juju deploy ceph hangs in pending in maas environment

Question

I have ubuntu 14.04.1 with maas 1.7.1 rc4 and juju 1.20.14. My client machine is commissioned and deployed. I can ssh [email protected] and I can juju ssh 1 successfully. I did "juju deploy --config config.yaml ceph --to 1".

juju status ceph looks like:

juju status ceph
environment: maas
machines:
  "1":
    agent-state: pending
    dns-name: cl1stn2.maas
    instance-id: /MAAS/api/1.0/nodes/node-caa78cae-a5d5-11e4-b918-180373b04ac9/
    series: trusty
    hardware: arch=amd64 cpu-cores=16 mem=24576M tags=auitsdisk
services:
  ceph:
    charm: cs:trusty/ceph-31
    exposed: false
    relations:
      mon:
      - ceph
    units:
      ceph/0:
        agent-state: pending
        machine: "1"
        public-address: cl1stn2.maas
networks:
  maas-em1:
    provider-id: maas-em1
    cidr: 10.20.0.0/24

I don't see any attempt to contact cl1stn2.maas and the status stays in pending for ever. I can ssh to cl1stn2.maas and juju ssh 1 successfully. I don't see anything in the all-machines.log.

Where do I look to find out what it is waiting for?

Answer 1

Juju attempts to start a new machine and use cloud-init to inject the script necessary to download and start a juju agent process. This process then contacts the juju state server (machine 0) and check in. The status remains in pending forever until the client checks in. If anything goes wrong with this process then the machine stays in pending state forever. The simplest solution with juju and it's cloud thinking, is to just remove the unit and wait for it to clean up and power down the server and then add a new unit and let it deploy it all again.
If it continues to fail after several retries, then it is more than likely that there is something wrong with your networking configuration.