Cluster connectivity fails after RCF C3232Q cluster switch update
Applies to
- Cisco NX3232C Cluster Network Switch (CNS)
- Cisco NX3132 Cluster Network Switch (CNS)
- NX-OS firmware update
- RCF firmware update
Issue
- Migrating cluster LIFs to the updated switch, the
cluster ping-cluster
shows broken paths. Example for 4 nodes:
::> set advanced
::*> cluster ping-cluster -node node_name-2
Host is node_name-2
Getting addresses from network interface table...
Cluster node_name-2_clus1 169.254.97.141 node_name-2 e4a
Cluster node_name-2_clus2 169.254.119.137 node_name-2 e4e
...
Basic connectivity succeeds on 6 path(s)
Basic connectivity fails on 6 path(s)
...
Larger than PMTU communication succeeds on 6 path(s)
RPC status:
6 paths up, 0 paths down (tcp check)
6 paths up, 0 paths down (udp check)
- Each node reports an error similar to:
[node_name-1: vifmgr: vifmgr.cluscheck.droppedall:debug]: Total packet loss when pinging from cluster lif node_name-1_clus2 (node node_name-1) to cluster lif node_name-2_clus1 (node node_name-2).
[node_name-1: vifmgr: callhome.clus.net.degraded:debug]: Call home for CLUSTER NETWORK DEGRADED: Total Packet Loss - Ping failures detected between node_name-1_clus1 ( 169.254.243.146 ) on node_name-1 and node_name-2_clus2 ( 169.254.43.139 ) on node_name-2
cluster ping-cluster
fails when ports ping others connected to a different switch.cluster ping-cluster
are ok when ports ping others connected to the same switch.- ISL ports in eth1/31-32 report properly and transmit data in the
# show interface eth1/31-32
output