CAPsMAN controller glitch: bandwith handicap

RouterOS Wireless Networking
1

CAPsMAN cAPs throughput is being occasionally handicapped at 10-11 MB/s (~85-90mbps), for all wireless clients with perfect SNR and signal (<110dBm, (-30dBm)-(-49dBm), 866Mbps), without a reason.

Rebooting cAPs doesn’t help. Rebooting clients, any manual connection drop through the client, or /caspman doesn’t help either. The problem stays until I reboot the controller - the router - router reboot restores standard throughput of 200-300 Mbps. That’s the only way of solving it.
RB3011+wAP AC, all 6.47.

Resetting configuration doesn’t help either - I have home - not even SMB - router application, so I don’t have a complicated setup.

Did anyone face this behavior before?

P.S. Yup, local forwarding is the only way I’m using it.

2

I would:

  • Upgrade to 6.47.3 (for both router and accesspoints)
  • test through iperf (I have no clue how you are currently testing?)
  • share the configuration (/export hide-sensitive file=anythingyoulike) to be sure everything is set correctly

And in addition:
What do you mean by occasionaly?
Is it solved without interference, or will it remain “capped”?
What is de CPU load on both RB3011 and wAP ac?
What client are you using for testing?
Prohibited to a single client, or do all clients suffer?
How about wired clients?

3 



/caps-man access-list
add action=reject allow-signal-out-of-range=always client-to-client-forwarding=yes disabled=no interface=all mac-address= signal-range=-120..-65 ssid-regexp=""
add action=reject allow-signal-out-of-range=always client-to-client-forwarding=yes disabled=no interface=all mac-address= signal-range=-120..-65 ssid-regexp="" 
add action=reject allow-signal-out-of-range=always client-to-client-forwarding=yes disabled=no interface=all mac-address= signal-range=-120..-59 ssid-regexp="" 

/caps-man configuration
add channel.band=2ghz-g/n channel.control-channel-width=20mhz channel.extension-channel=disabled channel.skip-dfs-channels=no channel.tx-power=13 datapath.bridge=bridge datapath.client-to-client-forwarding=yes datapath.local-forwarding=yes \
guard-interval=any mode=ap multicast-helper=full name="2.4 GHz - Configuration" rx-chains=0,1,2,3 security.authentication-types=wpa2-psk security.encryption=aes-ccm tx-chains=0,1,2,3
add channel.band=5ghz-n/ac channel.control-channel-width=20mhz channel.extension-channel=Ceee channel.tx-power=19 datapath.bridge=bridge datapath.client-to-client-forwarding=yes datapath.local-forwarding=yes guard-interval=any mode=ap \
multicast-helper=full name="5 GHz - Configuration" rx-chains=0,1,2,3 security.authentication-types=wpa2-psk security.encryption=aes-ccm tx-chains=0,1,2,3

/caps-man provisioning
add action=create-enabled hw-supported-modes=an,ac master-configuration="5 GHz - Configuration" name-format=prefix-identity name-prefix="5GHz - "
add action=create-enabled hw-supported-modes=g,gn master-configuration="2.4 GHz - Configuration" name-format=prefix-identity name-prefix="2.4GHz - "

Occasionaly means the problem reoccurs again between a couple of minutes after reset and a couple of hours, day even.
It is solved only when I reboot router - so I guess it remains without interference.
RB3011 - 0-1%, 33C; wAPs between 10-30% and 40-70%, 59-65C.

Client is PC, laptop and iPad Pro, testing methodology - simply download a .bin file from NAS 1Gbps wired.
All wireless clients suffer, wired seems to be fine.

4

I have some doubts about your accesslist, in comparison herewith mine:

/caps-man access-list
add action=accept allow-signal-out-of-range=10s disabled=no interface=any \
    signal-range=-80..-10 ssid-regexp=""
add action=reject allow-signal-out-of-range=10s disabled=no interface=any \
    signal-range=-120..-81 ssid-regexp=""

I am missing an allow action. And your thressholds are a bit of. Think i.e. Apple has a thresshold of around -75 dBi. Besides, why three accesslists?

And what processes causes the high CPU load on the wAP?

5

No sense in allow action - it is allowed by default, if not restricted
Those access rules are only applied for three clients, those are not general rules. Two Smart TV and PC.
Those three has to have access rule, since they all cannot negotiate with mikrotik like with a normal devices, so both TVs and PC kept connecting to far-out wAPs in my house, with signal being around -76dBm -86dBm, completely ignoring the closest one in the room.

And I have one living room without good wifi coverage, where clients barely live; -80dBm is at best, and no way I can cover it with 8p wire and additional wAP. But I'm not really concerned about that, since room is occupied with my family members, and they are not so geeky, bad singal doesn't bother them at all. Even if it is poor ~10mbps channel. Thus I can't apply general rule to kick out anyone with signal worse than -75dBm.

Besides, the clients I talk about are my devices, covered with wireless pretty well - -30/-50 dBm, I have stable 780/866Mbps channels, especially PC which is stationary and lives just up front of wAP AC, in ~literally 3 meters.

wAP AC CPU is VERY-VERY weak, general networking isn't trivial task for it - and all that with local forward, without capsman tunneling and etc.



P.S. Okay, so I restored my 2months old backup, 1 hour and no issues so far, wireless speed keeps above 30MB/s. It's obviously firmware bug, since I've changed literally nothing for the past two months, and this glitch occured only 2 days ago.

6

Send the supout file to Mikrotik team when the slowdown occurs so they can make a fix

7

Starting from august 21s, Ive got only 2 replies from them, last one was on Mon.
So I guess not a chance this time…