A “Wi-Fi isn’t working” report is the common name for five very different faults: failure to associate, authentication error, failure to get an IP, frequent disconnects and low speed. Each has its own order of checks and tools; approaching all of them with the same reflex, “restart the AP,” hides the problem for a few hours but doesn’t solve it. Below we explain how to classify enterprise Wi-Fi faults by symptom, which commands to look at on the client and controller side, and when wireless packet capture is needed, in the order of checks in the field.
Classifying problems by symptom
The first step of diagnosis is translating what the user describes into a connection stage. Until a client is working on Wi-Fi, it passes through these stages in order: finding the network (scanning), 802.11 authentication and association, 802.1X/PSK key setup, getting an IP address, and data transfer. Which stage the problem is in determines where you look.
| What the user says | Likely stage | First place to look |
|---|---|---|
| “The network isn’t in the list” | Scanning | SSID broadcast, band, client band support |
| “Can’t connect,” “wrong password” | Association / key setup | Security setting, RADIUS logs |
| “Connected but no internet” | Getting an IP | DHCP, VLAN mapping |
| “It keeps dropping” | Data stage / roaming | RF quality, roaming events, power saving |
| “It’s very slow” | Data stage | Data rate, channel utilization, wired side |
The second step is determining scope: is the problem with a single user, a device model, an AP or floor, or everywhere? A problem on a single device is usually caused by the client or driver; a problem concentrated in one area is RF or the AP; a problem across the whole network is authentication, DHCP or the wired infrastructure. Diagnosis doesn’t start without knowing the scope. For a general approach that also covers the wired side, see the layer-by-layer troubleshooting of network slowness article.
Association and authentication errors
If the client sees the network but can’t connect, first check that the security settings match. One cause you may encounter recently is the incompatibility of WPA3, PMF or 802.11r changes made on the SSID with old drivers. If the SSID settings were touched recently according to the change log, that’s the first suspect.
On enterprise networks using 802.1X, the order of checks:
- RADIUS logs: If Microsoft NPS is used, look at events 6272 (access granted) and 6273 (access denied) in the Security event log. The reason code and matching network policy in a 6273 event show whether the problem is the user’s credentials, a policy condition or a certificate. If there are no records at all, the request isn’t reaching RADIUS: check that the AP/controller is defined as a RADIUS client, the shared secret and UDP 1812/1813 access.
- Certificates: If the RADIUS server certificate has expired, or the client sees the renewed certificate from a CA it doesn’t trust, mass disconnects occur. With EAP-TLS, the expiry and revocation status of the client certificate are also checked.
- Controller events: 802.11 reason codes point the way. Reason code 15 indicates that the 4-way handshake timed out (often a wrong password or key mismatch), and reason code 23 that 802.1X authentication failed.
For a single client’s problem, match three records over the same time window: the client’s own log, the events the controller or access point keeps for that client’s MAC address, and the records on the RADIUS and DHCP servers. First verify that the device clocks are in sync via NTP. If a connection attempt appears on the client but there’s no association on the controller, look at RF or the security settings. If there is an association on the controller but no record in RADIUS, the request isn’t reaching the server. If RADIUS granted access but there’s no request on the DHCP server, check the VLAN mapping.
Failure to get an IP: DHCP and VLAN mapping
If the client appears connected but has a 169.254.x.x (APIPA) address or no IPv4 address at all, the wireless connection is fine; the DHCP stage is what failed. Stop looking at RF. Possible causes:
- VLAN mapping: In architectures using local bridging (local switching), if the SSID’s VLAN isn’t allowed on the trunk of the switch port the AP is connected to, client traffic is lost. If the problem appears only on specific APs, this is the first suspicion; it’s common on newly installed APs or ones that changed ports.
- Dynamic VLAN assignment: If the VLAN returned by RADIUS (the Tunnel-Private-Group-ID attribute sent with Tunnel-Type=VLAN and Tunnel-Medium-Type=802) isn’t defined on the controller or the AP uplink, only that user group can’t get an IP.
- DHCP scope exhaustion: Especially on guest and high-density networks, the scope fills up when the lease time is kept long.
- DHCP relay and snooping: A wrong helper address or an uplink marked untrusted prevents DHCP responses from reaching the client; details are in the DHCP relay and DHCP snooping article.
Frequent disconnects: RF, roaming and power saving
With a disconnect complaint, first clarify what the disconnect is: is the client really dropping off the network, or does it stay connected while the application times out? The client association history on the controller shows this distinction. The two have completely different solutions.
- Weak or unbalanced RF: If the signal and SNR the client sees are at cell-edge level, it may be a one-way coverage problem where the AP can’t hear the client.
- Roaming problems: A “sticky client” (a client that stays connected to a distant access point) stays on a distant AP; an aggressive minimum RSSI setting, on the other hand, keeps dropping the client. Incompatibility between 802.11r and some clients also causes disconnects during roaming. For the mechanism, see the roaming with 802.11k, 802.11v and 802.11r article.
