Increase threshold in TestHAL3Agent fullstack test

The HA router failover fullstack test has a maximum threshold of ICMP
packets lost after a failover has been executed. Based on the number of
failures observed in zuul recently, this threshold seems to be set at a
value too low. In this change, an increase in this value is proposed.

Change-Id: I4555ed821865db724c4e5fc502648fe71d5febea
Partial-Bug: #2104019
Partial-Bug: #2091351
(cherry picked from commit 557d57e)
Signed-off-by: Yatin Karel <[email protected]>

Author: Miguel Lavalle

neutron/tests/fullstack/test_l3_agent.py

@@ -566,8 +566,9 @@ def is_one_host_active_for_router():
         netcat_udp.stop_processes()
 
         # With the default advert_int of 2s the keepalived master timeout is
-        # about 6s. Assert less than 90 lost packets (9 seconds)
-        threshold = 90
+        # about 6s. Assert less than 90 lost packets (9 seconds) plus 30 to
+        # account for CI infrastructure variability
+        threshold = 120
 
         lost = pinger.sent - pinger.received
         message = (f'Sent {pinger.sent} packets, received {pinger.received} '