Swift

This is a role for setting up and managing OpenStack Swift clusters. Currently it supports PACO nodes running all Swift services; Proxy, Account, Container and Object.

It will:

• Prepare Swift nodes, including setting SELinux and ensuring SSH access
• Add repositories and install packages
• Configure dependent services, such as logging, rsyncd and memcached
• Configure keepalived for failover of proxy VIPs
• Configure Swift PACO services
• Create initial account, container and object rings
• Prepare disks on each node, format and mount according to the rings
• Build and distribute the rings
• Configure dispersion
• Simple operational tasks such as:
  • Update and distribute the rings
  • Re-configure PACO services
  • Generate dispersion and replication reports

Requirements

An inventory of pre-existing CentOS 8 Stream nodes with their networking configured. The cluster requires an admin box (included in swift_admin group) from which all the Swift nodes are managed. This can be one of your Swift nodes, if you don't have a separate admin host.

If building a virtual Swift cluster, consider using csmart.virt_infra Ansible role at https://github.com/csmart/ansible-role-virt-infra.

Sample inventory and playbooks can be found at https://github.com/csmart/virt-infra-swift for both csmart.virt_infra and csmart.swift.

Role Variables

This role has a number of default variables, which are broken out into individual files under defaults/main/. These include common settings for a Swift cluster, as well as defaults for specific Swift services.

• account-rings.yml
• account.yml
• container-rings.yml
• container.yml
• dispersion.yml
• hash.yml
• networks.yml
• object-rings.yml
• object.yml
• packages.yml
• proxy.yml
• swift.yml
• tempauth.yml

Required global variables that the user must define, include:

• swift_hash_suffix - the hash suffix of the cluster, once set it should not be changed
  • default is 07b4ef9c-2e01-4ea2-a109-5ffc5273225f
• swift_hash_prefix - the hash prefix of the cluster, once set it should not be changed
  • default is f9175259-ace0-48bb-af9d-e7ac505b89d2
• swift_outward_subnet - the routable CIDR subnet for external connections (for proxy nodes)
  • default is 203.0.113.0/24
• swift_cluster_subnet - the cluster communication CIDR subnet
  • default is 192.0.2.0/24
• swift_replication_subnet - replication CIDR subnet (can be the same as cluster)
  • default is 198.51.100.0/24

Required node specific variables that the user must define, include:

• swift_outward_ip - IP on the outward network

  • for example, 203.0.113.11

• swift_cluster_ip - IP on the cluster network

  • for example, 192.0.2.11

• swift_replication_ip - IP on the replication network

  • for example, 198.51.100.11

• swift_vips - list of proxy VIPs, 4th octet of IP

  • Each proxy node should list the 4th octet of the VIP IP in order of preference. For example, based on default outward subnet of 203.0.113.0/24, here's a node which wants VIPs 203.0.113.111, 203.0.113.112 and 203.0.113.113.

    swift_vips:
      - 111
      - 112
      - 113
    

• swift_rings_disks - list of dicts defining what disks to use for what ring

  • Each node's disks should include the path and what weight it should have for a ring. For example, here's a SCSI disk to be used for objects and an NVMe to be used for both acount and container.

    swift_rings_disks:
      - disk:
          device: sdb
          rings:
          - name: account
            weight: 0
          - name: container
            weight: 0
          - name: object
            weight: 100
      - disk:
          device: nvme0n1
          rings:
          - name: account
            weight: 100
          - name: container
            weight: 100
          - name: object
            weight: 0
    

Dependencies

None.

Example Playbook

The virt-infra-swift repository at https://github.com/csmart/virt-infra-swift provides a set of example playbooks and sample inventory.

Your inventory should include the following groups:

• swift (or all)
• swift_admin
• swift_proxy
• swift_account
• swift_container
• swift_object

Here is an example inventory for a three node PACO Swift cluster with required groups.

swift:
  hosts:
    swift-admin:
    swift-[01:03]:
  children:
    swift_admin:
      hosts:
        swift-admin:
    swift_proxy:
      hosts:
        swift-[01:03]:
    swift_account:
      hosts:
        swift-[01:03]:
    swift_container:
      hosts:
        swift-[01:03]:
    swift_object:
      hosts:
        swift-[01:03]:

Once you have a basic inventory, creating a basic playbook should be simple.

---
- hosts: swift
  tasks:
    - include_role:
        name: csmart.swift

Executing the playbook will run the whole role and default set of tasks, in order.

ansible-playbook -i inventory/ site.yml

The role also includes tags for each type of task, so you could create one or more playbooks for all or specific tags.

---
- hosts: swift
  tasks:
    - include_role:
        name: csmart.swift
      tags:
        - account
        - common
        - config
        - container
        - disks
        - dispersion
        - hosts
        - keepalived
        - logging
        - memcached
        - object
        - packages
        - prep
        - proxy
        - rings
        - rsyncd
        - selinux
        - services
        - system
        - update

Then run the playbook against specific tags, for example, to just re-configure account services.

ansible-playbook -i inventory/ site.yml --tags account

License

GPLv3+

Author Information

Chris Smart https://blog.christophersmart.com