Re 1) Yes, in principle, you could turn it off while it is relaying, and then restart it later. However, that would not be a good idea for two reasons… first, because it is in the middle of a relay, you MUST let it complete this relay and get back to a green state before you replace any drives, even the one you last installed. Second, turning it off is just tempting fate- testing DRI’s backup battery circuitry. It is supposed to recover from a loss of power during a relay (that’s more or less how you have to stop it). But I think you’ve been through enough that you should not be testing their backup battery at this point in time :-).
- In principle, you can copy the data off while it is relaying. However, personally I would not, for a couple reasons. First, the throughput performance will be terrible. It might be something like 4-5MB/s. If you have a few terabytes of data it will take a week to get the data off. Second, the relay, by itself, is very stressful on the drives. It seems to me that forcing the Drobo to do two very stressful things at once (a relay and a bulk copy) is just going to make it thrash harder. And third, it will probably take the same amount of time to get the data off either way because the drives are already working full bore.
Now, if you have a relatively small amount of very critical data that you want to pull off while it relays, that’s fine (in my opinion). The above is just my personal opinion regarding a full bulk copy during the relay. I would just use common sense to decide how much, if any, to pull off during the relay, considering what I just said, and what I am going to say below.
How much data is on your Drobo? When I last upgraded a couple of drives I had 3TB of data on the Drobo and each relay required about 63 hours. My number is consistent with other numbers I’ve seen.
It is likely that the majority of Drobo users do not back up their Drobo arrays. Your case is a good example of why even a Drobo is not fail safe. And personally, I think these Drobos (or at least my Drobo V2) is probably as close as you can get, considering all the possible things that can go wrong with any Raid-like technology. My V2 has been running 24/7 for over 2 years. I have never lost a drive but I have done at least 4 upgrades without a hitch. My Drobo is not very fast (I have performance issues with it) but it has been reliable. Your mileage varied.
The biggest problem with the Drobo, in my opinion, is the 65 hours of relay time (for me, YMMV) during which the Drobo is working the drives at 100%. If a remaining drive is marginal, this is when it’s going to fail. And this is why these can never be fail-safe devices. We live in a world of sub $100 drives that are built to a price. Just reading hard drive reviews on NewEgg or Amazon, you’d think it would be a miracle any of them work at all.
This may not be as true with a 5 bay unit run in dual redundancy but there are other things that can go wrong.
It was not clear to me if you tried to hot swap in the first 2TB drive that you first cold booted. It sounds like you tried it too, in a hot swap, and it failed? And is now unusable in your Mac?
One of two things happened here:
You were unlucky and actually had two drives fail quickly.
The Drobo has a hardware defect and killed the drives.
If #2, then you would expect the Drobo to kill the 1.5TB drive you are now adding to the array. So logic dictates that if this 3rd relay attempt succeeds then you just had two drives die at an inopportune time.
Once the Drobo recognizes the old Drobo drive and starts relaying the data, that old Drobo drive is no different than any other new or used drive (except for any wear and tear from prior use). Just in case you were thinking along those lines.
If you do not do so already, I would put an UPS between the Drobos and AC. I think power fluctuations are behind a lot of otherwise unexplainable problems with computers, in general. Evan an inexpensive $50 APC UPS will go a long way to providing clean power. The Drobo doesn’t pull much power although the associated Mac may so you should size it accordingly. I suggest an UPS because you say you have had many issues with reliability (which most of us don’t seem to have).
You can also see why the time to backup your data is not when you have a problem. Especially considering the amount of time it takes to relay and/or backup terabytes of data. Backup is something that needs to be kept up to date, daily if the data changes daily. Then, when you have a problem, a quick sync is all that’s needed to get it fully up to date. Even if you dump the Drobos, your next disk array will not be fail-safe either.
You should try to test the two drives with the drive maker’s test app. I use WD drives; they seem to have a decent drive diagnostic utility. They should also still be under warranty if the drive has a 3 year.