Interface naming

I am having some issues with ethernet interfaces not being consistently numbered:

For example after the latest boot:

[   22.263298] fsl_dpaa2_eth dpni.7 eth0: configuring for inband/qsgmii link mode
[   22.300689] fsl_dpaa2_eth dpni.8 eth1: configuring for inband/qsgmii link mode
[   22.335246] fsl_dpaa2_eth dpni.9 eth2: configuring for inband/qsgmii link mode
[   22.368789] fsl_dpaa2_eth dpni.6 eth3: configuring for inband/qsgmii link mode
[   22.402577] fsl_dpaa2_eth dpni.5 eth4: configuring for inband/qsgmii link mode
[   22.436457] fsl_dpaa2_eth dpni.4 eth5: configuring for inband/qsgmii link mode
[   22.471804] fsl_dpaa2_eth dpni.3 eth6: configuring for inband/qsgmii link mode
[   22.506846] fsl_dpaa2_eth dpni.2 eth7: configuring for inband/qsgmii link mode
[   22.541962] fsl_dpaa2_eth dpni.1 eth8: configuring for inband/10gbase-r link mode
[   22.575834] fsl_dpaa2_eth dpni.0 eth9: configuring for inband/10gbase-r link mode
[   84.647351] fsl_dpaa2_eth dpni.7 eth0: Link is Up - 1Gbps/Full - flow control rx/tx

eth0 in this case corresponds to the port labled ge2. This is a bit annoyig.

I will see if I can come up with some udev rules to have consistent naming (preferably with the same names as the box labels).

Has someone encountered (and solved?) this problem already?

It sounds like this issue first reported by SUSE. There is a udev patch for this.

Bug 1181614 - udev does not support predictable names for DPAA2

The firmware is set up so they will always be enumerated in proper order but it sounds like udev is deciding to re-order on it’s own

So the easiest would just be to tell udev to not reorder them, right?