MAAS Deployment KO with bonding

[Flint WALRUS]

Hi guys, I'm looking for some help regarding an issue that I'm facing with
MAAS at deployment step while trying to use bonding.

Here is my MAAS installation:

MAAS allInOne server (RackD+RegionD):
1x1Gbps NIC for BMC management. DHCP and DNS under MAAS supervision.
2x10Gbps NIC for PXE vlan management. VLAN DHCP and DNS under MAAS
supervision.

PXE VLAN = ID100
BMC VLAN = ID101
SRV VLAN = ID200

4x DELL PE R720 servers:
1xBMC 1Gbps (iDRAC 7 Enterprise) connected to the BMC VLAN configured to
get IP from DHCP and register itself with DNS.
4x10Gbps NIC.

4x HPE DL390 Gen9 servers:
1xBMC 1Gbps (iLO 4 HPE Enterprise) connected to the BMC VLAN configured to
get IPs from DHCP and register itself with DNS.
4x10Gbps NIC.

I'm currently heavily testing MAAS and I'm able to get all my servers
correctly detected, commissioned and the rescue mode is working fine.

However I stumble upon something weird while trying to deploy my server
with the following network configuration:

Server network topology:
eno1 and eno2 are aggregated as Bond0.
eno3 and eno4 are aggregated as Bond1.

Bond0 interface is set with LACP fast and XMIT policy to Layer2+3.
Bond0 interface is untagged, subnet unset and do not request for IP.
Bond0 is flagged as PXE Interface.

Bond1 interface is set with LACP fast and XMIT policy to Layer2+3.
Bond1 interface is untagged, subnet unset and do not request for IP.

Bond0 interface get attached to VLAN 200 with subnet CIDR 10.0.200.0/24.
Bond1 interface get attached to VLAN 101 with subnet CIDR 10.0.101.0/24.

Unfortunately, I'm unable to complete the deploy process because once the
server is install and boot bonds interfaces are created but it hang out
with the error: *a start job is running for raise network interfaces.*

Once the 5 mins countdown finish the cloud-init script fail with the
following error message: *Didn't find any datasource,
DataSourceNotFoundException.*

At this point I'm unable to know if this error is caused by the network
issue or if the network issue is caused by the node not able to contact the
MAAS Metadata.

What's weird is that when I'm booting in rescue mode, and mounting the node
LVM Volume and looking at the syslog it show me that bonds and nic are up
and running and working fine.

I've try to boot under rescue mode and to manually configure the nics,
unfortunately, I'm not able to do so as the Rescue Mode ephemeral image
doesn't load the 8021q module at all even if the vlan package is already
installed.
After looking at the module volume and the kernel module directory, it
seems that the 8021q directory is not there. As all my VLANs need to be
tagged I'm kind of stuck right now.

So, if someone is able to get me some hints on this topic or could give me
a hand investigating with this issue, I'll would be more than happy!

Thanks a lot!
G.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/maas-devel/attachments/20170821/d1212946/attachment.html>