MAAS wants to do a dhcp probe on a slave interface in a bond
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MAAS |
New
|
Undecided
|
Unassigned |
Bug Description
I see logs indicating MAAS is trying to do a dhcp probe using interfaces that are part of a bond. I *think* it's correct to say that MAAS should only use the bond, not the individual slave interfaces.
ifenslave shows that some interfaces are part of bond-lan:
$ ifenslave -a
1: lo: <LOOPBACK,
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: em1: <BROADCAST,
link/ether 44:a8:42:3a:63:c0 brd ff:ff:ff:ff:ff:ff
3: em2: <BROADCAST,
link/ether 44:a8:42:3a:63:c1 brd ff:ff:ff:ff:ff:ff
4: em3: <BROADCAST,
link/ether 44:a8:42:3a:63:c2 brd ff:ff:ff:ff:ff:ff
5: em4: <BROADCAST,
link/ether 44:a8:42:3a:63:c3 brd ff:ff:ff:ff:ff:ff
6: p1p1: <BROADCAST,
link/ether a0:36:9f:74:68:c2 brd ff:ff:ff:ff:ff:ff
7: p2p1: <BROADCAST,
link/ether a0:36:9f:74:68:c2 brd ff:ff:ff:ff:ff:ff
8: p1p2: <BROADCAST,
link/ether a0:36:9f:74:68:c2 brd ff:ff:ff:ff:ff:ff
9: p2p2: <BROADCAST,
link/ether a0:36:9f:74:68:c2 brd ff:ff:ff:ff:ff:ff
10: bond0: <BROADCAST,
link/ether e2:50:f9:ad:f1:87 brd ff:ff:ff:ff:ff:ff
11: bond-lan: <BROADCAST,
link/ether a0:36:9f:74:68:c2 brd ff:ff:ff:ff:ff:ff
12: conjureup0: <BROADCAST,
link/ether f6:8f:b0:43:c4:5a brd ff:ff:ff:ff:ff:ff
I think this bug is essentially duplicate of bug #1633717. Bond parents shouldn't have IP addresses assigned to them, so (for now) should be skipped.
After MAAS gains support for DHCP detection on interfaces without an IP address assigned, we'll need to keep this in mind though. I agree that we should probably not try to probe bond parents. (the only reason it would help is if the bond was somehow misconfigured, but then the results would be misleading anyway.)