Hi all,
Running a hAP be³ Media, RouterOS 7.23.3 (stable), wifi-qcom-be package (ipq5300 chipset), latest stable, no update available. Tri-band local radios (wifi1=2.4GHz, wifi2=5GHz, wifi3=6GHz) plus a CAPsMAN-managed second AP.
The problem: every few days, wifi1 (and separately, wifi3) stop accepting/serving clients entirely. wifi2 stays perfectly healthy. Nothing recovers on its own, even after hours.
The most interesting/confusing part: I have a second SSID (domozolay, for IoT devices) configured as a slave interface bound to wifi1's same physical radio (master-interface=wifi1), same channel (2412), same tx-power. Right now, while wifi1 (the master AP) has zero connected clients and refuses to serve, the slave AP on that exact same radio and channel just authorized a brand-new client with a clean signal reading. So the RF hardware, the channel, and the driver's general ability to serve clients on that hardware are all clearly fine — the fault is isolated to the specific wifi1 AP/BSS instance itself, not the underlying radio.
What it looks like when it happens: sometimes the interface loses its running state and channel entirely. Other times (most recent occurrence), /interface/wifi/monitor wifi1 once reports state: running with a valid channel and normal tx-power — looks healthy — but registered-peers/authorized-peers both sit at 0 and the interface's bridge port shows inactive. So the state machine believes it's fine, but it isn't actually accepting associations.
Triggers observed so far (~5 occurrences in ~10-11 days):
- After a power outage / reboot.
- Spontaneously, mid-session, no obvious trigger.
- Right after a live config change on the interface while it was up (e.g. adding a steering= property).
What I've tried:
- Disable/enable toggle — works roughly 1 out of 3-4 tries, unreliable.
- Full router reboot — 100% success so far (3/3), but heavy-handed, and on this same box it has a separate ~50% chance of triggering an unrelated USB-storage-detection issue, so I'd rather avoid rebooting every time.
I also want to rule out my first suspicion: I have 802.11k/v steering configured on wifi1 but never touched it on wifi3 — yet both fail together each time, so steering alone doesn't look like the common cause.
Has anyone seen this — one AP/BSS instance on a shared radio silently stops serving while a sibling BSS on the identical channel keeps working fine, with no distinguishing log entry? Any known cause or fix on wifi-qcom-be/ipq5300 short of a full reboot?
Thanks!