnspath)
+	if (!netnspath && fd_param == -1)
 		return NS_UNKNOWN;
-	fd = open(netnspath, O_RDONLY);
-	if (fd == -1)
-		return NS_UNKNOWN;
-
+	if (netnspath)  {
+		fd = open(netnspath, O_RDONLY);
+		if (fd == -1)
+			return NS_UNKNOWN;
+	} else if (fd_param != -1)
+		fd = fd_param;
 	/* netlink socket */
 	sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
 	if (sock < 0) {
 		flog_err_sys(EC_LIB_SOCKET, "netlink( %u) socket() error: %s",
 			     sock, safe_strerror(errno));
-		close(fd);
+		if (fd_param == -1)
+			close(fd);
 		return NS_UNKNOWN;
 	}
 	memset(&snl, 0, sizeof(snl));
@@ -192,7 +199,8 @@ ns_id_t zebra_ns_id_get(const char *netnspath)
 			     "netlink( %u) socket() bind error: %s", sock,
 			     safe_strerror(errno));
 		close(sock);
-		close(fd);
+		if (fd_param == -1)
+			close(fd);
 		return NS_UNKNOWN;
 	}
 
@@ -214,7 +222,8 @@ ns_id_t zebra_ns_id_get(const char *netnspath)
 	ret = send_receive(sock, nlh, seq, buf);
 	if (ret < 0) {
 		close(sock);
-		close(fd);
+		if (fd_param == -1)
+			close(fd);
 		return NS_UNKNOWN;
 	}
 	nlh = (struct nlmsghdr *)buf;
@@ -258,7 +267,8 @@ ns_id_t zebra_ns_id_get(const char *netnspath)
 				"netlink( %u) recvfrom() error 2 when reading: %s",
 				fd, safe_strerror(errno));
 			close(sock);
-			close(fd);
+			if (fd_param == -1)
+				close(fd);
 			if (errno == ENOTSUP) {
 				zlog_debug("NEWNSID locally generated");
 				return zebra_ns_id_get_fallback(netnspath);
@@ -279,7 +289,8 @@ ns_id_t zebra_ns_id_get(const char *netnspath)
 		ret = send_receive(sock, nlh, seq, buf);
 		if (ret < 0) {
 			close(sock);
-			close(fd);
+			if (fd_param == -1)
+				close(fd);
 			return NS_UNKNOWN;
 		}
 		nlh = (struct nlmsghdr *)buf;
@@ -310,16 +321,18 @@ ns_id_t zebra_ns_id_get(const char *netnspath)
 		} while (len != 0 && ret == 0);
 	}
 
-	close(fd);
+	if (fd_param == -1)
+		close(fd);
 	close(sock);
 	return return_nsid;
 }
 
 #else
-ns_id_t zebra_ns_id_get(const char *netnspath)
+ns_id_t zebra_ns_id_get(const char *netnspath, int fd __attribute__ ((unused)))
 {
 	return zebra_ns_id_get_fallback(netnspath);
 }
+
 #endif /* ! defined(HAVE_NETLINK) */
 
 #ifdef HAVE_NETNS
@@ -355,7 +368,7 @@ ns_id_t zebra_ns_id_get_default(void)
 		return NS_DEFAULT_INTERNAL;
 	}
 	close(fd);
-	return zebra_ns_id_get((char *)NS_DEFAULT_NAME);
+	return zebra_ns_id_get((char *)NS_DEFAULT_NAME, -1);
 #else  /* HAVE_NETNS */
 	return NS_DEFAULT_INTERNAL;
 #endif /* !HAVE_NETNS */
diff --git a/zebra/zebra_netns_id.h b/zebra/zebra_netns_id.h
index 7a5f6851f4..dd9eab18e0 100644
--- a/zebra/zebra_netns_id.h
+++ b/zebra/zebra_netns_id.h
@@ -24,7 +24,7 @@
 extern "C" {
 #endif
 
-extern ns_id_t zebra_ns_id_get(const char *netnspath);
+extern ns_id_t zebra_ns_id_get(const char *netnspath, int fd);
 extern ns_id_t zebra_ns_id_get_default(void);
 
