RouterOS 7.23.3. I've hit the same symptom twice, in two unrelated configurations:
- Standalone, no CAPsMAN: main router (hAP be³ Media) with 3 local WiFi 7 radios grouped into MLO via mld-interface. Clients near it would disconnect/reconnect repeatedly. Fixed by removing the mld-interface assignment from each radio directly.
- Via CAPsMAN: a second device (hAP BE Lite, mt7992) added as a CAPsMAN-managed CAP. CAPsMAN automatically grouped its two radios into MLO by default (mld-interface= on the dynamic /interface wifi entries). Same symptom: clients connect with strong signal (-23 to -36 dBm, confirmed proximity) but disconnect within 1-47 seconds ("connection lost"/"not responding"), bouncing repeatedly between this AP and others on the same SSID. Manager/CAPsMAN session itself stayed healthy the whole time — only client associations flapped. Fixed with multi-link-mode=disabled on the provisioning rule + forced reprovision (/interface wifi capsman/remote-cap/provision [find address~""], rule changes don't apply retroactively to an already-connected CAP).
None of my client devices (regular phones, sable, so they're not actually trying to useMLO — it looks like presenting an MLO-grouped AP confuses normal (non-MLO) client association/roaming behavior.
Question: is this a known limitation of the current MLO implementation with non-WiFi7 clients, or is there a correct
configuration so non-MLO clients aren't affee for actual WiFi 7 devices later on? Happyto share full config/log diffs for either case if useful.