Aruba Comware Advanced study materials
Aruba Comware Advanced study resources

Aruba Comware Advanced common mistakes

Aruba Comware Advanced · Understand the traps and how to avoid them.

Mistake 1

Applying IRF port bindings to physical interfaces before putting the member ports into a shutdown state, then wondering why the IRF fabric fails to form or flaps.

Candidates memorize the 'bind IRF port' command order but forget the Comware requirement that physical ports be shut down before being bound into an IRF port and only un-shut after both members are configured.

What to do instead

Memorize the sequence: shut down the physical ports, configure irf-port 1/1 or 2/1 on each member, bind the ports, save config, activate irf-port-configuration, then undo shutdown. Verify with display irf and display irf topology.

Mistake 2

Treating IRF as if it were a single control plane with no MAD protection, and failing to configure any multi-active detection method.

Candidates conflate IRF with stacking and assume split-brain is impossible or harmless; they skip BFD/LACP MAD because it looks optional.

What to do instead

Know that a split IRF produces two active fabrics with duplicate IPs and that MAD (LACP MAD, BFD MAD, or ARP MAD) is required. Be able to describe which MAD type works with which peer device and how the failed fabric recovers.

Mistake 3

Configuring an ACL but applying it in the wrong direction or on the wrong object type, and expecting traffic to be filtered.

Candidates learn ACL syntax but not the Comware model that ACLs are only enforced when attached to an interface, a user line, a QoS classifier, or a service; an unattached ACL does nothing.

What to do instead

For each scenario, state the attachment point explicitly (packet-filter on an interface, in/out direction, or classifier behavior). Remember ACL rules are matched top-down with an implicit deny only when the last rule is deny, and verify with display packet-filter.

Mistake 4

Mixing up the order of operations in OSPF/IS-IS area or level design and expecting a non-backbone area or Level-1 router to carry transit traffic.

Candidates remember 'areas must touch Area 0' and 'Level-1 must reach Level-2' as trivia but cannot apply it to a topology question about virtual links or route leaking.

What to do instead

Practice tracing which router originates which LSA/LSP and where summarization or leaking is needed. For OSPF, know when a virtual link is required; for IS-IS, know how route leaking from Level-2 to Level-1 is configured and why.

Mistake 5

Configuring BGP neighbor relationships without matching the update-source, peer AS, and authentication parameters, then blaming the routing protocol for an Idle state.

Candidates focus on route policy and forget that most BGP failures are session-level: wrong source address on loopback peering, AS mismatch, or missing MD5.

What to do instead

Systematically check: display bgp peer verbose shows state and last error. Confirm update-source loopback plus ebgp-max-hop for loopback peering, matching peer as-number, and matching authentication. Then move to policy.

Mistake 6

Deploying multicast and forgetting that PIM requires an RP (static or BSR) and that IGMP snooping must be enabled on the VLAN to constrain traffic.

Candidates treat multicast as 'enable PIM and it works' and miss the rendezvous point, RPF check direction, and Layer 2 snooping components.

What to do instead

For any multicast scenario, verify: PIM mode per interface, RP configuration (static-rp or BSR), RPF interface toward the source, and igmp-snooping on the access VLAN. Use display pim rp-info and display igmp-snooping group to confirm.

Mistake 7

Configuring IPv6 addresses and routing but forgetting that ND, RA, and DHCPv6 have distinct roles, and assuming SLAAC works without RA.

Candidates map IPv4 habits onto IPv6 and skip the neighbor discovery and router advertisement steps that make autoconfiguration function.

What to do instead

Know that SLAAC needs RA with the A flag, DHCPv6 needs the M/O flags, and ND handles address resolution and duplicate address detection. Be ready to verify with display ipv6 neighbors and display ipv6 interface.

Mistake 8

Building an MPLS L3VPN and forgetting to enable MPLS and LDP on every transit interface, or mismatching the VPN instance RD/RT.

Candidates learn the control-plane concepts (MP-BGP, VPNv4) but underestimate the requirement that the underlay label path be present hop by hop, and that RT import/export must match between PEs.

What to do instead

Verify the underlay first: mpls enable and mpls ldp enable on all core links, display mpls ldp peer and display mpls lsp. Then verify VPN instance route targets with display ip vpn-instance and check that import matches the remote export.

Mistake 9

Configuring QoS with a classifier and behavior but forgetting to bind them into a policy and apply the policy to the interface in the correct direction.

Candidates stop at 'create class and behavior' and miss the three-part Comware QoS model (traffic class, traffic behavior, qos policy) and the interface application step.

What to do instead

Always complete the chain: traffic classifier, traffic behavior, qos policy associating them, then qos apply policy on the interface with in or out. Confirm with display qos policy interface.

Mistake 10

Diagnosing a spanning tree or link aggregation problem without checking the operational state of the member ports and the aggregation mode on both sides.

Candidates jump to STP timers and root bridge questions when the real issue is a LACP mode mismatch or a member port that is down or in a different VLAN, causing the aggregate to be inactive.

What to do instead

Check display link-aggregation summary and verbose for member port status and LACP state, confirm both sides use the same mode (static vs dynamic LACP), and confirm member ports share the same configuration before touching STP.