 #ifdef __cplusplus
diff --git a/zebra/zebra_netns_notify.c b/zebra/zebra_netns_notify.c
index ec7681bf23..995fa6fb5a 100644
--- a/zebra/zebra_netns_notify.c
+++ b/zebra/zebra_netns_notify.c
@@ -72,13 +72,14 @@ static void zebra_ns_notify_create_context_from_entry_name(const char *name)
 	char *netnspath = ns_netns_pathname(NULL, name);
 	struct vrf *vrf;
 	int ret;
-	ns_id_t ns_id, ns_id_external;
+	ns_id_t ns_id, ns_id_external, ns_id_relative = NS_UNKNOWN;
+	struct ns *default_ns;
 
 	if (netnspath == NULL)
 		return;
 
 	frr_with_privs(&zserv_privs) {
-		ns_id = zebra_ns_id_get(netnspath);
+		ns_id = zebra_ns_id_get(netnspath, -1);
 	}
 	if (ns_id == NS_UNKNOWN)
 		return;
@@ -97,9 +98,21 @@ static void zebra_ns_notify_create_context_from_entry_name(const char *name)
 		ns_map_nsid_with_external(ns_id, false);
 		return;
 	}
+
+	default_ns = ns_get_default();
+
+	/* force kernel ns_id creation in that new vrf */
+	frr_with_privs(&zserv_privs) {
+		ns_switch_to_netns(netnspath);
+		ns_id_relative = zebra_ns_id_get(NULL, default_ns->fd);
+		ns_switchback_to_initial();
+	}
+
 	frr_with_privs(&zserv_privs) {
 		ret = vrf_netns_handler_create(NULL, vrf, netnspath,
-					       ns_id_external, ns_id);
+					       ns_id_external,
+					       ns_id,
+					       ns_id_relative);
 	}
 	if (ret != CMD_SUCCESS) {
 		flog_warn(EC_ZEBRA_NS_VRF_CREATION_FAILED,
diff --git a/zebra/zebra_ns.c b/zebra/zebra_ns.c
index 4e51437337..6462daf687 100644
--- a/zebra/zebra_ns.c
+++ b/zebra/zebra_ns.c
@@ -153,20 +153,25 @@ static int zebra_ns_disable_internal(struct zebra_ns *zns, bool complete)
 /* During zebra shutdown, do partial cleanup while the async dataplane
  * is still running.
  */
-int zebra_ns_early_shutdown(struct ns *ns)
+int zebra_ns_early_shutdown(struct ns *ns,
+			    void *param_in __attribute__((unused)),
+			    void **param_out __attribute__((unused)))
 {
 	struct zebra_ns *zns = ns->info;
 
 	if (zns == NULL)
 		return 0;
 
-	return zebra_ns_disable_internal(zns, false);
+	zebra_ns_disable_internal(zns, false);
+	return NS_WALK_CONTINUE;
 }
 
 /* During zebra shutdown, do final cleanup
  * after all dataplane work is complete.
  */
-int zebra_ns_final_shutdown(struct ns *ns)
+int zebra_ns_final_shutdown(struct ns *ns,
+			    void *param_in __attribute__((unused)),
+			    void **param_out __attribute__((unused)))
 {
 	struct zebra_ns *zns = ns->info;
 
@@ -175,7 +180,7 @@ int zebra_ns_final_shutdown(struct ns *ns)
 
 	kernel_terminate(zns, true);
 
-	return 0;
+	return NS_WALK_CONTINUE;
 }
 
 int zebra_ns_init(const char *optional_default_name)
@@ -183,12 +188,16 @@ int zebra_ns_init(const char *optional_default_name)
 	struct ns *default_ns;
 	ns_id_t ns_id;
 	ns_id_t ns_id_external;
+	struct ns *ns;
 
 	frr_with_privs(&zserv_privs) {
 		ns_id = zebra_ns_id_get_default();
 	}
 	ns_id_external = ns_map_nsid_with_external(ns_id, true);
 	ns_init_management(ns_id_external, ns_id);
+	ns = ns_get_default();
+	if (ns)
+		ns->relative_default_ns = ns_id;
 
