Solved Strange behaviour on devices using mesh

DJones

Regular Contributor
386.7 on main router and node then tried 386.7_1 on both as well.

So this may not be Merlin or ASUS related directly I’m not quite sure. I’ve attempted to narrow the issue down to it being only the devices connected to my mesh node. On my devices connected to my mesh node I’m recently unable to connect to my samba server (dedicated server not smb on router) and I’m unable to connect to webmin and netdata.

Initially I though okay maybe my servers firewall is acting up, so I disable UFW and Skynet firewall and my Norton smart firewall on my computer connected to the node. Figured if it was a firewall issue I’d be able to connect after but that doesn’t seem to be the case. I can connect to these services on the main router fine, but not the node and I’ve tried with multiple devices on the node

But it’s not that I cannot access my server I can ping it from the node and I run a dashboard Heimdall on Apache from the server and can access it no problem. And can even access portainer, but not the others. SMB just hangs until explorer on windows crashes it can detect the space on the server but not open the mapped drive.

I’ve looked at the dmesg server logs, router logs and windows event logs, but nothing stands out. Also no Firewall logs indicating anything is wrong. Nothings changed unless Ubuntu server updated something that broke, but since I can access everything fine on my main router I’m at a loss. I attempted to run the node in media bridge mode as well but experienced the same problem. 3 services I’m unable to access. The issue seems recent because a few days ago it worked fine.
Any advice would be appreciated if you can point me in a direction to look into.


Update: So I figured out the issue and it's something that makes me bang my head against my desk. I had to use tcp_mtu_probing = 1 on my server since I'm working with Jumbo frames and wifi mesh does not handle jumbo frames across the wifi bridge. The node is perfectly happy to use jumbo frames locally on the nodes switch but since it's not a wired backhaul to my main router it still has to fragment the packets back to 1500 I assume. As to why somethings were accessible and not others my guess is Docker for my dashboard / portainer probably has some similar function as tcp_mtu_probing = 1, but I don't know for sure. I caused the issue when I cleaned up my sysctl file on the server.

Controls TCP Packetization-Layer Path MTU Discovery. Takes three values: 0 - Disabled 1 - Disabled by default, enabled when an ICMP black hole detected 2 - Always enabled, use initial MSS of tcp_base_mss.
 
Last edited:

Similar threads

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top