[voq] Test plan and testcase updates for multiasic (#3463)

What is the motivation for this PR?
To run the init, neighbor and IP forwarding tests on chassis with multiasic linecards.

How did you do it?
Scripts in tests/voq were enhanced to handle multiasic cards:

All test cases were changed so that ASICs with no neighbors or router interfaces will be skipped.
The HIDE_INTERNAL route map is let in place and tests in ipfwd script that are affected by it are removed.
CLI and database checks where enhanced to look in the correct namespace on multiasic cards.
Other enhancements:

Neighbor script cleanup was enhanced be more robust when restoring configuration.
PTF port selection was centralized and now uses minigraph.
TTL checks were made optional for running ipfwd script on VS testbed.
Test plan was updated with these changes.

How did you verify/test it?
Ran test suite on VoQ chassis.

Co-authored-by: Tom Custodio <[email protected]>

Author: tcusto

ansible/roles/test/files/ptftests/voq.py

@@ -257,7 +257,7 @@ def runTest(self):
         print("Found %s ICMP ttl expired packets on ports: %s" % (result, str(src_rx_ports)))
         logger.info("Found %s ICMP ttl expired packets on ports: %s" % (result, str(src_rx_ports)))
         print("port: %s" % result.port)
-        if str(result.port) not in src_rx_ports:
+        if result.port not in src_rx_ports:
             self.fail("Port %s not in %s" % (result.port, src_rx_ports))
 
         verify_no_packet_any(self, send_pkt, dst_rx_ports)
@@ -289,6 +289,7 @@ def setUp(self):
         self.src_ptf_port_list = self.test_params.get('src_ptf_port_list')
         self.dst_ptf_port_list = self.test_params.get('dst_ptf_port_list')
         self.version = self.test_params.get('version')
+        self.ignore_ttl = self.test_params.get('ignore_ttl')
 
     def check_icmp_mtu(self):
         """Check ICMP/Ping to DUT works for MAX MTU. """
@@ -386,6 +387,9 @@ def check_ip_mtu(self):
 
             masked_exp_pkt = Mask(exp_pkt)
             masked_exp_pkt.set_do_not_care_scapy(scapy.Ether, "dst")
+            if self.ignore_ttl:
+                masked_exp_pkt.set_do_not_care_scapy(scapy.IP, "ttl")
+                masked_exp_pkt.set_do_not_care_scapy(scapy.IP, "chksum")
 
         else:
             pkt = simple_ipv6ip_packet(pktlen=self.pktlen,
@@ -403,12 +407,15 @@ def check_ip_mtu(self):
 
             masked_exp_pkt = Mask(exp_pkt)
             masked_exp_pkt.set_do_not_care_scapy(scapy.Ether, "dst")
+            if self.ignore_ttl:
+                masked_exp_pkt.set_do_not_care_scapy(scapy.IPv6, "hlim")
+                masked_exp_pkt.set_do_not_care_scapy(scapy.IPv6, "chksum")
 
         src_port = self.src_ptf_port_list[0]
         send_packet(self, src_port, pkt)
-        logging.info("Sending packet from port " + str(src_port) + " to " + ip_dst)
 
         dst_port_list = self.dst_ptf_port_list
+        logging.info("Sending packet from port " + str(src_port) + " to " + ip_dst + " expected ports " + str(dst_port_list))
         (matched_index, received) = verify_packet_any_port(self, masked_exp_pkt, dst_port_list)
 
         assert received

docs/testplan/Distributed-VoQ-Arch-test-plan.md

@@ -378,40 +378,7 @@ will allow the tests to validate the various table deletes before the entries ar
 
