Sometimes, we loose forwarding on caps

It looks like the data-path on the CAP isn't correctly provisioning, as soon as you select the affected CAP remove it from the (static) interface list and push (re-)provision, the dataforwarding is restored again, no need to reboot the cap itself.

Our setup is as follows:
Main CAPsMAN controller
2 datapaths, local-forwarding:
* 1 untagged
* 1 tagged

When the forwarding blocks, only the tagged datapath is affected, the plain untagged datapath still works.

We've time to time observed this behavior seemingly random on all location, the CAPsMAN controllers are different (CCR,RB1100AHx4, etc) and caps ranging from 18 (small) to 300 (larger) devices.

Anyone seen this behavior as well?