Hello I new to this forum and I’m new to Rockstor! I’m trying to build a new server / nas (hp gen8 microserver). My current idea is to use:
Hypervisor: ESXI (on SD-card)
Guest (Windows Server on SSD)
Guest (Rockstor on SSD)
2 HDD passthrough to rockstor guest.
My problem right now is that I need to figure out how to setup the HDDs, one 8TB and one 3TB. What’s the most practical, secure way to setup these if I don’t need more than, say 100GB of replication? (The rest will be non sensitive data like media stuff).
I’ve been thinking perhaps the simplest way is to format each drive to “single drive” btrfs, and use a dedicated folder/share on one of them for nightly backups for the other drives 100GB. Are snapshots something that could work for this task? Or is best to copy data using a simple backup script that would run every night?
I see that nobody else has weighed in after a few days, so I’ll put in my two cents. I think you’d be better off adding additional disks to create separate pools with varying redundancy. However, as you said, with only 100GB that needs to be protected, this is an expensive and wasteful option. In this case, then, I recommend you setup a script to do the dirty work. Seems much easier than the alternatives.
Thanks for your reply. I could add another disk, but I assume I need 8TB or 3TB to do a mirror as a redundancy, of type mirror, at least. If I only want to protect the 100GB, do what is the best approach? I have read little about snapshot send/receive. Is this a good way to protect data? To my understanding It will only be protected when it is snapshot.
You should first decide on what your requirements are.
If the data is not critical and downtime of the data is not a problem, you can just setup the drives independent from each other in whatever configuration you like, and sync the 100GB of data between them.
If you need to backup (i.e. you can’t lose it) the 100GB of data then I would recommend a simple external backup (external HDD, USB drive, NAS etc.) since Rockstor isn’t quite reliable enough to have it setup as the only copy of data in my opinion. Maybe you could add a small drive to the hypervisor (ESXi) and use it in a simple VM to rsync the data, but then again: all disks are mounted in the same hypervisor/host so it’s still no real backup. If you put the 100 GB backup somewhere else, you can use your remaining 10.9 TB of storage for the less important stuff. Of course this can also be combined with the previous bullet, then you would have three copies.
If the data must be available at all times, some form of raid or failover would be the obvious choice. Then you would have to get another 3TB or 8TB drive to complement the one your already have, or you would probably lose 5TB of usable storage.
By the way, on a sidenote: do you want to use Rockstor also for NFS and/or iSCSI storage ‘back to’ the hypervisor? In most cases this would be quite inefficient since you are running the VM on the storage that is provided by another VM on the same host. If multiple hosts (one for hypervisor, one for storage/nas) are not option, I would seriously consider running both on the same host, i.e. expanding the NAS with VM-capability (e.g. KVM/Xen on Rockstor) or expanding the hypervisor with NAS-capability (e.g. Proxmox/oVirt with LVM, RAID, SAMBA, NFS etc.). But this is of course more complex.
Thanks for your excellent reply. I have also looked at Proxmox, because of having storage directly at host (hardest). Another way was to just install Windows server 2016 with hypervisor and use drivepool + external backup at host level (easiest).
I’m leaning to bullet 1 and 2 because rockstor seems like a nice way to have storage separated from windows where I will do work. So this means I should sync the data between internal drives. I will also have a external backup (2.5" usb-hdd) so if rockstor dies and internal disks explode I have some backup. What is the best way to sync between the internal drives? Is a snapshot send/receive script ok?
I do not need to have the important data available directly if a drive fails. I’m not aiming to share data between the vm guests or to host, only between rockstor and my few other windows clients; tablet, laptop, workstation within the lan and trough vpn from internet.
To be honest, I don’t know what the best way of syncing data between internal drives on Rockstor is. I use Rockstor intensively for three days now and do not need internal syncing since I have other nodes for that (replication and rsync). Maybe someone that uses Rockstor for a long time now can better answer this question .
On first sight it seems that Rockstor doesn’t provide a handy mechanic to mirror data to other hard drives, you could always just use rsync (Rockstor seems to be CentOS with some configured software/packages, scripts and nice GUI on it). Rsync has proven itself for years and years now, and combined with a simple cronjob can make reliable copies of data on different hard drives .
Yes, I’ve used some form of rsync on windows for backups to a server. I think it was Robocopy. I will have to do some research on which way is most suited for my needs. I’m guessing that internal syncing might work with snapshot send/receive to a another pool might work. Otherwise I will fallback to rsync and cronjobs as you pointed out is a good proven idea.
.
.