Backup and restore

Critical systems are often backed up to ensure continuity in the event of an outage or other critical failure. Operators typically use backups for the following purposes:
  • to replicate a customer issue when debugging
  • to restore a cluster to a previous state (often via re-installation)
EDA's implementation for backup and restore include the following features:
  • You can perform backups at any time with no outages or maintenance actions required.
  • Backups are atomic and contain the lost known set of working configuration. The system waits until an in-progress transaction is completed before proceeding with a backup.
  • You can restore into a dirty cluster and revert the cluster back to the state provided in the backup, auditing any resources as necessary.
  • You can restore into a clean/freshly installed cluster.

Backups

In EDA, a backup is simply the copy of all git repositories in use at the time of the backup and an EngineConfig resource file from the source that can be optionally restored.

The edactl platform backup command is used to create a backup. At a high-level, the this command does the following:

  • Creates a tarball of all repositories, including the following:
    • Backup
    • Apps
    • User storage
    • Certificates
  • Adds to this tarball the current EngineConfig resource file.
  • Streams these files to the client performing the backup over gRPC.

The tarball created is created in the format eda-backup-<cluster-member-name>-<date-and-time>.tgz in the current directory. You can optionally provide a name for the tarball and provide an alternate destination by providing the name and the destination's path in the command.

Restore process

The edactl platform restore command restores a backup. When you initiate a restore operation, the ConfigEngine on the destination cluster performs the following tasks:
  • Receives the restore request via gRPC. The request contains the complete tarball generated via a backup.
  • Unpacks the backup, overwriting all content as it goes.
  • Pushes backed up repositories and files to any server identified in EngineConfig resource.
  • Restarts, relying on Kubernetes to restart.
  • Starts again as if it had started clean from the repositories.
Note: A restore operation restarts the ConfigEngine, so use the command with caution.

Creating backups

You can create a backup using the edactl platform backup command.

​
  1. From the eda-toolbox pod​ command line prompt, execute the edactl platform backup command.
  2. ​ Copy the backup from the eda-toolbox pod to a safe location.

Restoring backups

  • The destination cluster must be running the same version as the cluster from where the backup was created.
  • You must have rights and permissions on the cluster in which the backup is to be restored.
  1. Copy the saved backup to the eda-toolbox pod.
  2. From the eda-toolbox pod, execute the edactl platform restore <eda-backup.tar.gz> command.