ceph-osd does not configure devices if cluster and access network are different

[Narinder Gupta]

If ceph public and cluster network on different subnet then ceph-osd devices does not get initialized.

Here is the config I was trying. If i change the public-api binding to internel-api binding everything works.

    ceph-mon:
      charm: "./xenial/ceph-mon"
      num_units: 3
      bindings:
        public: public-api
        cluster: internal-api
      options:
        monitor-count: 3
        fsid: ecbb8960-0e21-11e2-b495-83a88f44db01
        monitor-secret: AQD1P2xQiKglDhAA4NGUF5j38Mhq56qwz+45wg==
        source: "cloud:xenial-updates/newton"
      to:
        - "lxd:nodes/4"
        - "lxd:nodes/2"
        - "lxd:nodes/3"
    ceph-osd:
      charm: "./xenial/ceph-osd"
      num_units: 3
      bindings:
        public: public-api
        cluster: internal-api
      options:
        osd-devices: /dev/sdb
        osd-reformat: 'yes'
        source: "cloud:xenial-updates/newton"
      to:
        - "nodes/0"
        - "nodes/1"
        - "nodes/2"
    ceph-radosgw:
      charm: "./xenial/ceph-radosgw"
      num_units: 1
      bindings:
        internal: internal-api
        admin: admin-api
        public: public-api
      options:
        region: RegionOne
        operator-roles: "Member,admin"
        source: "cloud:xenial-updates/newton"
        ceph-osd-replication-count: 3
      to:
        - "lxd:nodes/0"

[David Ames (thedac)]

Narinder,

If this is still a problem, please provide the logs from all ceph-osd and ceph-mon nodes for us to debug further.

Marking incomplete. Please add the logs and mark as new.

[James Page (james-page)]

Marking Invalid as no updates in nearly 2.5 years