 	default_ns = ns_lookup(ns_get_default_id());
 	if (!default_ns) {
diff --git a/zebra/zebra_ns.h b/zebra/zebra_ns.h
index dc79a83db0..f7d1f40782 100644
--- a/zebra/zebra_ns.h
+++ b/zebra/zebra_ns.h
@@ -67,9 +67,12 @@ struct zebra_ns *zebra_ns_lookup(ns_id_t ns_id);
 int zebra_ns_init(const char *optional_default_name);
 int zebra_ns_enable(ns_id_t ns_id, void **info);
 int zebra_ns_disabled(struct ns *ns);
-int zebra_ns_early_shutdown(struct ns *ns);
-int zebra_ns_final_shutdown(struct ns *ns);
-
+int zebra_ns_early_shutdown(struct ns *ns,
+			    void *param_in __attribute__((unused)),
+			    void **param_out __attribute__((unused)));
+int zebra_ns_final_shutdown(struct ns *ns,
+			    void *param_in __attribute__((unused)),
+			    void **param_out __attribute__((unused)));
 int zebra_ns_config_write(struct vty *vty, struct ns *ns);
 
 #ifdef __cplusplus
diff --git a/zebra/zebra_vxlan.c b/zebra/zebra_vxlan.c
index 13a58bc34a..61498973e9 100644
--- a/zebra/zebra_vxlan.c
+++ b/zebra/zebra_vxlan.c
@@ -782,6 +782,43 @@ static void zl3vni_print_hash_detail(struct hash_bucket *bucket, void *data)
 		vty_out(vty, "\n");
 }
 
+static int zvni_map_to_svi_ns(struct ns *ns,
+			      void *_in_param,
+			      void **_p_ifp)
+{
+	struct zebra_ns *zns = ns->info;
+	struct route_node *rn;
+	struct zebra_from_svi_param *in_param =
+		(struct zebra_from_svi_param *)_in_param;
+	struct zebra_l2info_vlan *vl;
+	struct interface *tmp_if = NULL;
+	struct interface **p_ifp = (struct interface **)_p_ifp;
+	struct zebra_if *zif;
+
+	if (!in_param)
+		return NS_WALK_STOP;
+
+	/* TODO: Optimize with a hash. */
+	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
+		tmp_if = (struct interface *)rn->info;
+		/* Check oper status of the SVI. */
+		if (!tmp_if || !if_is_operative(tmp_if))
+			continue;
+		zif = tmp_if->info;
+		if (!zif || zif->zif_type != ZEBRA_IF_VLAN
+		    || zif->link != in_param->br_if)
+			continue;
+		vl = (struct zebra_l2info_vlan *)&zif->l2info.vl;
+
+		if (vl->vid == in_param->vid) {
+			if (p_ifp)
+				*p_ifp = tmp_if;
+			return NS_WALK_STOP;
+		}
+	}
+	return NS_WALK_CONTINUE;
+}
+
 /* Map to SVI on bridge corresponding to specified VLAN. This can be one
  * of two cases:
  * (a) In the case of a VLAN-aware bridge, the SVI is a L3 VLAN interface
@@ -791,15 +828,11 @@ static void zl3vni_print_hash_detail(struct hash_bucket *bucket, void *data)
  */
 struct interface *zvni_map_to_svi(vlanid_t vid, struct interface *br_if)
 {
-	struct zebra_ns *zns;
-	struct route_node *rn;
 	struct interface *tmp_if = NULL;
 	struct zebra_if *zif;
 	struct zebra_l2info_bridge *br;
-	struct zebra_l2info_vlan *vl;
-	uint8_t bridge_vlan_aware;
-	int found = 0;
-
+	struct zebra_from_svi_param in_param;
+	struct interface **p_ifp;
 	/* Defensive check, caller expected to invoke only with valid bridge. */
 	if (!br_if)
 		return NULL;
@@ -808,33 +841,19 @@ struct interface *zvni_map_to_svi(vlanid_t vid, struct interface *br_if)
 	zif = br_if->info;
 	assert(zif);
 	br = &zif->l2info.br;
-	bridge_vlan_aware = br->vlan_aware;
-
+	in_param.bridge_vlan_aware = br->vlan_aware;
 	/* Check oper status of the SVI. */
-	if (!bridge_vlan_aware)
+	if (!in_param.bridge_vlan_aware)
 		return if_is_operative(br_if) ? br_if : NULL;
 
