Ceph Fact

The ceph-fact tool is an ansible/puppet facter script based on 42on to gather diagnostic information from a Ceph cluster.

The output from the tool is json to assist in writing ansible/puppet for managing your ceph cluster.

This tool does NOT collect any user (object) data contents nor authentication credentials from a Ceph cluster.

Requirements

To run the tool, the following requirements need to be met:

Python 3.5 or higher
python-rados, ceph and ceph-common RPM or DEB installed
client.admin keyring present in /etc/ceph
/etc/ceph/ceph.conf configured to connect to Ceph cluster

You can test this by running:

ceph health

This should output either HEALTH_OK, HEALTH_WARN or HEALTH_ERR.

Where to run

This tool should be run on a machine which is able to connect to the Ceph cluster and has the client.admin keyring.

In most of the situations the monitors are the right location to run this tool.

Running

When the requirements are met, simply run the ceph-fact tool:

root@mon01:~# ./ceph-fact.py
{
  "ceph": {
    "status": {
      "fsid": "50c175aa-18d2-4182-899a-95a73cafe6da",
      "health": {
        "status": "HEALTH_OK",
        "checks": {},
        "mutes": []
      },
      "election_epoch": 1380,
      "quorum": [
        0,
        1,
        2,
        3,
        4
      ],
      "quorum_names": [
        "marvin06",
         ...
root@mon01:~#

ansible examples

wait 1000 seconds until cluster is healthy

  tasks:
    - name: "Waiting til cluster is healthy!!!"
      ansible.builtin.setup:
        filter: "facter_ceph"
      delegate_to: "{{ cluster_health_host }}"
      until:
        - facter_ceph['status']['health']['status'] == 'HEALTH_OK'
      retries: 100
      delay: 10

check that there are more than one ceph-manager

    - name: "Fail, if there is only one ceph-manager!"
      fail:
        msg: "There must be at least 2 Managers!"
      when:
        - lookup('ansible.builtin.env', 'FAILOVERRIDE') | default("") | length  == 0 
        - facter_ceph['mgr_metadata'] | length < 2

Fail the active ceph manager when working on it

    - name: "Failover to another Ceph-Mgr if we want to reboot the active one"
      shell: "ceph mgr fail {{ item.name }}"
      when:
        - facter_ceph['mgr_stat']['active_name']  == item.name
      loop: "{{ facter_ceph['mgr_metadata'] | selectattr('container_hostname', 'equalto', inventory_hostname_short) }}"

License

This tool is based on [https://github.com/42on/ceph-collect] by 42on to help Ceph customers quickly.

The tool is free to use for all other purposes. It's licensed GPLv2, so please send changes back to us!

pengutronix/ceph-fact