@Viacheslav I can confirm the boot race issue is gone on the latest rolling release. Thanks!

Correction - change 2 above was too broad, and rule 3 in the resulting list with it.

Three changes came out of review and testing on PR #5392. Two of them narrow or widen what verify() accepts relative to the original description above, so the rules listed there are superseded by the list at the end of this comment.

A PR has been raised for this here: https://github.com/vyos/vyos-1x/pull/5422

I dropped a PR for this change, here: https://github.com/vyos/vyos-1x/pull/5392

This is the same boot-time interface-naming race tracked by T3871, here triggered by a slow-initializing NIC (an empty-cage Intel ice E823-L, which delays bring-up with a DDP-package load plus a firmware "module not present" media check) rather than a multi-vendor probe-order difference. Related interface-ordering reports for cross-reference:

A PR has been opened with these changes.

PR submitted for this: https://github.com/vyos/vyos-1x/pull/5296

Submitted a PR on this.

Looking at this a bit further, this appears to be related to the VRF. Something between earlier versions of 1.5 and now have made VRF'ing for the API either unstable or unusable. I can commit any key name succesfully as long as I don't have a vrf specified, and then if I come back in after the first commit and add the vrf, VyOS accepts it - however the performance of the API then becomes extremely spotty.

Sure thing!

I'll get a new build going and give this another test.

@c-po apologies for the delay, missed this over the holidays.