Network Infrastructure and Connectivity practice questions

Autonegotiation against a hard-set peer: speed is sensed, duplex falls back to half at 10/100 Mb/s.

Two facts combine. First, a port with speed and duplex hard-coded stops sending autonegotiation advertisements, so the autonegotiating NIC uses parallel detection: it senses 100 Mb/s from the signal, but it cannot learn the duplex and, per IEEE 802.3, defaults to half duplex at 10 or 100 Mb/s. Second, the resulting duplex mismatch shows CRC errors and runts on the full-duplex end (the switch) and late collisions on the half-duplex end (the NIC). The tempting near-twin puts the late collisions on the wrong end.

1. Recognise that forcing speed and duplex disables autonegotiation on the switch port.
2. Recall parallel detection: speed is sensed, duplex is not.
3. Apply the fallback rule: 10/100 Mb/s falls back to half duplex.
4. Place the counters: CRC/runts on the full-duplex end, late collisions on the half-duplex end.

Exam tip: Hard-set one end and autonegotiate the other, and you build a duplex mismatch: set both to auto or both to the same fixed values.

When one end must stay hard-coded, the other end must be hard-coded to the same speed and duplex.

Two facts combine. The 'a-half a-100' output shows the port autonegotiated but could only sense the speed: a peer with autonegotiation disabled sends no advertisement, so the port detected 100 Mb/s and fell back to half duplex. Late collisions are logged by the half-duplex end of a duplex mismatch, so the appliance must be running full duplex. Because the appliance cannot be changed, the fix is to force the switch port to 100 Mb/s full duplex. Re-enabling auto changes nothing, forcing half duplex keeps the mismatch, and forcing 1000 Mb/s takes the link down.

1. Read 'a-half a-100': the speed was sensed, half duplex is the fallback.
2. Read the late collisions: this end is the half-duplex side of a duplex mismatch, so the appliance is full duplex.
3. Match the switch port to the fixed peer exactly: 100 Mb/s full duplex.
4. Reject options that keep half duplex or change the speed.

Exam tip: Both ends auto, or both ends forced to identical values; never a mix.

Speed mismatch and duplex mismatch fail differently.

Ethernet ends must use the same signalling rate to establish a link. When both ends are forced to different speeds there is no negotiation to rescue them, so the physical link never comes up and the interface shows down/down. Disabling autonegotiation is fine when both ends are forced to the same values, and a router-to-switch link uses a straight-through cable whether or not Auto-MDIX is active. 'Administratively down' would indicate a shutdown.

1. Note that the cable worked before the change.
2. Read 'down down': the physical layer is not up.
3. Compare the forced speeds: 10 versus 100.
4. Conclude speed mismatch; set matching speeds or autonegotiate both ends.

Exam tip: Mismatched speed: link down. Mismatched duplex: link up with errors.

Interface status 'administratively down' is a configuration state, not a physical fault.

'Administratively down' appears only when the interface has been disabled with the shutdown command, which Cisco routers apply to physical interfaces by default. No physical or speed/duplex problem can produce that wording, so the first step is no shutdown; only if the interface then shows down/down should the cabling be investigated.

1. Read the Status column: administratively down.
2. Recall that only shutdown produces that state.
3. Enable the interface with no shutdown.
4. Re-check the status before touching the cable.

Exam tip: 'Administratively down' always means shutdown is configured.

'Down, line protocol down (notconnect)' means no physical link signal is received.

notconnect is the Catalyst status for a port that sees no link partner: no cable, a broken cable, a far end that is shut down or powered off, or a speed mismatch. Both a disconnected cable and a shut-down router interface remove the link signal. Duplex, IP subnet and native VLAN problems all happen on a link that is already up, so they show as errors or log messages, not as down/down.

1. Read 'down, line protocol is down (notconnect)': no physical link.
2. List Layer 1 causes: cable missing or faulty, far end shut or powered off, speed mismatch.
3. Eliminate conditions that require an up link: duplex, IP, VLAN.
4. Check the far end and the cable first.

Exam tip: notconnect points to Layer 1 or the far end being down, never to IP or VLAN settings.

CRC errors mean frames arrived with bits changed in transit.

A CRC (FCS) error means the frame's checksum did not match what was received, so bits changed on the wire. With duplex and speed matched on both ends and no collisions, the remaining causes are physical: damaged cable or connectors, or electromagnetic interference. The errors rising only when the elevator motor runs is the giveaway for EMI on unshielded cable. Duplex and speed problems are ruled out by the outputs, and MTU problems show as giants.

1. Confirm duplex and speed match on both ends.
2. Note there are no collisions or late collisions.
3. Correlate the CRC increase with the motor activity.
4. Conclude interference and plan shielded cable, fiber or a new route.

Exam tip: CRC errors with matched duplex and no collisions point to the cable or its environment.