LCPE common mistakes
LCPE · Understand the traps and how to avoid them.
When asked which route a router will use, the candidate picks the route with the lowest administrative distance instead of comparing prefix length first, so a /24 learned via OSPF is chosen over a /16 static route.
Studying AD and metric as a single ranking list is easier to memorize than the two-stage process, and many practice scenarios only show routes of equal prefix length, so the first stage never gets exercised. Candidates also confuse 'longest match' with 'best metric'.
What to do instead
Drill the decision order explicitly: (1) longest prefix match wins regardless of AD or metric; (2) only among routes for the same prefix do you compare AD; (3) only among routes from the same protocol do you compare metric. Build a 10-row table of mixed prefixes and protocols and justify each selection in writing before checking the answer.
The candidate computes subnet boundaries by hand and misplaces the block size, for example treating a /27 as 64 addresses or putting 192.168.1.96/27 in the wrong subnet, and then answers the VLSM or summarization question from that bad boundary.
Rote memorization of the mask table without a working method breaks under time pressure, and candidates count host bits starting from the wrong end or forget to subtract the network and broadcast addresses when the question asks for usable hosts.
What to do instead
Use one consistent method every time: block size = 256 minus the interesting octet of the mask; subnets step by that block size; usable hosts = 2^host bits minus 2. Practice 20 random prefix-length conversions until each takes under 15 seconds, and always write the binary boundary for the octet in question rather than doing it mentally.
On switching questions the candidate assumes a port in a VLAN is automatically trunking, or forgets that a trunk carries the native VLAN untagged, and so predicts the wrong frame tagging or the wrong VLAN for untagged traffic arriving on an 802.1Q link.
Access and trunk behaviour is taught as configuration syntax rather than as frame-forwarding behaviour, and the native VLAN detail is a single line in most study material, so it is skipped. Candidates also assume VLAN 1 is always the native VLAN after it has been changed.
What to do instead
Trace frames explicitly: for each ingress port state whether the frame arrives tagged or untagged, which VLAN it is assigned to, and whether it leaves the egress trunk tagged. Do this for the native VLAN, a non-native VLAN, and an access port on the same switch. Confirm with 'show interfaces trunk' output that you can read the native VLAN and the allowed list.
The candidate reads a spanning-tree topology and identifies the root bridge or the blocked port by physical cabling rather than by comparing bridge IDs and path costs, so the port they expect to forward is actually the alternate port.
Root election and port-role election are multi-step and easy to short-circuit: candidates compare MAC addresses before priorities, or forget that the root path cost is cumulative and that the lowest sender bridge ID breaks ties on a shared segment.
What to do instead
Work the four-step sequence in order for every topology: elect root by lowest priority then lowest MAC; on each non-root switch elect one root port by lowest root path cost, then lowest neighbor bridge ID, then lowest port ID; on each segment elect one designated port by lowest path cost to root; everything else is non-designated. Redraw the topology with costs annotated before answering.
When a question describes a host that cannot reach a server, the candidate jumps to a routing-protocol conclusion instead of walking the layered path, and ignores that the default gateway is missing, the ARP entry is incomplete, or the return path is asymmetric.
Troubleshooting questions reward pattern recognition, and candidates pattern-match on the most recently studied topic. They also treat 'ping fails' as one symptom rather than separating name resolution, local subnet reachability, default gateway reachability, and remote reachability.
What to do instead
Apply a fixed checklist: link state, then IP address and mask correctness, then same-subnet ping, then default gateway ping, then remote IP ping, then name resolution. For each question, state which layer the symptom isolates before choosing an answer, and explicitly check whether the described traffic is one-way.
The candidate configures or interprets OSPF without accounting for the network type and DR/BDR election, and assumes two neighbors on the same segment will always form a full adjacency with each other.
Neighbor-state diagrams are memorized as a linear list, so the two-way state between two DROTHERs looks like a failure rather than normal behaviour. Broadcast versus point-to-point network types and the effect of the priority field are treated as trivia.
What to do instead
Learn the states as a progression with the trigger for each transition, and separately learn which pairs reach FULL on a broadcast segment: DR and BDR with every DROTHER, but DROTHER-to-DROTHER stays 2-WAY. Practice predicting the DR from priority then router ID, including the case where all priorities are equal and the highest router ID wins.
The candidate answers NAT questions using the private addresses on the inside instead of the translated addresses, or mixes up inside local, inside global, outside local, and outside global when reading a translation table.
The four NAT terms are defined relative to the router, not to the direction of traffic, and candidates learn them as synonyms for 'private' and 'public', which fails whenever the outside network is also privately addressed.
What to do instead
Anchor each term to a question: whose address is it, and from which device's perspective? Inside local is the inside host as seen inside; inside global is that host as seen outside. Rebuild a translation table for a sample topology with both sides privately addressed and label all four columns before answering.
On first-hop redundancy and DHCP questions the candidate assumes the virtual IP and the physical interface addresses behave identically, or forgets that a relay agent is needed when the DHCP server is on another subnet.
HSRP, VRRP, and GLBP are studied as configuration templates, so the behaviour of the virtual MAC, the priority and preemption rules, and the tracking object are not connected to forwarding. DHCP broadcast behaviour is likewise memorized without the relay step.
What to do instead
For each protocol, write down the virtual MAC format, the default priority, whether preemption is on by default, and what happens on the failed device's recovery. For DHCP, trace the DORA exchange and mark exactly where a broadcast is stopped by the router and replaced by a unicast from the relay's interface address.
The candidate treats port security, DHCP snooping, and dynamic ARP inspection as independent features and cannot say which one requires which other, or what happens to a frame when a violation mode is triggered.
Layer 2 security features are taught one at a time with their own configuration, and the dependencies between them, such as DAI relying on the DHCP snooping binding table, are mentioned only in passing. Violation modes are memorized as three words without their differing actions.
What to do instead
Build a dependency map: port security binds MACs to ports; DHCP snooping builds a binding table of MAC, IP, lease, VLAN, and port; DAI consults that table to validate ARP. Then state the exact action for protect, restrict, and shutdown, including whether traffic is dropped, whether a log or counter increments, and whether the port is err-disabled.
On automation questions the candidate assumes a RESTCONF or NETCONF call returns the same structure as the CLI show command, or assumes model-driven telemetry and SNMP polling produce equivalent data at equivalent intervals.
The programmability domain is often studied last and treated as vocabulary, so the data encoding, the transport, and the difference between a pull model and a push model are not tied to concrete request and response examples.
What to do instead
Compare the mechanisms in a table with columns for transport, encoding, and data model: SNMP uses UDP with MIB OIDs and polling; NETCONF uses SSH with XML and YANG; RESTCONF uses HTTPS with JSON or XML and YANG. Then write one sample request and the shape of its response for each, and note that telemetry pushes on a configured subscription rather than waiting to be polled.