lag corner case: active-standby mlag without any lag specific config on the switches

[jbemmel]

PR #1624 contains a special use case involving active-standby lag to a pair of switches.
This use case can work without any lag support on the M side

It is a corner case, and I could imagine we may not want to mix it up with adding Linux lag support

Thoughts / opinions? Too much of a corner case to worry about?

If we do decide to support it: How would we model this? Some kind of dummy interfaces on the M-side that get removed during cleanup perhaps?

[ipspace]

We had that discussion already (#1504). Do you think something has changed in a month?

I told you that active-standby bonding in Linux is not LAG, and I told you to implement it in a plugin. This will not get merged as part of the lag module because it is not LAG (and I just realized we might have to adjust the LAG module documentation).

[jbemmel]

Then we should probably remove ‘’’active-backup’’’ as a lag mode option value too?

I can make the changes, probably Friday

[ipspace]

Then we should probably remove ‘’’active-backup’’’ as a lag mode option value too?

Is it used anywhere else? If not, let's remove it (although active-backup is a configuration in some LAG implementations).

I can make the changes, probably Friday

No rush. Add them to the Linux LAG PR whenever that one is ready.