January 21Jan 21 I run Unraid 7.2.0 on a Dell T440 server with two Gigabit Ethernet ports. They are bonded as bond0 using 802.1ad LAG. The two ports connect to a Cisco SG200-26 26-Port Gigabit Smart Switch, on ports GE25 and GE26. They are also configured as LAG. When I first built the system, I did an informal test copying large files from network hosts to the array. On the Dashboard Interface, I saw the Inbound go well above 1000 Mbps, so I thought Bonding was OK. Now I've run a more precise test with iperf. Running a single iperf test with the client on a system on the LAN to the Unraid Server running iperf server shows throughput of 950 Mbps. I expected running simultaneous iperf (from separate hosts on the LAN to two different iperf server ports on Unraid) would show a speed around 1900 Mbps due to LAG. What actually happens is each iperf test slows down to about 300 Mbps when run concurrently. Gemini suggested changing Unraid xmit_hash policy. It is currently layer2:root@T440:~# cat /sys/class/net/bond0/bonding/xmit_hash_policylayer2 0The recommendation is ifconfig bond0 down echo layer3+4 > /sys/class/net/bond0/bonding/xmit_hash_policy ifconfig bond0 upI have not done that because the only way I access Unraid is by the network, so I'd have to add a monitor and keyboard to retain connectivity. Is this the first thing to look at for poor network performance with Bonding and Lag?
January 22Jan 22 Community Expert Bonding is just bad. It doesn't do much if anything to increase real world performance. You would be better served to install a higher speed nic if supported by your network.
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.