portworx/ansible-portworx-defaults

Ansible role for deploying Portworx

Install Portworx through Ansible

[ experimental ] Portworx is typically deployed through/with the container orchestrator. For Kubernetes, Portworx is deployed as a daemonset through the Portworx installer For DCOS, Portworx is deployed through Universe

This ansible script is intended to help install Portworx for HashiCorp Nomad, Docker Swarm, or for other environments where Portworx is to be deployed directly.

Sample Playbook

Here is a sample playbook, that uses the portworx role to create a custom ClusterID and install Portworx on all nodes:

---
- hosts: localhost
  tasks:
  - name: Set Unique ClusterID Once
    command : "uuidgen"
    register: clusterID
- hosts: all
  vars:
    - kvdb: "-k etcd://70.0.1.2:2379"
    - data_if: "-d eth1"
    - mgmt_if: "-m eth1"
    - device_args: "-s /dev/sdb"
    - cluster_id: "-c {{ hostvars['localhost']['clusterID'].stdout }}"
  roles:
     - ../../ansible-portworx-defaults

The following variables are required:

• kvdb : IPaddress:port of 'etcd' or 'consul' kvdb
• cluster_id
• device_args : should correspond to list of devices (each with a -s), or should be -a to use all available devices

Please note that the corresponding prefix must be included in the variable clause (ex: -s /dev/sdb)

The cluster_id must be the same for all nodes in the Portworx cluster, but unique among any clusters sharing the same kvdb. The actual value can have a friendly name (ex: -c my-portworx-cluster)

For a complete list of command line arguments, please see https://docs.portworx.com/runc/options.html

Notes

The actual installation task is long running and does not provide incremental output

TASK [portworx : debug] ********************************************************
ok: [test1] => {
    "msg": "Command output does not display.  Command takes ~3-5 minutes to complete.  Please be patient ..."
}
...

Please be patient.

ansible-portworx-defaults's Issues

Idepondency of playbook

Greetings. I really doubt that launching bash script makes any use of this playbook at all. I mean what's the point of this? Just to bind this line?

curl -fsSL https://get.portworx.com | sh -s -- -k etcd://188.166.xx.xx:2379 -c portworx-swarm -d eth1 -m eth1 -a

If I can't launch playbook on top of it to make sure that everything is in the place, then there's no point to have automation. Why you need bash script if you have ansible, at least use some package manager if you want to maintain installation in one place. Now I've got a failed script in one of 20 nodes, and when relaunching playbook it fails to reinstall portworx. And I have no idea exactly in which place this failed.