search

halmartin / meraki-builder

Scripts and tools to assemble firmware images for various Meraki MS switches and MX routers
GNU General Public License v3.0
77 stars 17 forks source link

DHCP Broadcast not passed through #13

Closed ben-oxley closed 3 years ago

ben-oxley commented 4 years ago

Whilst I've not fully tested this yet, I seem to see a weird behaviour when following these steps:

  1. Connect DHCP Server and upstream network to Port 1.
  2. Boot up MS220-8P
  3. Connect POE powered device (MR-18 flashed with OpenWRT) to Port 2.
  4. Device on port 2 does not acquire an IP address.

I suspect that either the switch is not passing on DHCP packets or probably more likely isn't forwarding broadcast packets?

I'll continue to investigate with some wireshark etc.

halmartin commented 4 years ago

Can you MITM the upstream network port of the MS220-8P and see if you see the DHCPREQUEST from the MR-18?

I have not had any issues with downstream devices obtaining a DHCP lease on my MS220-8P.

ben-oxley commented 4 years ago

Will do! I'll get back to you with it, as soon as I can get something to MITM the traffic.

halmartin commented 4 years ago

You don't need to necessarily MITM the traffic, you can run dnsmasq (and optionally tcpdump) on the upstream device and you'll have a view of DHCP request/replies.

e.g. on the upstream device connected to the MS220:

ip addr add 192.168.10.1/24 dev eth0
dnsmasq -a 192.168.10.1 -d -F 192.168.10.10,192.168.10.20,2h -i eth0

dnsmasq will then show DHCPDISCOVER, DHCPOFFER, DHCPREQUEST, and DHCPACK in stdout. Replace eth0 with your Ethernet interface device name.

ben-oxley commented 3 years ago

I don't have the results of a tcpdump/similar for you yet, but check this out. The issue turns out to only be on port 2 of the switch (Ports 1, 3-8 all seem to work fine!). It's possible that its just a dodgy port and that port 2 just isn't usable.. I've got another switch that I need to flash at some point soon so I'll try that and see if it has the same problem and report back.

halmartin commented 3 years ago

This is the same switch you were having PoE issues on, right? If I remember correctly, you said that you tried other ports and PoE was still non-functional, so there may indeed be something wrong with the 220-48LP.

It's strange though, because I would expect the same issues on the 320-48LP/FP, and other people have not reported any issues with these models. The only differences I know of between the 220-48 and 320-48 are the hot-swap power supplies.

What does /click/switch_port_table/dump_pports say when you have a device connected to port 2? I assume the port is at least negotiating a link.

It might also be worth taking a look at /click/switch_port_table/dump_pport_vlans; maybe I've got an error in an initscript and the port is not being initialized on VLAN/PVID 1.

ben-oxley commented 3 years ago

Sorry no, this is on an MS220-8P (not 48).

I'll take a look at the dump_pports tomorrow and see what it says.

ben-oxley commented 3 years ago

Seems to be a dodgy port on the 8p, just flashed a second one and port works just fine!