Intermittent AP Drops and PoE Resets in a University Library
Edge University's library has 30 EX4300-48P switches and 220 APs. Over two weeks, APs on floors 2-4 drop for 3-7 minutes and then recover. Mist Wired Assurance shows switch health at 71 percent for the building and six ports with repeated port flap and PoE reset insights. Marvis flags one port as bad cable and another as duplex mismatch. Port statistics show CRC and alignment errors rising on ge-0/0/17; ge-0/0/23 is linked at 100 Mbps half-duplex even though the AP supports 1 Gbps. The building PoE budget is at 78 percent. The APs are business-critical for exams, and the network team wants to avoid replacing all cabling. The six affected ports are split across three switches on the same closet uplink.
1. Prioritize the first three troubleshooting actions using the Mist evidence. Explain why you would not start by replacing every cable.
Read the worked answer
First, check the shared closet uplink and the three switches because the affected ports are spread across switches but share an uplink. Port flaps on multiple switches point more to an uplink, power, or upstream issue than to six independent bad cables. Second, act on ge-0/0/17: rising CRC and alignment errors plus the bad cable insight justify replacing that patch cord and testing the port. Third, fix ge-0/0/23: a 100 Mbps half-duplex link to a 1 Gbps AP is a negotiation or cable problem, so check the cable, replace it if needed, and confirm autonegotiation before hardcoding speed and duplex. Do not start by replacing all cabling because the evidence is clustered, not random. The uplink and two specific ports explain more of the symptom pattern, and mass cable replacement consumes the maintenance window without proving root cause. Treat Marvis insights as hypotheses to confirm with port counters, LLDP, PoE status, and switch logs.
2. What commands and Mist views would you use to confirm the root cause on ge-0/0/17 and the shared uplink?
Read the worked answer
In Mist, open the switch details, port insights, and Marvis actions for the affected ports. Compare CRC, alignment, flap counts, PoE reset counts, and LLDP neighbor data over time, not just the current value. On the CLI, use show interfaces ge-0/0/17 extensive to check CRC, alignment, carrier transitions, and speed/duplex; show lldp neighbors to confirm the AP is on the expected port; show poe interface to see power draw and resets; and show log messages | match ge-0/0/17 for link events. For the uplink, check show interfaces extensive, show lacp interfaces if bundled, show chassis alarms, and show system alarms. If the uplink is optical, check diagnostics optics for receive power. The goal is to correlate the port flap timestamps with uplink errors or power events. If uplink counters are clean, move down to the individual patch cords and AP ports; if uplink counters are dirty, fix the shared path first.
3. What configuration or monitoring changes would reduce recurrence?
Read the worked answer
Set PoE priority high on AP ports, keep the building PoE budget below 80 percent, and alert on PoE reset and power-deny events. Enable port flap and CRC error monitoring through Mist alarms or webhooks so the team sees degrading links before exams. Replace the bad patch cord on ge-0/0/17 and correct the duplex mismatch on ge-0/0/23; avoid hardcoding speed and duplex unless autonegotiation repeatedly fails with a known-good cable. Verify the closet uplink is correctly sized and error-free, and monitor it with the same alarm policy. Add a monthly check of switch health, port error counters, and PoE utilization. After changes, compare the affected ports for a full week and confirm that AP offline duration falls and that no new port flap or PoE reset insights appear.