On 3.4.2, if a known machine’s VLAN is changed on the switch, with the next PXE boot attempt the whole (known) subnet that machine is part of is moved to another VLAN, affecting all other machines in that subnet and old VLAN.
The machine itself is then not finishing the PXE boot as DHCP doesn’t work correctly. In addition, for the machines in the old subnet, the DHCP server now assigns the DNS address of the new subnet.
Putting the machine back on the old VLAN now moves the subnet that was in the new VLAN to the old VLAN, breaking the DHCP for the subnet in the new VLAN.
The only way to prevent this is to delete the machine before moving to another VLAN. Is there any other way to deal with this? We put machines into different VLANs based on user privilege. When the machine gets assigned to a user with a different privilege level than before, it breaks the DHCP for the whole subnet.
Steps to recreate:
Create VM test in VLAN 6 with subnet 10.60.0.0/24. Let MAAS install via PXE.
Now move test to VLAN 5, which has subnet 10.50.0.0/24 and reboot two times. Now the VLAN and subnets in MAAS are not correct anymore and test doesn’t PXE boot anymore.
