Configuring LACP

    • OMV 2.x
    • Configuring LACP

      While I found some posts here discussing LACP/trunking/bonding, I really have not found a guide/instructions how to set it up from the start.

      I have a HP microserver Gen8 with OMV 2.2.5. Both NICs are connected to a Netgear smart managed switch (GS108Tv2). (The third NIC for ILO is connected to an unmanaged switch).

      I actually don't have a communication partner, that uses LACP. Nevertheless, from my understanding, LACP should help me to get better performance, when serving two or more single NIC clients. (I am well aware, that there are other restrictions. Now I get about 100+ MByte/s read/write speed from one disk, I would expect, that another disk should yield about the same speed in parallel. I use 4 HDs, no RAID, no logical volumes).

      For installing OMV I selected eth0 (there was no choice, to select two NICS, at least, I did not find one). For the moment, I am using DHCP, however I think to switch to static adresses soon. In the WebGUI I can add eth1, but I guess, that won't buy much or even make things fail because the other NIC would get a separate IP in the same net with the options I saw.

      So my question: Can anybody show or give a link, how to set this up from scratch? (at least the OMV side, for the switch side the manual seems to give sufficient information)
      OMV 4.1.13-1 (typically everything up to date), only plugin: flash memory; HP Microserver, 4 internal ext4 HDDs, SSD for OS, SD-Card for booting (can't boot on SSD with 4 HDD used …), external USB3 HDDs (ext4 + NTFS)
    • Thankyou very much for your answer. My switch does support LACP. The description behind your link seems easy enough to follow. However, I fear, this wouldn't work without any problems under OMV.

      As far as I understand, the /etc/network/interfaces file is created by OMV from information given inside the WebGUI (and during installation).

      When I edit /etc/network/interfaces, wouldn't it get overwritten again at some point?
      Am I safe, when I don't touch the interfaces inside the GUI?

      (I am very new to OMV, so I am not sure about its inner workings)
      Perhaps I'll just go ahead and try ...

      Update: For the moment, I tried mode bond 0, seems to work very well. In very limited tests, I got cummulated transfer rates of around 170 MByte/s - about what I had expected and more than would be possible without bonding on 1 GB NICs. Maybe bond 0 is even best for me (without clients, that use bonding) and I really don't want LACP besides my curiosity.

      Concering my doubt about automatic generation of /etc/network/interfaces and possible interference with manual editing this file, I found Customize /etc/network/interfaces the OMV way [>= 1.11] . I'll rearead this - it is not clear yet to me, how to handle this. Nevertheless, happy that the first tests showed excellent performance.

      OMV 4.1.13-1 (typically everything up to date), only plugin: flash memory; HP Microserver, 4 internal ext4 HDDs, SSD for OS, SD-Card for booting (can't boot on SSD with 4 HDD used …), external USB3 HDDs (ext4 + NTFS)

      The post was edited 2 times, last by buers ().