4-may
I'm trying to work with a TBF queue
to enforce bandwidth limits on a linux
bridge on an HP box running Ubuntu 20.04
w/ kernel 5.8.0-50-generic.
I set the bridge up as follows:
****
#!/bin/sh
ip link add name br0 type bridge
ip link set dev br0 up
ip link set dev ens1 master br0
ip link set dev ens2 master br0
****
I'm using an Intel EXPI9301CT NIC for
both ens1 & ens2. I have multicast video
(about 4.5 Mbps) entering the bridge on
ens1 and exiting on ens2.
I tried setting up a TBF queue with
bandwidth limit (max 3Mbps) on ens1:
****
#!/bin/sh
/usr/sbin/tc qdisc del dev ens1 root
/usr/sbin/tc qdisc add dev ens1 handle 10: root tbf rate 2.5mbit \
burst 5kb latency 70ms peakrate 3mbit \
minburst 1540
****
I ensured the TBF kernel module:
sch_tbf
was loaded. Also the kernel config file has:
CONFIG_NET_SCHED=y
The TBF queue definitely seems to be in place:
****
root@rorem-HP-Compaq-8000-Elite-SFF-PC:~# tc qdisc show dev ens1
qdisc tbf 10: root refcnt 2 rate 2500Kbit burst 5Kb peakrate 3Mbit
minburst 1539b lat 70.0ms
I left the ens2 NIC in its default state:
root@rorem-HP-Compaq-8000-Elite-SFF-PC:~# tc qdisc show dev ens2
qdisc fq_codel 0: root refcnt 2 limit 10240p flows 1024 quantum 1514
target 5.0ms interval 100.0ms memory_limit 32Mb ecn
****
When I examine the NICs using the
iptraf-ng tool, it shows no difference
(in the bit rate) before or after running
the script setting up the ens1 tbf queue at
the ens2 (output) interface.
Can anyone help point me to what I'm
doing wrong, or misunderstanding?
thanks,
Doug Rorem
