I’m trying to deploy a duplicate Drobo 5N with identical disks and configuration to make sure I always have two complete and updated disk packs along with a hot spare of the drobo hardware just in case something goes catastrophically wrong. With over 10TB of available data even after dual disk redundancy is enabled it’s hard not to.
That said, I’m still not quite sure what the best way to do this is.
Drobo Sync is inexplicably only available on a couple enterprise offerings, so I guess that leaves me with just rsync?
All of the documentation I’ve found for it so far refers to optimizations for the DroboFS. Is there something better to use for the 5N’s? At this time they are both living on the same LAN but at some point in the near future one will be at an off-site location and they’ll have to sync over the Internet.
Ideas for most reliable and secure method to do this would be well appreciated, as right now all I can think of is rsync over SSH.[hr]
Ugh…it’s 5:30AM and I’ve got a headache.
I got as far as installing OpenSSH and rsync on both of my 5N’s, but I can’t get any further. Rsync definitely makes my head hurt, and I’m still not completely sure what commands to use to get them to talk to each other. I guess OpenSSH doesn’t include a client? Any particular ssh client that’s best for the 5N?
Okay, I guess OpenSSH did come with the client…I just didn’t set up my path in my profile. Let’s see how many more tries it takes for me to stop missing small punctuation and nuking my shell.[hr]
Alright, I decided to be stubborn and stayed up for another 45 minutes and I finally have rsync working over SSH but…
IT’S SO F*CKING SLOW!
It seem silly that it’s actually faster for me to use a computer as an intermediary instead of having the data flow directly from one device to another. I assume the bottleneck is the SSH tunnel. Is there any way to speed this up?
If you run “top” or “vmstat” on the Drobo, that will tell you something about resource usage. It’s likely that CPU is the limiting factor. rsync is very demanding on the CPU, so I won’t guess whether rsync or SSH is causing more pain here.
If CPU is your bottleneck, you could use arcfour encryption instead of stronger stuff, and disable compression. rsync -ave “ssh -c arcfour -o compression=no” copy1 backup@otherend:copy2
The first copy will take longest. You could do that first one through an intermediate computer (enabling compression with rsync -z), then use your Drobo to keep things updated from there.
I read about SyncBack on this forum while searching around for answers as well, but the “computer” that is attached is a notebook computer and because it’s my primary system it’s always with me. I have many TB of data to move and it’ll take months if I only do it while I’m at home.
Can the two Drobo’s talk to each other over SMB on the LAN at least? Could I do the initial copy that way?
hmm, as far as i know, if you already have setup a secure login to your 5n, (for example secure ftp) then i think synchback can support ssh (but i think it’s only in the paid version, but might be wrong)
one idea was to try that, and to simply “pause” and “resume” your syncback profile whenever you need to use your main laptop… (pause and resume works fine for me) - but that would need to use your internet connection.
If your main objective is essentially to Mirror your 1st 5n’s data, onto your 2nd 5n, and as quickly as posisble (from a safety point of you), then maybe the best thing is to find or borrow a friends machine, so that you simply do leave a dedicated machine located at the same location and network as your 5n’s, and to let it chug away with a basic syncback or similar.
10TB is a lot of data, but if i can do 2TB in a few days on a Gen1/Gen2 than im sure you can do it in under a week for yours. (the initial replication will take time for sure, but subsequent “syncs” will be very quick and you could even do that locally on your laptop again say once a week or something)
sorry thats all i can suggest atm, but i hope it helps a bit in the meantime[hr]
(update: if you only have your data currently on one 5n, even with dual redundancy mode, be very careful before trying to setup new stuff or trying new commands etc… if it’s your only copy of the data, i would stick with the slower method in case something gets screwed up)