ERROR: juju.worker exited: setting up container dependencies on host machine: could not find unused subnet
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Expired
|
High
|
Unassigned |
Bug Description
OS: Ubuntu 16.04 LTS
Juju Version: 2.7.3
I have deployed the Openstack base and it is able to deploy to all the bare-metal server, however anything on a lxd container is not starting to be deployed. Running juju debug-log this is the error received for all machines.
machine-2: 20:48:26 INFO juju.utils.
machine-2: 20:48:27 INFO juju.tools.
machine-2: 20:48:27 INFO juju.container.lxd LXD_IPV4_ADDR is not set; searching for unused subnet
machine-2: 20:48:27 WARNING juju.provisioner not stopping machine agent container watcher due to error: setting up container dependencies on host machine: could not find unused subnet
machine-2: 20:48:27 ERROR juju.provisioner starting container provisioner for lxd: setting up container dependencies on host machine: could not find unused subnet
machine-2: 20:48:27 INFO juju.worker stopped "2-container-
machine-2: 20:48:27 ERROR juju.worker exited "2-container-
That just loops over and over. The network address for everything is 10.0.0.0/16 and my MaaS server is providing the DHCP and a gateway is 10.0.0.1
I had this exact setup for my previous version of OpenStack also deployed with Juju a few months ago, but I just wanted a clean start. I have attached a screenshot of the debug-log as well as the output of juju status. At this point it has been running for 30 minutes and will go no further.
summary: |
- Juju error setting up container + ERROR: juju.worker exited: setting up container dependencies on host + machine: could not find unused subnet |
Changed in juju: | |
importance: | Undecided → High |
status: | New → Incomplete |
We don't have a 2.7.3 version of Juju, so I'm guessing this was 2.3.7.
It seems we're trying to initialize an lxdbr0 even though we wouldn't be using it for containers that are being deployed on maas managed machines.
Do you know what version of Juju you were running while it was working in the past? I do believe the auto-detection- of-subnet code scans through 10.0/16 looking for an unused subnet, which you obviously have already preallocated to MAAS.