Lagg interface anyone?

[Patrick M. Hausen]

Hi all,

has anyone managed to configure a lagg interface with the TN SCALE alpha?

When I use the console to create a Lagg, both physical interfaces can be added but no lagg0 appears. Then adding an IP address or a VLAN tag does not work because there is no interface but lo.

Thanks,
Patrick

 

[HPloco]

I found out the hard way that you can only make 1 change to the network per reboot. I currently have a bond0, you have to name it bond#, they are the onboard NIC's from a X8DT6-F

 

[Patrick M. Hausen]

Using bond0 (and using the UI instead of the console menu to start with) finally did the trick. Thanks.
Silly me, I thought I could configure the switch in advance and set up everything at the console without initial network connectivity ...

Yes, I know, this is supposed to be a work in progress. Port-channel working, everything's well.

 

[Patrick M. Hausen]

Related - do you happen to know why the network interfaces are called ens161 and ens192, now? I am familiar with enp0s4 after eth0 was gone for $reasons, but this is new - again. Can't they keep an API for a couple of years in penguinland?

And what do the numbers mean and can I expect them to remain constant at least?

 

[rssfed23]

And what do the numbers mean and can I expect them to remain constant at least?

The numbers changed because your hardware did I imagine or the distribution used changed its configuration slightly (as vendors can choose which ordering scheme takes dominance and there are some differences between systemd-networkd and other network managers).
The API is the same as when things first switched away from ethX. The $reasons are valid ones with the new scheme making the lives of multi-homed sysadmins easier, once they get used to it.

You can read about it in more details at https://en.wikipedia.org/wiki/Consistent_Network_Device_Naming, but in essence:

• Onboard interfaces at firmware index numbers eno[1-N]
• Interfaces at PCI Express hotplug slot numbers ens[1-N]
• Adapters in the specified PCI slot, with slot index number on the adapter enp<PCI slot>s<card index no>
• If firmware information is invalid or rules are disabled, use traditional eth[0-N]

And yes, you can expect them to remain constant. That's the whole point of using them instead of ethX they are guaranteed to remain constant upon reboot unless you change hardware or linux distro.