KNOWN-ISSUES

General
    - The nav bar allows you to step back in the process, but not jump forward. The user must click next to continue within the workflow, since this
    action triggers the 'save' of state
    - TODO: need to rationalise the css classes - reduce any duplication, and misnomers
    - the tool doesn't account for a pre-existing inventory
    - TODO: the plugin doesn't flag an error if the runner-service is not there! 
    - cockpit -> flask. On older versions of flask, the cockpit.http calls receive a RST (reset) packet from flask, so calls to the API fail. When this
    happens the UI stops working due to the js error. flask versions found to work flask >= 1.0.2, flask_restful 0.3.6
    - ceph-ansible provides .sample named playbooks that need to be copied/renamed before use
      - this should be done by the plugin post-install script
    - test is hitting a j2 template error with undefined ansible variable disk_list. Workaround applied in 
      ansibleMap.js - KNOWN ISSUE in ceph-ansible, resolved on Master (Seb)
    - docker pull timeout shows this error
      ceph-3 failed in role 'ceph-container-common', task 'pulling docker.io/ceph/daemon:latest image'. Error msg - non-zero return code
      WORKAROUND: all.yml now written with 600s timeout (default is 300s)

hostspage.jsx
    - TODO: if the probe playbook fails, there are no indications other than looking at the browser console log
    - Once collocation rules are applied, and the inventory created, the user could go back to the environment page and
      select a different cluster type, which would not match with the new collocation rules. If the user selects a different cluster or install
      type, the inventory should probably be wiped and they should start again from scratch
    - if you get ssh auth issues, the modal dialog presents the path to the prod mode ansible-runner-service file - even when the daemon is actually
      running in dev mode
    - BUG: using a mask that has 0 prefixed numbers doesn't work. The first number in the series is correct, but subsequent numbers drop the '0'.
      
validatepage.jsx
    - BUG: selecting rows, also triggers the requirement to probe again even when the roles haven't changed
    - TODO: for testing checks are currently bypassing with skipChecks=true in the class definition - which just allows the 
            next button to work, without any validity checks
    - TODO: Add some 'business logic' to analyse the cluster as a whole. checkrole focuses only at the per host level.
            if one osd node has ssd and hdd, they all must for example
            all monitors on the same public network
            is there a min of 3 mons defined
            is there a min of 3 osd hosts defined
            total host count >= 5 in prod mode or 3 in POC mode

networkpage.jsx
    NIC bandwidth is reported as unknown, when ansible returns a speed of 0.
        this happens on some interfaces; docker bridge, libvirt bridge or when the target machine is a VM

ansibleMap.js
    - allVsars() : for rgw use case there are a number of settings that would be beneficial, but adding them
      requires changes to the jinja2 template for ceph.conf first
      - "objecter_inflight_op_bytes": 1048576000,
      - "objecter_inflight_ops": 102400
      - "rgw_ops_log_rados": false,
      - "rgw_dynamic_resharding": false