 ### Test cases
 
-#### Test Case 1. Neighbor established from a remote card.
-
-##### Test Objective
-Verify local neighbor behavior is correct when ARP/NDP request is triggered by a packet from a remote line card.
-
-##### Test Steps
-* Send ping to from linecard 1 to a VM attached to linecard 2.  This will establish a local neighbor on linecard 2 and a remote neighbor on linecard 1.
-
-* On linecard 2:
-    * Verify ASIC DB entries on local linecard.
-        * MAC address matches MAC of neighbor VM.
-        * Router interface OID matches back to the correct interface and port the neighbor was learned on.        
-    * Verify on local linecard CLI, show arp/ndp, ip neigh commands.
-        * MAC address matches MAC of neighbor VM.
-    * Verify neighbor table in linecard appDB.
-        * MAC address matches MAC of neighbor VM.
-
-* On supervisor card:
-    * Verify SYSTEM_NEIGH table in Chassis AppDB (redis-dump -h <ip> -p 6380 -d 12 on supervisor).
-        * Verify encap index and MAC address match between ASICDB the Chassis AppDB
-
-* On linecard 1:
-    * Verify ASIC DB entries as a remote neighbor.
-        * Verify impose index=True in ASIC DB. 
-        * Verify MAC address in ASIC DB is the remote neighbor mac.
-        * Verify encap index for ASIC DB entry matches Chassis App DB.
-        * Verify router interface OID matches the interface the neighbor was learned on.
-    * Verify on linecard CLI, show arp/ndp, ip neigh commands.
-        * For inband port, MAC should be inband port mac in kernel table and LC appDb.
-        * For inband vlan mode, MAC will be remote ASIC mac in kernel table and LC appdb.
-    * Verify neighbor table in linecard appdb.
-    * Verify static route in kernel with /32 for entry.
-
-#### Test Case 2. Clear ARP, single address.
+#### Test Case 1. Clear ARP, single address.
 ##### Test Objective
 Verify tables, databases, and kernel routes are correctly deleted when a single neighbor adjacency is cleared.
 ##### Test Steps
@@ -428,7 +395,7 @@ Verify tables, databases, and kernel routes are correctly deleted when a single
 * Repeat with IPv4, IPv6, dual-stack.
 
 
-#### Test Case 3. Clear ARP table via sonic command.
+#### Test Case 2. Clear ARP table via sonic command.
 ##### Test Objective
 Verify tables, databases, and kernel routes are correctly deleted when the entire neighbor table is cleared.
 ##### Test Steps
@@ -446,41 +413,24 @@ Verify tables, databases, and kernel routes are correctly deleted when the entir
 * Repeat with IPv4, IPv6, dual-stack.
 
 
-#### Test Case 4. Front panel port link flap
+#### Test Case 3. Front panel port link flap
 ##### Test Objective
-Verify tables, databases, and kernel routes are correctly deleted when the front panel port flaps.
+Verify tables, databases, and kernel routes are unaffected when the front panel port flaps and restores.
 ##### Test Steps
 * Admin down interface on fanout to cause LOS on DUT.
 * On local linecard:
-    * Verify ARP/NDP entries are removed from CLI for neighbors on down port.
-    * Verify table entries in ASIC, AppDb are removed for addresses on down port.
+    * Verify local interface is down, verify ARP is still present in local database.
 * On Supervisor card:
-    * Verify Chassis App DB entry are removed for only the cleared address.  Entries for addresses on other line cards
-    should still be present.
+    * Verify Chassis App DB entry are still present.
 * On remote linecards:
-    * Verify table entries in ASICDB, APPDB, and host ARP table are removed for cleared addresses.
-    * Verify kernel routes for cleared address are deleted.
+    * Verify table entries in ASICDB, APPDB, and host ARP table are present for affected addresses.
+    * Verify kernel routes are present for affected addresses.
 * Admin interface up, verify recreation after restarting traffic.
+* Verify ARP entries on linecards and supervisors are still correct.
 * Repeat with IPv4, IPv6, dual-stack.
 
