TrilioVault data protection

Important

This page has been identified as being affected by the breaking changes introduced between versions 2.9.x and 3.x of the Juju client. Read support note Breaking changes between Juju 2.9.x and 3.x before continuing.

Overview

TrilioVault is a data protection solution that integrates with OpenStack. It allows end-users to back up and restore point-in-time snapshots of their own workloads (cloud instances). TrilioVault is implemented via the Trilio charms.

Note

TrilioVault is not part of the OpenStack project. It is a commercially supported propriety product.

Prerequisites

  • Ubuntu 18.04 LTS or 20.04 LTS

  • OpenStack Queens, Stein, Train, Ussuri or Wallaby

  • an NFS server for snapshot storage or S3 compatible storage

  • a license (see the project’s homepage)

Supported combinations

The below table shows the support status for versions of TrilioVault for each version of Ubuntu and OpenStack.

Supported combinations

Ubuntu

OpenStack

TrilioVault 4.0

TrilioVault 4.1

TrilioVault 4.2

Bionic

Queens

Bionic

Stein

Bionic

Train

Bionic

Ussuri

X

Focal

Ussuri

X

Focal

Wallaby

X

X

Deployment

The TrilioVault solution consists of three core services:

  • TrilioVault Workload Manager: the main API - used to manage snapshots (e.g. creation and restores). This is implemented with the trilio-wlm charm.

  • TrilioVault Data Mover: deployed alongside OpenStack Nova - responsible for managing the instance snapshot process at the cloud level. This is implemented with the trilio-data-mover charm.

  • TrilioVault Data Mover API: the API for managing the TrilioVault Data Mover services - used by the Workload Manager. This is implemented with the trilio-dm-api charm.

An OpenStack Dashboard plugin is also provided, allowing for snapshot management to take place via a Web UI. This is implemented with the trilio-horizon-plugin charm.

The Trilio charms are now managed in channels within the CharmHub. The Trilio charms have a track which corresponds to the version of Trilio they are compatable with, so each charm has a 4.0, 4.1 and 4.2 channel. The charms are also compatable with the previous version of Trilio so that they can provide an upgrade path.

An overlay bundle is used to deploy to an existing OpenStack cloud. An example is provided below.

Note

Ensure that the value for openstack-origin matches the currently deployed OpenStack release.

The ceph-mon:client relation is only needed if the cloud is already configured to use Ceph-backed VM images (via either Cinder or Nova).

series: bionic

applications:

  trilio-wlm:
    charm: ch:trilio-wlm
    channel: 4.2/stable
    num_units: 1
    options:
      openstack-origin: cloud:focal-wallaby
      triliovault-pkg-source: 'deb [trusted=yes] https://apt.fury.io/triliodata-4-2/ /'

  trilio-data-mover:
    charm: ch:trilio-data-mover
    channel: 4.2/stable
    options:
      triliovault-pkg-source: 'deb [trusted=yes] https://apt.fury.io/triliodata-4-2/ /'

  trilio-dm-api:
    charm: ch:trilio-dm-api
    channel: 4.2/stable
    num_units: 1
    options:
      openstack-origin: cloud:focal-wallaby
      triliovault-pkg-source: 'deb [trusted=yes] https://apt.fury.io/triliodata-4-2/ /'

  trilio-horizon-plugin:
    charm: ch:trilio-horizon-plugin
    channel: 4.2/stable
    options:
      triliovault-pkg-source: 'deb [trusted=yes] https://apt.fury.io/triliodata-4-2/ /'

relations:

  - - trilio-horizon-plugin:dashboard-plugin
    - openstack-dashboard:dashboard-plugin
  - - trilio-dm-api:identity-service
    - keystone:identity-service
  - - trilio-dm-api:shared-db
    - percona-cluster:shared-db
  - - trilio-dm-api:amqp
    - rabbitmq-server:amqp
  - - trilio-data-mover:amqp
    - rabbitmq-server:amqp
  - - trilio-data-mover:juju-info
    - nova-compute:juju-info
  - - trilio-wlm:shared-db
    - percona-cluster:shared-db
  - - trilio-wlm:amqp
    - rabbitmq-server:amqp
  - - trilio-wlm:identity-service
    - keystone:identity-service
  - - trilio-data-mover:ceph
    - ceph-mon:client
  - - trilio-data-mover:shared-db
    - percona-cluster:shared-db

Note

The trilio-wlm and trilio-dm-api charms must be deployed with openstack-origin >= ‘cloud:bionic-train’ - even for Queens deployments. These parts of the TrilioVault deployment are Python 3 only and have dependency version requirements that are only supported from Train onwards.

