caps error - "removing stale connection" since 6.49.x

RouterOS Wireless Networking
1

Hello community,

by all of my installations, where CapsMan is running, after Upgrading rom 6.47.x/6.48.x to 6.49.x/7.1.x we get really annoying error. the LOG is RED, full of it.
for example:

removing stale connection [74:4D:28:E7:C4:9C/16/4d9c,Run,[74:4D:28:E7:C4:9C]] because of ident conflict with [74:4D:28:E7:C4:9C/16/61dc,Join,[74:4D:28:E7:C4:9C]]
removing stale connection [08:55:31:B0:CD:40/20/fc02,Run,[08:55:31:B0:CD:40]] because of ident conflict with [08:55:31:B0:CD:40/20/a1c2,Join,[08:55:31:B0:CD:40]]
removing stale connection [C4:AD:34:5B:B5:46/16/be70,Run,[C4:AD:34:5B:B5:46]] because of ident conflict with [C4:AD:34:5B:B5:46/16/77fb,Join,[C4:AD:34:5B:B5:46]]
removing stale connection [74:4D:28:E7:C2:10/20/e0b,Run,[74:4D:28:E7:C2:10]] because of ident conflict with [74:4D:28:E7:C2:10/20/c703,Join,[74:4D:28:E7:C2:10]]

and so on. every of 24 Cap AC will bring it. Anybody else, who got em and could solve it?

What I already did, but please, have in mind, EVERYthing worked like a charm, till the “stable” Update… Last Update was necessary, because of wireguard…

  1. checked Admin Mac on CCR1009
  2. just to proof, deleted admin mac on Cap AC / Hap Ac2
  3. checked ALL my VLAN config
  4. upgraded System->Routerboard
  5. changed Combo Port to copper only, on CCR1009

Like some Hex/Hex S, after Upgrade to 7.1.1 and some restarts, CCR1009 was dead, had to recover it with netinstall.

customers with hex/hex s and hap ac2/3 as router have same problems. Capsman is on, to manage cap ac and hap’s → removing stale connection… I dont geht it, sorry.
I will also open ticket with support file…

Thanks a lot…for any advise!

Greets
Alex

2

So far no CAPsMAN issues for me on 6.49…

Those kind of symptoms point to intermitent L2 (or L3) issues: flapping port, bridge… are SFP modules involved?

Have you tried to set Admin Mac address on cAPs to the wired ether one??

3

Hi and thank you for a fast answer

  • I tried, I think, nearly EVERYthing, before the post here.

Actually, since BRIDGE mac address is same to the ETH1 Port on Cap AC, I could not set any other mac as showed.
This issue is with VLAN, and without VLAN, with Firewall enabled, without Firewall, customers with SFP and without.

One customer, although the capsman is on main POE CRS328 and SFP only connected to a CSS326, so all 5 Cap’s AC are on the POE Switch…Firewall disabled, has this issue too! but also on 6.48.4
VLAN is implemented just the way WIKI shows…
My problem is, that everything worked not bad, never saw the “removing stale connection” issue before…

thank you for any suggestion



4

Not sure if this will somehow help but I have the same issues at CCR1009 all the CAPs are disconnecting when the load is too big. That only happens on big installations with a few hundreds of CAPs.

The only solution from Mikrotik support was to get a separate ARM machine (4011 was suggested) to run as a Capsman server.

5

Well .. I also get thus error with 5 Caps…
This cant be the reason I think…

Greets

6

SOLUTION:

after 2 Month I finally found the bad boy.

Since my configs and scripts are 2 years old, and back then VLAN interfaces have been configurated as MEMBER PORTS of the BRIDGE (by default)… caused the issue with “removing stale connection”

/interface vlan
add interface=bridge name=vlan1 vlan-id=1
add interface=bridge name=vlan11 vlan-id=11
add interface=bridge name=vlan12 vlan-id=12
add interface=bridge name=vlan13 vlan-id=13

/interface bridge port
add bridge=bridge interface=vlan1
add bridge=bridge frame-types=admit-only-vlan-tagged interface=vlan11 pvid=11
add bridge=bridge frame-types=admit-only-vlan-tagged interface=vlan12 pvid=12
add bridge=bridge frame-types=admit-only-vlan-tagged interface=vlan13 pvid=13


Had to remove em all from over 100 customers… :astonished:

Hopefully, I could someone help , cause Mikrotik couldnt :laughing:

7

Hey bud, not quite sure whether it was since 6.49, however, once I disabled the bridge VLANs, the issue is gone and moreover, there is no reconnection from time to time.

Thank you for your the-only-solution advice.
Dunno if this might be helpful for someone or not, but the disconnects mostly appeared on 2+ port routers. One port solutions not suffering that much, if ever.

Edit: it is possible to remove the non-dynamically-added Bridge VLANs on the APs. But start with the CAPs manager router (which definitely does not necessarily need to be an AP).

8

Hi shurik3, could you please write wat exactly you had in the config, what escactly you removed and what exactly you recreate?
On whitch interface are the vlans of the new configuration connected?
I have the same problem but cannot understand your solution.
Thank you!

9

This is a standard problem of not properly configuring the capsman.
When local forwarding is enabled, physical radio interfaces wlan 1/2 must be excluded from the bridge.