 
-#### Test Case 5. Gratuitous ARP - Previously Known IP
-##### Test Objective
-Verify tables, databases, and kernel routes are correctly updated when receiving a unsolicited ARP packet for a previously known IP address..
-##### Test Steps
-* Clear ARP table on a line card.
-* Send unsolicited ARP packet into linecard for an IP that was known on that card and is now stale.
-* On local linecard:
-    * Verify table entries in local ASIC, APP, host arp table are recreated.
-* On supervisor card:
-    * Verify Chassis App DB entry is correct for the relearned address.  
-* On remote linecards:
-    * Verify table entries in remote hosts/ASICs in ASICDB, APPDB, and host ARP table are recreated.
-    * Verify kernel routes in remote hosts are still present.
-* Verify that packets can be sent from local and remote linecards to learned address.
-* Repeat with IPv4, IPv6, dual-stack.
-
-#### Test Case 6. Gratuitous ARP - Known IP - Mac change
+#### Test Case 4. Gratuitous ARP - Known IP - Mac change
 ##### Test Objective
 Verify tables, databases, and kernel routes are correctly updated when a unsolicited ARP packet changes the MAC address of learned neighbor.
 ##### Test Steps
@@ -497,7 +447,7 @@ Verify tables, databases, and kernel routes are correctly updated when a unsolic
 * Verify that packets can be sent from local and remote linecards to learned address.
 * Repeat with IPv4, IPv6, dual-stack.
 
-#### Test Case 7. ARP Request/Reply - Mac change
+#### Test Case 5. ARP Request/Reply - Mac change
 ##### Test Objective
 Verify tables, databases, and kernel routes are correctly updated when the MAC address of a neighbor changes and is updated via request/reply exchange.
 ##### Test Steps
@@ -514,7 +464,7 @@ Verify tables, databases, and kernel routes are correctly updated when the MAC a
 * Verify that packets can be sent from local and remote linecards to the learned address.
 * Repeat with IPv4, IPv6, dual-stack.
 
-#### Test Case 8. Disruptive Events
+#### Test Case 6. Disruptive Events
 ##### Test Objective
 Verify port, router interface, and neighbor recovery after disruptive events.
 ##### Test Steps
@@ -570,6 +520,9 @@ Please reference the [T2 topology](https://github.com/Azure/sonic-mgmt/pull/2638
 
