It is strongly recommended that you take backups while an application server is not running. Make sure that your application server is stopped before taking any backups.
Backups are used for copying and archiving target data (target is either a disk or an application server as a single whole of all disks used).
- Images menu lists normal backups of an application server
- Files menu list application server's incremental backups
- Schedules menu allows you to schedule automatic backups for application server. See Schedules Settings section of this guide for details.
OnApp supports two backup types: normal and incremental:
Normal - simple method of taking backups by making full copy of target data and storing it in an archive.
Ensure that you do not use XFS or other filesystems not supported by OnApp for Linux backups as OnApp will address them as ext3/4 filesystems.
Incremental - advanced method of taking backups. During the incremental backup, only the changes made after the last backup are archived instead of backing up the whole target. You must have dedicated backup servers configured in your cloud to be able to utilize the incremental backups functionality. Incremental backups are enabled via Settings > Configuration > Backups/Templates menu.
It is not possible to take incremental backups if you are using location group functionality without a backup server added to the group - the following error message will appear:
"Backup cannot be made at this time: This disk cannot be backed up, check Location Group settings."
This issue will be fixed in next releases. As a workaround, add an empty backup server zone to your location group.
Each backup type can be taken in two ways:
Manually - the user logs into OnApp CP and clicks the “Take backup” button.
Automatically - the user enables backup schedule (daily, weekly, monthly, yearly). To enable auto-backups for application servers that support incremental backups which used auto-backups option before the upgrade, re-enable automatic backups by switching them off and on again.
If you are using incremental backups option, you should either enable dedicated backup servers in your cloud or share the backups and templates folders (paths) between your compute resources. SSH file transfer option will be skipped for application servers using incremental backups. Existing full backups will be still accessible via Backups > Images menu.
How do incremental backups work?
For example, we have a disk with three files:
File1 - 4Gb
File2 - 2Gb
File3 - 3Gb
The first incremental backup will be 9 GB (sum of all files). If you decide to take another incremental backup soon thereafter, the backup size will be equal to 0 , as the files have not been changed since the first backup (if your backup has complicated directory structure, it could be more than 0, as file system could store some system data).
If the user decides to delete File2, the target size will now be 7Gb. The subsequent incremental backup size will be 0, as new data has not been added.
If the user adds File4 of 4 GB size, the subsequent incremental backup will equal 4 GB (the size of new data added).
If the user increases File3 disk size to 6 GB, the subsequent incremental backup size will equal 6 GB, although the target is increased by 3 GB. This happens because the incremental system takes the update of the existing file as the deletion of the existing file and adding the new file with the same name (the first version of File3 has been deleted and the new one with 6GB size has been added).
Backups can be saved either to a compute resource or to a dedicated backup server. When saving a backup, the system calculates if user has enough physical/ billing plan resources to save a backup in the selected destination.
When saving a backup to a compute resource, the system does not check if compute resource has enough disk space to save a backup and only checks if user has enough billing plan limits.
When saving a backup to a dedicated backup server, the system checks both disk space and billing plan limits.
Free disk size on a target must be at least equal to the disk’s size for which the backup is taken (or to a size of all application server disk for incremental backup).
In some cases (for example, if a user has scheduled several disk backups simultaneously but there’s only free space/billing limits for the first one) the system may allow taking all the backups but will not be able to save them. This will result in a system error and over-billing.