I have three RBs (2x 751U-2HnD and 1x 411R with additional AR5007G). These all are connected to a dynamic-mesh (& ap bridges). The setup has been running for a long time without issues. Basically following the examples here in forums. The latest incarnation of this setup had 751Us running os 5.24 (cannot remember the fw version sorry) and 411R running 6.4 (with 2.x something as the fw). This was a stable setup and ran a mixture of public/private IPv4 and public IPv6. Recently I upgraded all routers to 6.12 (751U has fw 3.13, 411R has 3.10). Note.. I do not enjoy configuring & experimenting with the setup & os versions.. I really don't.

Anyway, now I am left with a setup that runs fine.. until it suddenly dies. Symptons are 100% CPU occasionally (both 411R and 751U) , mesh interfaces not forwarding or dropping from the mesh. 100% CPU seems to get less frequent and reboot fixes that. The mesh either drops entirely or typically starts flapping i.e. some router pair can exchange packets while others don't. Sometimes mesh heals itself after a while but usually I need to disable/enable the WLAN master interface to get it join the mesh again. This is getting annoying. I am fully aware that the issue most probably is me doing goofy stuff.

Before I start posting further logs/configs any immediate ideas what to look after? Or what to capture?

From 751U:
0 R name="util-WLAN" mtu=1500 mac-address=00:0C:42:E4:AF:D7 arp=enabled
interface-type=Atheros AR92xx mode=ap-bridge ssid="tube" frequency=2442 band=2ghz-b/g
channel-width=20mhz scan-list=default wireless-protocol=802.11 antenna-mode=ant-a
wds-mode=dynamic-mesh wds-default-bridge=wds-mesh wds-ignore-ssid=no bridge-mode=enabled
default-authentication=yes default-forwarding=yes default-ap-tx-limit=0
default-client-tx-limit=0 hide-ssid=no security-profile=default compression=no

From 411R:
2 R name="util-WLAN" mtu=1500 mac-address=00:0C:42:28:08:9F arp=enabled
interface-type=Atheros AR5212 mode=ap-bridge ssid="tube" frequency=2442 band=2ghz-b/g
channel-width=20mhz scan-list=default wireless-protocol=802.11 antenna-mode=ant-a
wds-mode=dynamic-mesh wds-default-bridge=wds-mesh wds-ignore-ssid=no bridge-mode=enabled
default-authentication=yes default-forwarding=yes default-ap-tx-limit=0
default-client-tx-limit=0 hide-ssid=no security-profile=default compression=no

Mesh setting:
name="wds-mesh" mtu=1500 arp=enabled mac-address=00:0C:42:E4:AF:D7 auto-mac=yes
admin-mac=00:00:00:00:00:00 mesh-portal=no hwmp-default-hoplimit=32 hwmp-preq-waiting-time=4s
hwmp-preq-retries=2 hwmp-preq-destination-only=yes hwmp-preq-reply-and-forward=yes
hwmp-prep-lifetime=5m hwmp-rann-interval=10s hwmp-rann-propagation-delay=5
hwmp-rann-lifetime=22s reoptimize-paths=no

Mesh port setting:
0 interface=util-WLAN mesh=wds-mesh path-cost=10 hello-interval=10s port-type=auto active-port-type=wireless


UPDATE #3:

Tried also static-mesh.. not much help since that increased the number of 100% CPU cases
When looking at the profile the "unclassified" takes over 96% of CPU.
Just noticed that 411R reboots quite often when I fiddle around with the mesh links on _other_ 751U routers.. there is no note about the crash in the 411R log. I am now back to ROS6.4 on all RBs. Lets see if this gets more stable (the 6.12 is still left as-is on the other partition).

Right.. downgrading back to 6.4 did not help too much. I get 100% CPU after a while, which basically then kills the mesh. Next is, I guess, trying to downgrade the firmware to something below 3.x. Not sure how to do that "by the book" without turning the RB a brick (last time I tried had to do netinstall to get RB back to life, hehe).

You could go safely to 4.17 on those boards i think

You could go safely to 4.17 on those boards i think

Hmm.. I meant downgrading the "current-firmware", not the RouterOS. Version 6.4 of the RouterOS was good for me for a long time.

Ok.. I give up with this. Maybe it is 751U wireless issue in general. I did notice a "mesh" (pun intended) with only two nodes i.e. just a p2p thing sort of works more than 24h without a hick-up. Anything else kills the mesh interface after a while and the WDS connections drop.