 VMs attached to line card 1 and line card 2 will be used for this test.
 DUT Port A&B are on line card 1, D is on line card 2.
+
+The HIDE_INTERNAL route policy will prevent inband and interface address from being advertised to EBGP peers. Looback addresses will be used to test traffic flows across cards and VMs.
+
 ```
                       ---------- DUT ----------    
                       |--- LC1 ---|--- LC2 ---|
@@ -637,7 +590,6 @@ Verify the kernel route table is correct based on the topology.
 * Verify routes for local inband interfaces are directly connected.
 * Verify BGP established between line cards.
 * Verify routes of remote linecard inband interfaces are connected via local linecard inband interface.
-* Verify IP interface addresses on remote network ports have a next hop of their inband IP. On linecard 1, route 10.0.0.64/31 next hop is 133.133.133.5. 
 * Verify all learned prefixes from neighbors have their neighbors as next hop.
 * Repeat for IPv4 only, IPv6 only, dual-stack.
 
@@ -647,9 +599,6 @@ Verify Host IP forwarding for IPv4 and IPv6 for various packet sizes and ttls to
 ##### Test Steps
 * On linecard 1, send ping from:
     * DUT IP interface A to DUT IP Interface B. (10.0.0.0 to 10.0.0.2) 
-    * DUT IP interface A to DUT IP Interface D. (10.0.0.0 to 10.0.0.64)
-* On linecard 2, send ping from:
-    * DUT IP interface D to DUT IP Interface A.
 * Repeat for TTL 0,1,2,255
 * Repeat for 64, 1500, 9100B packets
 * Repeat for IPv6
@@ -660,13 +609,8 @@ Verify Host IP forwarding for IPv4 and IPv6 for various packet sizes and ttls to
 ##### Test Steps
 * On linecard 1, send ping from:
     * DUT IP Interface on port A to directly connected neighbor address. (10.0.0.0 to 10.0.0.1)
-    * DUT IP Interface A to neighbor address on port B. (10.0.0.0 to 10.0.0.3)
-    * DUT IP Interface A to neighbor address on port D. (10.0.0.0 to 10.0.0.65)
-* On linecard 2, send ping from:
-    * DUT IP interface D to neighbor address on port A. (10.0.0.64 to 10.0.0.1)
 * On Router 01T3, send ping from:
     * Router IP interface to DUT address on port A. (10.0.0.1 to 10.0.0.0)
-    * Router IP interface to DUT address on port D. (10.0.0.1 to 10.0.0.64)
 * Repeat for TTL 0,1,2,255
 * Repeat for 64, 1500, 9100B packets
 * Repeat for IPv6
@@ -677,13 +621,8 @@ Verify Host IP forwarding for IPv4 and IPv6 for various packet sizes and ttls to
 ##### Test Steps
 * On linecard 1, send ping from:
     * DUT IP Interface A to routed loopback address from router 01T3. (10.0.0.0 to 100.1.0.1)
-    * DUT IP Interface A to routed loopback address from router 02T3. (10.0.0.0 to 100.1.0.2)
-    * DUT IP Interface A to routed loopback address from router 01T1. (10.0.0.0 to 100.1.0.33)
-* On linecard 2, send ping from:
-    * DUT IP interface D to routed loopback address from router 01T3. (200.0.0.1 to 100.1.0.1)
 * On Router 01T3, send ping from:
     * Router loopback interface to DUT address on port A. (100.1.0.1 to 10.0.0.0)
-    * Router loopback interface to DUT address on port D. (100.1.0.1 to 10.0.0.64)
 * Repeat for TTL 0,1,2,255
 * Repeat for 64, 1500, 9100B packets
 * Repeat for IPv6
@@ -694,21 +633,7 @@ Verify IP connectivity over inband interfaces.
 ##### Test Steps
 * On linecard 1 send ping from:
     * Inband interface F0 to inband interface F1 (133.133.133.1 to 133.133.133.5)
-    * Inband interface F0 to interface D (133.133.133.1 to 10.0.0.64)
     * Inband interface F0 to neighbor on port A (133.133.133.1 to 10.0.0.1)
-    * Inband interface F0 to neighbor on port D (133.133.133.1 to 10.0.0.65)
-    * Inband interface F0 to routed loopback from router 01T3 (133.133.133.1 to 100.1.0.1)
-    * Inband interface F0 to routed loopback from router 01T1 (133.133.133.1 to 100.1.0.33)
-* On linecard 2, send ping from:
-    * Inband interface F1 to inband interface F0 (133.133.133.5 to 133.133.133.1)
-    * Inband interface F1 to interface D (133.133.133.5 to 10.0.0.64)
-    * Inband interface F1 to neighbor on port A (133.133.133.5 to 10.0.0.1)
-    * Inband interface F1 to neighbor on port D (133.133.133.5 to 10.0.0.65)
-    * Inband interface F1 to routed loopback from router 01T3 (133.133.133.5 to 100.1.0.1)
-    * Inband interface F1 to routed loopback from router 01T1 (133.133.133.5 to 100.1.0.33)
-* On Router 01T3, send ping from:
