More about backup & restore

We don’t use S3/GCS to store our backup but it could be done.

For now our backup proc is :

A kubernetes job that is doing the following steps:

Start a new fresh & empty mesh instance, that will rejoin mesh cluster and will download /graphdb database.
Stop the mesh instance,
Backup /graphdb folder with graphdb backup tools, produce a tar/zip package
Store it to a shared NFS (could be stored to a GCS/S3 bucket)
Package and store a tar/zip of the shared /uploads folder too.
End.

Our restore proc is :

(Cold restore)
Stop all the node from the cluster you want to restore.

Start a pod,
In init container download graphdb zip and uploads zip from NFS (could be dl from GCS/S3 bucket)
In main container, start mesh server, with the restored data.

Then, start the other pods (replicas), they ll rejoin the cluster with the new data