- DFS events: When an AP detects radar and changes channels, clients on that AP experience a brief interruption. If radar events in the controller logs are concentrated on specific APs, those APs can be moved to non-DFS channels.
- Power saving: The wireless adapter’s power saving mode on a laptop can cause latency and session drops, especially on battery. If the problem increases on battery and decreases when plugged in, the wireless adapter settings in the power plan and the driver version are checked.
Low speed: data rate, channel utilization and the client
With slowness, the first question is whether the bottleneck is really on the wireless side. If a test done with a wired computer at the same spot is normal, the problem is on the wireless side; if not, blaming Wi-Fi is a waste of time. Then look at:
- Connection data rate: If the receive/transmit rate in the
netsh wlan show interfacesoutput is low, the client has connected with a weak signal or an old standard. - Channel utilization: If the channel utilization of the AP radio on the controller is high, the channel is being shared: co-channel interference, a large number of clients or low-rate broadcast traffic.
- Retransmissions: A high retransmission rate points to interference or a cell-edge client.
- Client limitation: A single-stream (1×1) phone and a 2×2 laptop get very different speeds at the same spot; this isn’t a fault.
- Wired side: The speed/duplex match of the AP uplink, port errors and distribution layer utilization are checked. An AP uplink that has dropped to 100 Mbps can’t be compensated by any setting on the wireless side.
Diagnostic commands on the client and controller side
On a Windows client, the most comprehensive source is the wireless connection report run as administrator. The report is produced as HTML with connection and disconnect events of recent days together with reason descriptions:
REM At an administrator command prompt
netsh wlan show wlanreport
REM Output: C:\ProgramData\Microsoft\Windows\WlanReport\wlan-report-latest.html
REM SSID, BSSID, channel, signal, rate, authentication method
netsh wlan show interfaces
REM All APs seen (BSSID), channel and signal
netsh wlan show networks mode=bssid
REM IP, gateway, DNS and DHCP server
ipconfig /all
On Linux clients, iw and NetworkManager tools give similar information:
iw dev wlan0 link # connected BSSID, signal, tx/rx rate
iw dev wlan0 station dump # retransmission and error counters
nmcli dev wifi list # nearby SSIDs/BSSIDs, channel, signal
journalctl -u NetworkManager --since "1 hour ago"
On current versions of macOS, the sudo wdutil info command gives connection details. On the controller side, the detailed status is queried with the client’s MAC address. On Cisco Catalyst 9800 controllers:
show wireless client summary
show wireless client mac-address aaaa.bbbb.cccc detail
This output shows which AP and radio the client is on, its association state, assigned VLAN and IP address, authentication method and signal values. On other vendors, equivalent commands and client event timelines are found in the management interface. The first thing to check on the console is the last successful stage in the client’s event timeline; the fault is in the next step.
When is wireless packet capture needed?
Client and controller logs solve most problems. Wireless packet capture is for cases where the logs contradict each other or fall short:
| Status | Why is capture needed? | Frames to look at |
|---|---|---|
| It’s unclear who initiated the disconnect | To see the source and reason code of deauth/disassoc | wlan.fc.type_subtype == 0x0c and 0x0a |
| The handshake doesn’t complete | To see at which EAPOL message it stops | eapol |
| Roaming takes long | To measure the time difference between the old and new AP | Authentication, reassociation, EAPOL |
| High latency, loss | To see retransmissions and airtime congestion | wlan.fc.retry == 1 |
Wireless capture requires an adapter that supports monitor mode or the AP’s sniffer mode; capture is done on the client’s channel. Capturing on the wrong channel produces an empty file. In roaming analysis, because the client moves between multiple channels, more than one capture point may be needed. For filters and capture setup, see the network packet analysis with Wireshark article, and for 802.11 field names, the Wireshark documentation.
We address recurring problems or ones whose source can’t be found under our network performance analysis and troubleshooting service, and problems arising from RF design under our wireless network optimization service.
Frequently asked questions
Where should you start if Wi-Fi appears connected but there’s no internet?
First check whether the client got a valid IP address. An address starting with 169.254 shows that the DHCP stage failed; check whether the SSID’s VLAN is allowed on the AP uplink, the DHCP scope and the relay setting.
What is netsh wlan show wlanreport for?
Run with administrator rights on Windows, this command produces an HTML report containing the wireless connection, disconnect and error events of recent days with reason descriptions. It’s the first client-side source to check to see the time and cause of disconnects.
If all users get authentication errors at the same time, what is the likely cause?
Most often it’s a change on the RADIUS side: a server certificate that expired or was renewed and isn’t trusted by clients, a changed network policy or lost access to RADIUS. If NPS is used, the reason code in 6273 events points the way.
Is wireless packet capture needed for every problem?
No. Most problems are solved with client reports and controller logs. Capture is for cases where it’s unclear who initiated the disconnect, the handshake doesn’t complete, or roaming time needs to be measured.