Comment 6 for bug 1921658

I've tried replicating this issue.

In my setup (bionic, 2.8.6 (8602-g.07cdffcaa-0ubuntu1~18.04.1) with logical storage pool setup as per the bug.

My setup consistently fails with
2021-06-08 18:31:41 provisioningserver.rpc.pods: [critical] rare-kodiak: Failed to compose machine: RequestedMachine(hostname='sure-cattle', architecture='amd64/generic', cores=1, memory=2048, block_devices=[RequestedMachineBlockDevice(size=8000000000, tags=['maas_guest_lvm_vg'])], interfaces=[RequestedMachineInterface(ifname=None, attach_name=None, attach_type=None, attach_options=None, requested_ips=[], ip_mode=None)], cpu_speed=None, known_host_interfaces=[])

In
File "/usr/lib/python3/dist-packages/provisioningserver/drivers/pod/virsh.py", line 815, in configure_pxe_boot
            doc = etree.XML(xml)

There pexpect appears to be consistently returning what was input to the 'virsh #' prompt as well as the output. On occasion appears to be returning a truncated version of the input commencing with '<' or some left over output from a previous command.

This is the rackd log with added print statements:
https://paste.ubuntu.com/p/7Kyj3hN66K/

Both the volume and the vm are defined/created on the node. Sometimes they even manage to partially commission despite the UI error.