-    * Router loopback interface to DUT inband address on linecard 1. (100.1.0.1 to 133.133.133.1)
-    * Router loopback interface to DUT inband address on linecard 2. (100.1.0.1 to 133.133.133.5)
 * Repeat for TTL 0,1,2,255
 * Repeat for 64, 1500, 9100B packets
 * Repeat for IPv6

tests/common/helpers/redis.py

@@ -227,6 +227,7 @@ def __init__(self, host):
         self.hostif_table = []
         self.system_port_key_list = []
         self.port_key_list = []
+        self.lagid_key_list = []
 
     def get_switch_key(self):
         """Returns a list of keys in the switch table"""
@@ -257,8 +258,11 @@ def get_hostif_list(self):
         cmd = self._cli_prefix() + "KEYS %s:*" % AsicDbCli.ASIC_HOSTIF_TABLE
         return self._run_and_raise(cmd)["stdout_lines"]
 
-    def get_asic_db_lag_list(self):
+    def get_asic_db_lag_list(self, refresh=False):
         """Returns a list of keys in the lag table"""
+        if self.lagid_key_list != [] and refresh is False:
+            return self.lagid_key_list
+
         cmd = self._cli_prefix() + "KEYS %s:*" % AsicDbCli.ASIC_LAG_TABLE
         return self._run_and_raise(cmd)["stdout_lines"]
 
@@ -309,9 +313,10 @@ def get_neighbor_value(self, neighbor_key, field):
         """
         cmd = ["/usr/bin/redis-cli", "-n", "1", "HGET", neighbor_key, field]
         if self.host.namespace is not None:
-            cmd = ["sudo", "ip", "netns", "exec"] + cmd
+            cmd = ["/usr/bin/redis-cli", "-h", str(self.ip), "-n", "1", "HGET", neighbor_key, field]
+            cmd = ["sudo", "ip", "netns", "exec"] + [self.host.namespace] + cmd
+
         result = self.host.sonichost.shell(argv=cmd)
-        logger.debug("neigh result: %s", result['stdout'])
         return result['stdout']
 
     def get_hostif_table(self, refresh=False):
@@ -390,21 +395,26 @@ def get_rif_porttype(self, portid, refresh=False):
         Returns:
             "hostif" if the port ID has a host interface
             "sysport" if it is a system port.
-            "port" if the port ID is in local port table but has no hostif
+            "port" if the port ID is in local port table but has no hostif (inband)
+            "lag" if the portid is a portchannel.
             "other" if it is not found in any port table
         """
-        # could be a localport
 
         port_key_list = self.get_port_key_list(refresh=refresh)
         system_port_keylist = self.get_system_port_key_list(refresh=refresh)
+        lag_keylist = self.get_asic_db_lag_list(refresh=refresh)
 
+        # could be a frontpanel port
         if "%s:%s" % (
                 AsicDbCli.ASIC_PORT_TABLE,
                 portid) in port_key_list and portid in self.get_hostif_portid_oidlist():
             return "hostif"
         # could be a system port
         elif "%s:%s" % (AsicDbCli.ASIC_SYSPORT_TABLE, portid) in system_port_keylist:
             return "sysport"
+        # could be a lag
+        elif "%s:%s" % (AsicDbCli.ASIC_LAG_TABLE, portid) in lag_keylist:
+            return "lag"
         # could be something else
         elif "%s:%s" % (AsicDbCli.ASIC_PORT_TABLE, portid) in port_key_list:
             return "port"

tests/voq/conftest.py

@@ -38,24 +38,6 @@ def all_cfg_facts(duthosts):
     return results
 
 
-@pytest.fixture(scope="module", autouse=True)
-def bgp_redistribute_route_lo(duthosts, all_cfg_facts):
-    for a_host in duthosts.frontend_nodes:
-        for a_asic in a_host.asics:
-            asic_asn = all_cfg_facts[a_host.hostname][a_asic.asic_index]['ansible_facts']['DEVICE_METADATA']['localhost']['bgp_asn']
-
-            send_command = a_asic.get_docker_cmd(
-                "vtysh -c 'configure terminal' -c 'router bgp " + asic_asn + "' -c 'address-family ipv4 unicast' -c 'no redistribute connected route-map HIDE_INTERNAL' -c 'redistribute connected'",
-                "bgp")
-
-            send_command_ipv6 = a_asic.get_docker_cmd(
-                "vtysh -c 'configure terminal' -c 'router bgp " + asic_asn + "' -c 'address-family ipv6 unicast' -c 'no redistribute connected route-map HIDE_INTERNAL' -c 'redistribute connected'",
-                "bgp")
-
-            a_host.command(send_command)
-            a_host.command(send_command_ipv6)
-
-
 @reset_ansible_local_tmp
 def _get_nbr_macs(nbrhosts, node=None, results=None):
     vm = nbrhosts[node]