+	in_param.vid = vid;
+	in_param.br_if = br_if;
+	in_param.zif = NULL;
+	p_ifp = &tmp_if;
 	/* Identify corresponding VLAN interface. */
-	/* TODO: Optimize with a hash. */
-	zns = zebra_ns_lookup(NS_DEFAULT);
-	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
-		tmp_if = (struct interface *)rn->info;
-		/* Check oper status of the SVI. */
-		if (!tmp_if || !if_is_operative(tmp_if))
-			continue;
-		zif = tmp_if->info;
-		if (!zif || zif->zif_type != ZEBRA_IF_VLAN
-		    || zif->link != br_if)
-			continue;
-		vl = &zif->l2info.vl;
-
-		if (vl->vid == vid) {
-			found = 1;
-			break;
-		}
-	}
-
-	return found ? tmp_if : NULL;
+	ns_walk_func(zvni_map_to_svi_ns, (void *)&in_param,
+		     (void **)p_ifp);
+	return tmp_if;
 }
 
 static int zebra_evpn_vxlan_del(zebra_evpn_t *zevpn)
@@ -846,18 +865,22 @@ static int zebra_evpn_vxlan_del(zebra_evpn_t *zevpn)
 
 	return zebra_evpn_del(zevpn);
 }
-/*
- * Build the VNI hash table by going over the VxLAN interfaces. This
- * is called when EVPN (advertise-all-vni) is enabled.
- */
-static void zevpn_build_hash_table(void)
+
+static int zevpn_build_hash_table_zns(struct ns *ns,
+				     void *param_in __attribute__((unused)),
+				     void **param_out __attribute__((unused)))
 {
-	struct zebra_ns *zns;
+	struct zebra_ns *zns = ns->info;
 	struct route_node *rn;
 	struct interface *ifp;
+	struct zebra_vrf *zvrf;
+
+	zvrf = zebra_vrf_get_evpn();
+
+	if (!zvrf)
+		return NS_WALK_STOP;
 
 	/* Walk VxLAN interfaces and create EVPN hash. */
-	zns = zebra_ns_lookup(NS_DEFAULT);
 	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
 		vni_t vni;
 		zebra_evpn_t *zevpn = NULL;
@@ -874,7 +897,15 @@ static void zevpn_build_hash_table(void)
 
 		vxl = &zif->l2info.vxl;
 		vni = vxl->vni;
-
+		/* link of VXLAN interface should be in zebra_evpn_vrf */
+		if (zvrf->zns->ns_id != vxl->link_nsid) {
+			if (IS_ZEBRA_DEBUG_VXLAN)
+				zlog_debug(
+					"Intf %s(%u) VNI %u, link not in same "
+					"namespace than BGP EVPN core instance ",
+					ifp->name, ifp->ifindex, vni);
+			continue;
+		}
 		/* L3-VNI and L2-VNI are handled seperately */
 		zl3vni = zl3vni_lookup(vni);
 		if (zl3vni) {
@@ -943,7 +974,7 @@ static void zevpn_build_hash_table(void)
 					zlog_debug(
 						"Failed to add EVPN hash, IF %s(%u) L2-VNI %u",
 						ifp->name, ifp->ifindex, vni);
-					return;
+					return NS_WALK_CONTINUE;
 				}
 
 				if (zevpn->local_vtep_ip.s_addr !=
@@ -985,6 +1016,19 @@ static void zevpn_build_hash_table(void)
 			}
 		}
 	}
+	return NS_WALK_CONTINUE;
+}
+
+/*
+ * Build the VNI hash table by going over the VxLAN interfaces. This
+ * is called when EVPN (advertise-all-vni) is enabled.
+ */
+
+static void zevpn_build_hash_table(void)
+{
+	ns_walk_func(zevpn_build_hash_table_zns,
+		     (void *)NULL,
+		     (void **)NULL);
 }
 
 /*
@@ -1617,14 +1661,22 @@ static int zl3vni_del(zebra_l3vni_t *zl3vni)
 	return 0;
 }
 
-struct interface *zl3vni_map_to_vxlan_if(zebra_l3vni_t *zl3vni)
+static int zl3vni_map_to_vxlan_if_ns(struct ns *ns,
+				     void *_zl3vni,
+				     void **_pifp)
 {
-	struct zebra_ns *zns = NULL;
+	struct zebra_ns *zns = ns->info;
+	zebra_l3vni_t *zl3vni = (zebra_l3vni_t *)_zl3vni;
 	struct route_node *rn = NULL;
 	struct interface *ifp = NULL;
+	struct zebra_vrf *zvrf;
+
+	zvrf = zebra_vrf_get_evpn();
+
+	if (!zvrf)
+		return NS_WALK_STOP;
 