Configure storage

Once the deployment completes the trilio-wlm and trilio-data-mover applications will be in a blocked state (see juju status). To rectify this, both applications must be have their workload backup storage configured.

TrilioVault supports NFS and S3 backends for storing workload backups. The storage type used by TrilioVault is determined by the backup-target-type configuration option in the trilio-data-mover and trilio-wlm charms.

Warning

Switching between S3 and NFS backups types is not supported or tested.

NFS

To configure for an NFS backend:

juju config trilio-data-mover backup-target-type=nfs
juju config trilio-wlm backup-target-type=nfs

Secondly, point both the trilio-wlm and trilio-data-mover applications to the same NFS share(s):

juju config trilio-data-mover nfs-shares=10.40.3.20:/srv/triliovault
juju config trilio-wlm nfs-shares=10.40.3.20:/srv/triliovault

Multiple NFS shares can be specified by using a comma separated list:

juju config trilio-data-mover nfs-shares="10.40.3.20:/srv/triliovault,10.40.3.30:/srv/triliovault2"
juju config trilio-wlm nfs-shares="10.40.3.20:/srv/triliovault,10.40.3.30:/srv/triliovault2"

Mount settings for the NFS shares can be passed via the nfs-options configuration option in the trilio-wlm and trilio-data-mover charms.

juju config trilio-data-mover nfs-options="nolock,soft,timeo=180,intr,lookupcache=none"
juju config trilio-wlm nfs-options="nolock,soft,timeo=180,intr,lookupcache=none"

S3

To configure for an S3 backend:

juju config trilio-data-mover backup-target-type=s3
juju config trilio-wlm backup-target-type=s3

Parameters that describe the S3 service are passed with configuration options available to both the trilio-wlm and trilio-data-mover charms:

  • tv-s3-endpoint-url the URL of the S3 storage

  • tv-s3-secret-key the secret key for accessing the S3 storage

  • tv-s3-access-key the access key for accessing the S3 storage

  • tv-s3-region-name the region for accessing the S3 storage

  • tv-s3-bucket the S3 bucket to use to storage backups in

  • tv-s3-ssl-cert the SSL CA to use when connecting to the S3 service

Options are set to the same value for both applications. For example:

juju config trilio-data-mover tv-s3-endpoint-url=http://s3.example.com/
juju config trilio-data-mover tv-s3-secret-key=superSecretKey
juju config trilio-data-mover tv-s3-access-key=secretAccessKey
juju config trilio-data-mover tv-s3-region-name=RegionOne
juju config trilio-data-mover tv-s3-bucket=backups
juju config trilio-wlm tv-s3-endpoint-url=http://s3.example.com/
juju config trilio-wlm tv-s3-secret-key=superSecretKey
juju config trilio-wlm tv-s3-access-key=secretAccessKey
juju config trilio-wlm tv-s3-region-name=RegionOne
juju config trilio-wlm tv-s3-bucket=backups

The required parameters are dependent upon the given S3 service, making the setting of some charm options unnecessary.

Authorisation

The TrilioVault service account must be granted the authorisation to access resources from across users and projects to perform backups. This will involve providing it with the cloud’s admin password (set up by the keystone application). This is done with the trilio-wlm charm’s create-cloud-admin-trust action:

juju run trilio-wlm/leader create-cloud-admin-trust password=cloudadminpassword

Licensing

The TrilioVault deployment must be licensed. This is done by uploading the license file (attaching it as a charm resource) and running the trilio-wlm charm’s create-license action:

juju attach trilio-wlm license=mycorp_tv.lic
juju run trilio-wlm/leader create-license

The trilio-wlm and trilio-data-mover applications should now be in the ‘active’ state and ready for use.