I've been successful at installing the snapshot for MX5500 on a couple of nodes, but I've noticed something extremely odd.
AT&T RG uses 192.168.1.254
as RG IP for DHCP, so default OpenWRT install uses 192.168.1.1 and correctly (and seamlessly) gets connectivity through the RG without issue. However, if I connect my PC (regardless of which NIC) without the MX5500 in the line, so, 'virtually' direct to the RG, I get full gigabit duplex connectivity (minus standard overhead).
Having flashed 2 MX550 boxes now, if I put either one of them in front of my RG, my speeds drop staggeringly - about 25% down and 30% up. I tested this with all three NICs as well as with 2 Win11 Enterprise VMs in Hyper-V with virtual switches attached to the 2 onboard NICs.
My first thought is that, being a basic install, I have a lot of configuration (and possibly some additional package installation) before I get it working correctly. However, I want confirmation either that that is true, or else that I might be pushing past the limits of my run. So, here is what it looks like:
RG in basement -->
Cat 6A run to NetGear 8 port Gb switch -->
Cat 6A run to patch panel -->
Cat 5e run from patch panel to 'office' 2 stories up -->
replaced termination jack -->
Cat 6A run to Tp-Link SG116 16-pot switch -->
Cat 6A run from 3x switch ports to my PC NICs
(2x on board, Gb and 1.5 Gb, and 10.0 Gb PCIe addon)
When setting up and trying to configure the MX5500s, I've taken 2 of the runs from the switch to my onboard NICs and used them as WAN connection for the MX5500s, and then connected my NICs using additional Cat 6A cables to the LAN ports, one NIC per MX5500.
Notes:
-
Cat 5e in the walls was installed by builder, and not much I can do about it, but I did use a Southwire tester and all 4 pair tested fine one both ends.
-
All Cat 6A I built myself, so many are 2 m lengths, but one is at least a 5 m length (termination jack to switch). All were also tested with same Southwire cabler tester.
-
Just to be sure that my testing was not goofed, or reliant on a specific NIC, I swapped in every combination I could think of, and tested from PC using Speedtest.net, after disabling all NICs except the one being explicitly tested.
-
Even with some basic setup, like assigning my own subnet to the LAN on one of hte MX5500, nothing changes.
So, I want to know - is it possible that I'm simply adding the nodes into a chain that is exceeding some theoretical limit of devices in a chain (I seriously, seriously doubt that, but stranger and stupider things have happened), or does the basic install require a lot of configuration post installation before I can even begin setup to achieve relatively decent throughput (I'd be happy with 80%, but would obviously prefer as close to native as possible).
Example results:
hamstringed
'direct'