I have read up on setting up kopia server and have one running on a server on my local lan. However, I am curious to know if there is any reason I should setup a kopia server on a server on my local lan vs just keeping the repo on local disk and then using rclone to keep a copy of the repo on that same server and a copy in B2. Any opinions on this would be appreciated.
You don’t need rclone
at all. You might want to use option sync-to
directly to create/update spare backup repositories, locally (a spare local repository on some external hard drive) as well directly to B2. Just run after all snapshots completed sync-to
to update spare repositories.
Same question, except, my local Repo will be network storage (not direct attached HDD).
Only question is, do I setup a Kopia Server, or just a NAS?
I’m thinking that running RClone from the NAS to cloud storage adds a layer of security: it would be very unlikely that ransomware on the backup client could compromise the NAS which is the only device having the cloud storage API key. Of course RClone is still going to clone. Real solution is retention periods enforced by cloud. Then, sure a single backup client could do sync-to, but letting either NAS+RClone or Kopia Server do it relieves the client of the extra work. The NAS is already setup, already a target for an other backup product that offers full system bare metal recovery, but other product deduplication is poor, so plan is to use it for system volumes and use Kopia for data.
Still: I think I’m hearing that if we want to share the Repo between two or more backup clients, then maybe we need to use Kopia Server. Is this true?
I find kopia vastly faster than rclone, which is not a backup. Kopia server does help with visiblity into multiple backup clients backups. But you can do backup without Kopia server, as I used to, even with multiple clients. Depends what transport you use. However, when using Kopia, I found the maintenance which then runs on the client to be too slow. When using Kopia server, the maintenance runs on the server and it doesn’t have to send stuff across the lan to do so.
I think I was less than clear describing our environment.
We have Windows Server that hosts data.
We have a NAS that is strictly used as backup target.
Sometimes that’s it: single backup client, NAS as target. Occasionally an additional backup client.
Currently using a commercial backup product to push full server snapshots to the NAS.
Offsite copy: the NAS OS then uses RClone to replicate backups to cloud storage.
While the commercial product offers bare metal recovery, it lacks deduplication. Kopia seems a far better tool to backup volumes that are purely user data.
Obvious path to use Kopia would be to place Kopia repo on the NAS and also RClone the Repo to cloud.
Documentation for running Kopia Server Mode seems thin at best.
We’re already offloading offsite replication to the NAS so client doesn’t have to do it. Was asking: is there any compelling case for us to create a new Kopia Server host?
Rcloning the complete repository to the cloud is the easiest solution. Another solution is using kopia repository sync-to
.
I would definitely recommend enabling object lock or snapshots, if your cloud provider supports that.