Yesterday I had 3×1Tb and 1×300Gb in my Drobo, and 1×1.5Tb in the post. So I pulled out the 300Gb and my data became unprotected. The Drobo is still working on that 13 hours later. I’m feeling nervous, because this unprotected period also corresponds with the most furious disc activity that my Drobo sees, so this is a likely time for a drive to fail. Eventually all my data will be on the remaining 3 drives, and will be protected. Then I can insert the new 1Tb drive.
Would it be possible (by tweaking the software) for me to tell the Drobo that I’m about to extract a drive, so it should move stuff off it while it still has it, but in a way that maintains protection?
Then, when the nominated drive is empty, I could pull it out, and insert the new drive very quickly. I’d have been protected throughout.
I realise that this means the code in the Drobo getting quite a bit smarter. I’m avoiding the alternatives which mostly involve my bank balance getting smaller.
preparation for drive removal has been a feature request for a while now
why did you remove a drive… then wait for it to rebuild onto the remaining ones before inserting your new drive.
usually you would remove a drive and immediately insert your new drive. this would mean that drobo would just rebuild the “missing” 300gb onto your new 1tb drive. and i would also suspect it would be faster (since it would only be reading from your 3 old disks and writing onto your new one - at the moment it is having to do reads from all 3 disks, then writes to those same disks, to re0-create the lost data…
and the “at risk” periods during rebuilds is why the make the s/pro/elite with dual disk redundancy - so you are still protected even during rebuilds.[hr]
personally if im swapping out a disk, i tend to do it with drobo powered down, so it powers on to find one disk missing and a new one in its place then can immediately commence a rebuild onto the new one.
it also reduces the possibility of any issues to do with knocking or dislodging a second disk during removal/installation
I’m a little confused now. I haven’t swapped the drives yet in my Drobo V2 but I wonder how it looks like. So, could you guys please clarify what happens when I remove the smallest drive and then immediately put some larger one in its place, assuming that current data layout allows the Drobo to recreate redundancy (eventually) within the remaining old drives? I mean, let’s assume there’s enough free space for the Drobo to go either from “3-disk RAID5” (yes, I know that’s a wide-ranging simplification) to “RAID1” or from “4-disk RAID5” to “3-disk RAID5”. What it would do:
[*]Ignore the almost immediately inserted new drive, start relaying the data out using the “old ones minus the removed one”, and only then, when finished, begin moving the stuff once more, now using the new drive
[*]Wait for some predefined timeout before it starts doing anything, gracefully allowing some minute or two to insert a new drive, immediately utilizing it while starting the relayout
[*]In order to protect the data, immediately start the relayout using what’s left and just after a while, when the new disk appears, immediately adjust to the new situation, taking the new disk into account and making a good use of it
I just hope that’s not the first one…
Remove the drive you are replacing.
The remaining 2 drives will start to blink amber/green.
Once that happens, insert new drive.
All 4 drives will blink amber/green.
Dashboard will give time frame of the rebuilding process.
My Drobo mostly contains data that can be reconstituted. It’s not recovering the data that concerns me, but the time and effort of doing so. I have two local copies of everything that I value, across Drobo a “singleton” USB backup drive and on another continent. There are other backups as well. I’m very risk averse.
I waited for the reconstitution to complete before inserting the new drive as I’d been suggested to do that in this forum. I’m also a software engineer, and I know that you’re likely to prompt failures by interrupting a process (reconstitution) before it completes. So…
Thanks for the suggestion of swapping the drives while the unit is powered off. An elegant solution that I will deploy, next time I’m running low on space.
If Swifty’s data totalled to less than 900GB then what you suggest would be optimum & hopefully how Drobo is designed. But what if the drives were full? (This is the situation I am currently in.) Would it really just rebuild that missing 300GB first to ensure the data is protected & optimise later? Or does Drobo try to do both concurrently?
Brand new drives have a higher than average fail rate so I want my Drobo to be wary when they first meet.
common sense would be that it tries to get back to redundant as soon as possible then it optimises layout, i cant imagine them designing it any other way.
the only way to know for sure is to put, say, 3 x 2tb and 1 x 100gb drive in it, fill it, then replace the 100gb drive. if it takes 2 hours, then its just rebuilding the missing data, if it takes 48 hours, then its doinga full optimise / rebuild
Jennifer, when replacing an existing drive with a larger one, does the position of the drive matter? i.e. does it matter if the drive being replaced/upgraded is in the bottom bay, top bay, or any other bay?
Yes, we had too many people deciding to pop up drives during rebuild to “test” the dual disk redundancy. And sometimes that can cause issues, especially if you have a 2nd disk in your pack that might be on the verge of failing and you decide to pull a different drive then the one that is about to fail.