Ok so this one has me stumped. I’ve got a port that I’ve limited to 10baseT so that the customer who is providing us with tower access can have a 10mbps connection. As soon as he maxes out the port latency to the router and to anything else past the router spikes dramatically from 1-2ms to upwards of 100. CPU usage on the router stays low and there is plenty of backhaul capacity. If I open the port up to 100baseT and let him pull 20-30mbps the latency falls back down to normal.
It and three other ports all belong to one bridge. Ethernet 2 is for our access point so it has vlan251 on it for management traffic. And ether10 is bridged to v251 to allow crews access to the management network from their laptops. The bridge has a /28 ip address as well as a dhcp server on it. Below is the interface export
/interface ethernet
set [ find default-name=ether1 ] comment=\
"to 4-mile v287 is transit for bgp peering"
set [ find default-name=ether2 ] comment="access point"
set [ find default-name=ether3 ] advertise=\
10M-half,10M-full comment="joes connection"
set [ find default-name=ether4 ] comment="computer access"
set [ find default-name=ether5 ] comment="v251 access"
set [ find default-name=ether10 ] comment="sitemonitor, v251"
/interface vlan
add interface=ether1 l2mtu=1594 name=e1-v251 vlan-id=251
add interface=ether1 l2mtu=1594 name=e1-v287 vlan-id=287
add interface=ether2 l2mtu=1594 name=e2-v251 vlan-id=251
/interface bridge
add name=/28-customers protocol-mode=none
add name=loopback
add name=null
add name=v251
/interface bridge port
add bridge=v251 interface=e1-v251
add bridge=v251 interface=ether10
add bridge=/28-customers interface=ether2
add bridge=/28-customers interface=ether3
add bridge=v251 interface=e2-v251
add bridge=v251 interface=ether5
add bridge=/28-customers interface=ether4
/interface bridge settings
set use-ip-firewall=yes
The rb2011 is peered with a ccr upstream, ospf is anouncing the transit network and the loopback. BGP is doing the customer/traffic network and is peering on the loopback.
Disabling firewall/contracking had no effect either.
I’ve tried getting rid of the bridge for the customer network altogether and simply doing a master/slave setup for the interfaces but nothing changed.
If anyone has any ideas on this it would be greatly appreciated, as of right now I’m at a loss.