"Failed to create temporary directory" error with incus-deploy

Incus
1

Hello all,
I’ve been trying to install incus on a 4 node homelab using incus-deploy and keep running into an issue. I swear it’s something simple I’m missing, but have yet to find it.
I’ve installed Ubuntu 22.04 on each and am able to use ansible to update the packages and network configuration. When I start to use the ansible-deploy, it fails. If I add -vvv to the ansible-playbook command, I get the following error:
“msg”: “Failed to create temporary directory. In some cases, you may have been able to authenticate and did not have permissions on the target directory. Consider changing the remote tmp path in ansible.cfg to a path rooted in "/tmp", for more error information use -vvv. Failed command was: ( umask 77 && mkdir -p "echo ~/.ansible/tmp"&& mkdir "echo ~/.ansible/tmp/ansible-tmp-1722203858.4888856-46107-89077470102481" && echo ansible-tmp-1722203858.4888856-46107-89077470102481="echo ~/.ansible/tmp/ansible-tmp-1722203858.4888856-46107-89077470102481" ), exited with result 1”

I’ve tried this on 4 physical nodes with Ubuntu22.04 installed from MAAS. I’ve tried the same thing from 4 virtual nodes all running 22.04 installed from ISO. I get the same error.

Other things I’ve tried with no success:

  1. setting ansible_remote_tmp: ‘/tmp’ - this changes the path included in the error, but does not fix it.
  2. changing ansible_user from root to ubuntu
  3. cloning and running ansible from different servers

I’m just not sure what I’m missing or what the next step is.

Thanks for taking a look and I appreciate any help I can get with this.

2

Can you show your hosts.yaml?

3

Thanks for the quick reply!

These “servers” are Optiplex 3050 with only one NIC. So I created a VLAN interface with VLAN10 for OVN setup. There’s minimal local storage so I removed the ceph bits. I’m connecting using lvm to a TrueNAS iSCSI share.
I’ve adjusted the server names and IP range of course, but the rest is what I’m using.

all:
  vars:
    incus_name: "incus-dev"
    ovn_name: "incus-dev"
    incus_release: "stable"
    ovn_release: "ppa"
  children:
    baremetal:
      vars:
        ansible_connection: incus
        ansible_incus_remote: local
        ansible_user: ubuntu
        #ansible_user: root
        ansible_become: no
        ansible_incus_project: dev-incus-deploy
        #ansible_remote_tmp: '/tmp/ansible/'
        incus_roles:
          - cluster
          - ui
        incus_init:
          network:
            LOCAL:
              type: macvlan
              local_config:
                parent: br-enp1s0
            UPLINK:
              type: physical
              config:
                ipv4.gateway: "192.168.10.0/26"
                ipv6.gateway: "fd00:1e4d:637d:1234::1/64"
                ipv4.ovn.ranges: "192.168.10.30-192.168.10.40"
                dns.nameservers: "192.168.1.1"
              local_config:
                parent: br-enp1s0.10
            default:
              type: ovn
              config:
                network: UPLINK
              default: true
          storage:
            local:
              name: sharediscsi
              driver: lvmcluster
              local_config:
                source: "incus_vg"
              default: true
        ovn_roles:
          - host
      hosts:
        server1.local:
          ovn_roles:
            - central
            - host
        server2.local:
          ovn_roles:
            - central
            - host
        server3.local:
          ovn_roles:
            - central
            - host
        server4.local:
4

Right so the issue is that you’re trying to connect to them using the Incus connection plugin instead of SSH.

Remove all the ansible_ lines from the config and instead put in:

ansible_user: ubuntu
ansible_become: yes

That’s all you should need for SSH to do its thing. The ansible_connection is the main issue in your config and the other two ansible_incus_ entries become pointless when not using the Incus connection plugin.