I would like to use restic to backup to a tiered storage system that has a very aggressive policy to move data to tape from disk. I have the ability to control what is archived to tape and what is left on disk. I would like to be able to store the backup data on tape and free it from disk and leave the files needed to generate the next backup on disk. Does anyone know what is needed in the first backup to derive the changes to the next backup? I experimented with a 200GB dataset and I noticed that the second and third backups went very quickly when only a few changes where made. This led me to believe that all of the 16MB chunks stored in the data directory where not being parsed and could possibly be archived off to tape to save disk space and to generate a durable long lasting backup. Thanks!
hello @jpmitchell, from your story i conclude you have cold storage for the repository?
I do not have personal experience with that, but if you search this forum and the internet with restic rustic cold storage keywords then you will find more information. i.e.
restic faq - are-cold-storages-supported?
Thank you for pointing me to those resources. I will check them out.
@jpmitchell About the question: You should keep all metadata and can move all pure data pack files to the tape. The problem is that under /data there are tree pack files and data pack files and you cannot distinguish them without reading the index.
Moreover, besides backup there are other operations you may want to consider: Do you want to run regular checks to test (metadata) integrity of your repository? Do you want to run forget/prune to clean up old data or do you want to have some kind of append-only backup? What about restore?
restic assumes to have read-access to all files, so the above commands most likely won’t work in your setup.
I’ll write you a PM about possibilities with rustic.