Backups during upgrade
During upgrade, backups are placed in the
rollback directory of the installer directory
A logical backup is an
.sql file listing all the instructions that need to be run to recreate the DB from a fresh state. They should be used when migrating DB versions as internal file structures will likely not be the same in different DB versions.
Restoring a logical backup created during upgrade
It is best practice to ensure that your backup works in a non-production environment db BEFORE you delete the current database. proceed with caution so data is not lost.
sudo docker kill openelisglobal-webapp external-fhir-api
- replace file name
sudo docker cp ./rollback/<backup-name.sql> openelisglobal-database:backup.sql
sudo docker exec openelisglobal-database psql -U admin -d postgres -f backup.sql
sudo docker-compose up -d
A physical backup is a copy of all the database files in the structure that they are found on disk. It is not recommended to use them when migrating database versions as the file structure is likely to change between versions.
Restoring a physical backup created during upgrade
previous state is preserved in
/var/lib/openelis-global/data-old/ in case restoring does not go as planned. Move this directory back in place if you'd like to continue using it.
sudo docker kill openelisglobal-database
sudo mv /var/lib/openelis-global/data/ /var/lib/openelis-global/data-old/
- replace directory name
sudo cp -r ./rollback/<backup-directory-name> /var/lib/openelis-global/data/
sudo docker start openelisglobal-database
Currently backups are being handled by a script that takes a logical backup periodically and copies them offsite. This script must be configured to have a destination. This script will be getting replaced in 2.7 to use physical backups with continuous WAL archiving that are similarly copied offsite (for reading on logical backups vs physical backups here is a good intro. This will still require configuration to choose an offsite location and files will not be automatically deleted from the local server, so it is important to have someone go in to the local server, identify which files are no longer necessary to keep and remove them from the server. It is recommended for organizations to look at the backup mechanism/files and test the recovery process regularly. If the supplied system isn't considered robust enough for an organization (hot standby required, point in time recovery required, automatic failover, 0 data loss, etc.) it is recommended to configure a separate backup system to ensure it meets organization requirements. In particular, barman is recommended for managing backups.