Cisco ENNA 300-445 study materials
Cisco ENNA 300-445 study resources

Warehouse Wi-Fi: Scanner Roaming and a Renewed Internal CA

Cisco ENNA 300-445 · Put your knowledge to work in context.

All case studies

Warehouse Wi-Fi: Scanner Roaming and a Renewed Internal CA

A 60,000 m² distribution centre uses 38 Catalyst 9130 APs on a 9800-L controller. All APs use 80 MHz channels on 5 GHz, and 2.4 GHz is still broadcasting. The 1,400 clients are Zebra handheld scanners (5 GHz, 802.11ac), forklift-mounted tablets and 120 laptops doing 802.1X EAP-TLS through ISE. Scanners drop sessions for 3–8 seconds at aisle transitions. The controller reports retry rates around 38% and 5 GHz channel utilisation of 72–84% during shifts. RSSI at the worst aisle end is −71 dBm. RRM has been left in its default startup mode and has never been re-run. After the internal CA was renewed last month, about 40 laptops fail authentication with 'EAP-TLS handshake failed - unknown CA' while the other 4,960 connect normally. Warehouse management has refused funding for additional APs this fiscal year, so the fix has to come from configuration and channel planning.

1. Diagnose the 3–8 second scanner dropouts and propose a concrete RF configuration change covering channel width, data rates and roaming features. Support the channel-width decision with the number of non-overlapping channels available in the 5 GHz UNII bands.

Read the worked answer

The numbers point to co-channel interference, not coverage. A 5 GHz deployment has roughly 24–25 non-overlapping 20 MHz channels across UNII-1, UNII-2A, UNII-2C and UNII-3 (exact counts depend on the regulatory domain). At 80 MHz that collapses to about six usable channels for 38 APs: a 6:1 reuse pattern with heavy overlap. Overlapping 80 MHz cells raise the noise floor, push retries to 38% and utilisation to 72–84%, and force clients to hold on to a serving AP far too long because the client only roams when the signal is nearly gone.

Recommended configuration:

  • Move the aisles to 20 MHz. Scanners need range and reliability, not throughput. Twenty-four 20 MHz channels for 38 APs gives roughly 1.5:1 reuse. Use 40 MHz only in the low-density office corner if a survey supports it.
  • Disable the 2.4 GHz radios unless a documented legacy device needs them. They add airtime overhead and are not used by the scanners.
  • Disable the low rates (1, 2, 5.5, 11, 6, 9, 12, 18, 24 Mbps) on 5 GHz so clients roam at a sensible SNR instead of at the edge of the cell.
  • Enable 802.11k neighbour lists and 802.11v BSS Transition Management on the SSID. Enable 802.11r only after confirming the scanner and tablet firmware supports fast transition; a failure there is worse than slow roaming.
  • Take RRM out of startup mode, run a manual channel plan for the aisles, then leave RRM in monitor mode so it does not reassign channels under load.

Narrowing from 80 to 20 MHz also buys roughly 6 dB of noise-floor improvement, which directly addresses the −71 dBm worst-case reading.

2. Forty laptops now fail with 'unknown CA' after the internal CA renewal while the remaining 4,960 connect. Explain the most likely cause and the fix, working through the certificate chain and the ISE configuration.

Read the worked answer

EAP-TLS is mutual authentication: ISE validates the client certificate chain against a trusted root in its own trusted-certificate store, and the supplicant validates the ISE server certificate against the client's trust store. A renewal can break either direction.

The most likely cause given that this is a small subset is a chain mismatch introduced by the renewal, in one of two forms:

  1. The laptops that fail had their certificates reissued by the new issuing CA, but the new root or intermediate is not yet in ISE's trusted certificate store. ISE cannot build a chain and returns the unknown-CA style failure.
  2. The old root was removed from ISE's trusted store immediately, so the laptops still holding old certificates fail. The 4,960 that work are the ones whose chain happens to terminate in something ISE still trusts.

Work the problem in order:

  • Read the actual failure reason in ISE Live Logs (Operations > RADIUS > Live Logs) for one failing MAC. It distinguishes an untrusted client chain from a server-certificate or CRL problem.
  • Check the chain ISE presents with openssl s_client -connect <psn-fqdn>:8443 -showcerts and confirm the PSN certificate's SAN matches the FQDN the supplicant is configured to validate.
  • Confirm the new root and intermediate are loaded in ISE under Trusted Certificates, and keep the old root trusted until every endpoint has re-enrolled.
  • Verify the CRL and OCSP URLs in the new certificates are reachable from the PSN, because an unreachable revocation point also fails the handshake.
  • Push the new root to clients by GPO and let autoenrollment reissue the 40 certificates.

Not a fix: disabling certificate revocation checking to make the errors go away.

3. What measurements would you use to prove the RF change worked after 48 hours, and what thresholds define success?

Read the worked answer

Measure the same counters before and after, at the same time of day, because shift patterns change the load dramatically.

  • Retry rate per AP and per SSID: target under 15% from the current 38%.
  • 5 GHz channel utilisation at peak: target under 50%. Above 60% with 20 MHz channels means the channel plan still has a reuse problem.
  • Roam latency: run a continuous ping or ARP test from a scanner walking a full aisle loop and count gaps over 200 ms. Scanner sessions should not need a full re-authentication on every roam if 802.11k/v are working.
  • RSSI and SNR at the worst aisle end: aim for −67 dBm or better and SNR of 25 dB or more. The 4 dB improvement from 80 to 20 MHz should show here.
  • Per-radio channel assignment: confirm no channel is used more than twice in the same physical area, and that no AP is on a channel adjacent to a strong neighbour where it matters.
  • ISE authentication success rate for the laptop SSID and the number of EAP-TLS failures per hour.

If retries fall but roam gaps stay above a second, the remaining problem is client behaviour rather than RF: check whether the scanner firmware supports 802.11v and whether the WLC is actually sending BSS Transition requests. Report improvements as a before/after table so the operations team can defend the change without new APs.