Database Backup and Restore
zCICD lets you back up and restore instance databases from the interface, with no
manual pg_dump or pg_restore commands. Backups are tied to a specific branch,
which makes it easy to test something destructive, then roll straight back.
Restoring requires the Key User role or above. See User Roles.
Creating a backup
Open the branch and use the Actions area:
- Backup — dumps the database together with the filestore
- Backup w/o Filestore — dumps the database only
Leaving out the filestore is much faster and produces a far smaller file. Use it when you only care about data and not about attachments or images.
The backup runs as a background task. Watch the Tasks tab; the entry turns green when the dump is complete.
Automatic backups
Autobackup on the branch toggles scheduled backups for that instance, so each deployment has a recoverable point without anyone remembering to press a button.
Restoring a dump
- Open the branch you want to restore into.
- In the Actions area, choose Restore.
- Pick the dump from the list and confirm.
A queue job is created and the restore runs in the background. How long it takes scales with the size of the dump. The restore appears in the branch's Tasks tab and turns green when it finishes.
If the list of dumps is empty, no dump exists yet. Create one from another branch first — see above — or check the top-level Dumps menu, which shows every dump in the system.
After restoring production data
A dump taken from a production system still contains that system's mail servers
and active scheduled actions. Before working on the instance, run Turn into
DEV from the Actions area. It deactivates every ir.cron record and
repoints the mail servers at the local test mail host.
Skipping this step means the instance can send mail to real customers.
Related settings
Some behaviour is configured per repository rather than per branch, under Configuration > Repositories:
| Setting | Effect |
|---|---|
| Remove Webassets | Removes generated web assets after a restore, so they regenerate cleanly. Default: on |
| New Instance Mode | Whether new branches get a restored dump or an empty demo database |
| Dump for new instance | Which dump new instances are seeded from |
| Make Dev Dumps | Whether development dumps are produced |
Snapshots are not backups
Snapshot and Restore Snapshot (Developer role) work at the filesystem level and are much faster than a dump, but they are tied to the machine the instance runs on. Use snapshots for quick throwaway checkpoints during a working session, and dumps for anything you need to keep or move between instances.
Smaller dumps for development
Production databases are often too large to work with comfortably. A Compressor is a scheduled job that takes a large dump, shrinks and anonymises it, and distributes the result — giving developers a realistic but small and privacy-safe database. Compressors have their own top-level menu.