 	/* loop through all vxlan-interface */
-	zns = zebra_ns_lookup(NS_DEFAULT);
 	for (rn = route_top(zns->if_table); rn; rn = route_next(rn)) {
 
 		struct zebra_if *zif = NULL;
@@ -1639,13 +1691,39 @@ struct interface *zl3vni_map_to_vxlan_if(zebra_l3vni_t *zl3vni)
 			continue;
 
 		vxl = &zif->l2info.vxl;
-		if (vxl->vni == zl3vni->vni) {
-			zl3vni->local_vtep_ip = vxl->vtep_ip;
-			return ifp;
+		if (vxl->vni != zl3vni->vni)
+			continue;
+
+		/* link of VXLAN interface should be in zebra_evpn_vrf */
+		if (zvrf->zns->ns_id != vxl->link_nsid) {
+			if (IS_ZEBRA_DEBUG_VXLAN)
+				zlog_debug(
+					"Intf %s(%u) VNI %u, link not in same "
+					"namespace than BGP EVPN core instance ",
+					ifp->name, ifp->ifindex, vxl->vni);
+			continue;
 		}
+
+
+		zl3vni->local_vtep_ip = vxl->vtep_ip;
+		if (_pifp)
+			*_pifp = (void *)ifp;
+		return NS_WALK_STOP;
 	}
 
-	return NULL;
+	return NS_WALK_CONTINUE;
+}
+
+struct interface *zl3vni_map_to_vxlan_if(zebra_l3vni_t *zl3vni)
+{
+	struct interface **p_ifp;
+	struct interface *ifp = NULL;
+
+	p_ifp = &ifp;
+
+	ns_walk_func(zl3vni_map_to_vxlan_if_ns,
+		     (void *)zl3vni, (void **)p_ifp);
+	return ifp;
 }
 
 struct interface *zl3vni_map_to_svi_if(zebra_l3vni_t *zl3vni)
@@ -3987,11 +4065,10 @@ int zebra_vxlan_local_mac_add_update(struct interface *ifp,
 		return -1;
 	}
 
-	zvrf = vrf_info_lookup(zevpn->vxlan_if->vrf_id);
+	zvrf = zebra_vrf_get_evpn();
 	if (!zvrf) {
 		if (IS_ZEBRA_DEBUG_VXLAN)
-			zlog_debug("        No Vrf found for vrf_id: %d",
-				   zevpn->vxlan_if->vrf_id);
+			zlog_debug("        No Evpn Global Vrf found");
 		return -1;
 	}
 
@@ -5447,6 +5524,25 @@ stream_failure:
 	return;
 }
 
+static int macfdb_read_ns(struct ns *ns,
+			  void *_in_param __attribute__((unused)),
+			  void **out_param __attribute__((unused)))
+{
+	struct zebra_ns *zns = ns->info;
+
+	macfdb_read(zns);
+	return NS_WALK_CONTINUE;
+}
+
+static int neigh_read_ns(struct ns *ns,
+			 void *_in_param __attribute__((unused)),
+			 void **out_param __attribute__((unused)))
+{
+	struct zebra_ns *zns = ns->info;
+
+	neigh_read(zns);
+	return NS_WALK_CONTINUE;
+}
 
 /*
  * Handle message from client to learn (or stop learning) about VNIs and MACs.
@@ -5499,10 +5595,10 @@ void zebra_vxlan_advertise_all_vni(ZAPI_HANDLER_ARGS)
 			     zebra_evpn_gw_macip_add_for_evpn_hash, NULL);
 
 		/* Read the MAC FDB */
-		macfdb_read(zvrf->zns);
+		ns_walk_func(macfdb_read_ns, NULL, NULL);
 
 		/* Read neighbors */
-		neigh_read(zvrf->zns);
+		ns_walk_func(neigh_read_ns, NULL, NULL);
 	} else {
 		/* Cleanup VTEPs for all EVPNs - uninstall from
 		 * kernel and free entries.