Since I replaced one of my four 1.5tb drives with a 2tb drive and it has been rebuilding for the past eight days (it was supposed to take 72hrs, but after 72hrs, it rebooted itself and has continued to be flashing orange/green every since and doesn’t appear in the dashboard at all), I’m wondering if it is safe to pull the power cord (since there’s no way to do a safe shutdown) and remove the 2tb drive that was just added?
I presume it should be okay, but I’m not sure what the rebuild after expanding space entails. Does the data remain static on the other three drives and they just rebuild the fourth drive using a combination of that data?
Or does it rebuild/restructure data across ALL of the drives during a rebuild – meaning that if I removed this new drive, a future rebuild would be impossible?
I only ask, because since it is going on the ninth day, I’d like to try adding in a new-new drive. Especially since I’m still waiting on a reply from Drobo support.
Do not remove any drives while it is blinking green/orange.
It does not build data on the blank drive just based on the added drive, since the fault-tolerance data is distributed.
If you remove the new drive or any drive, you will (worst case) lose all your data, or (best case) get the all-red-blinking and Drobo Dashboard will tell you too many drives have been removed - replace the missing drive. Either way, don’t do it.
It is troubling that it’s been running this long, but until you hear back from support (best to phone them, at this point) I wouldn’t do anything more without their instruction.
If it’s actually restructuring data on all four drives, that seems incredibly inefficient. When you swap a drive in RAID, it uses the inherent redundant data on the remaining drives to rebuild the replaced drive. I can’t imagine what data would need to be physically altered on the remaining drives since they should just be serving as sources for the data to be rebuilt on the added drive.
Then again, RAID doesn’t allow this time of expansion, either, so perhaps there is a re-alignment of actual data involved in the process in order to incorporate the new space that was added.
Make sure, if you can, that the drives are actually spinning. My Drobo went into a state where it was in relayout, but Dashboard just gave a barberpole for progress and never resolved to a time. I was suspicious because I hadn’t replaced any drives! This relayout, if due to a bad spot, should have resolved after about four or five minutes. After several days I declared failure and, at Drobo Inc’s advice, unplugged and reset the Drobo. When it came back it did a five-minute relayout and that was that.
Just a thought on this from a future standpoint - and maybe DRI can confirm…
but there must be a way to make a failsafe feature so that if you pull a drive out which was being relaid out, (or powercuts etc) that it should be able to recover when you put it back in (and not perform a simply wipe etc)
eg, old defrag and drive utilities used to go through a “read, write, verify, delete” set of steps for each piece of data they manipulated, so that if something went wrong, nothing would actually be deleted until the previous steps were completed properly etc. - i think part of this is built in for powercuts etc but im think there must be a way to increase the “intelligence” of a drobo unit, eg by way of firmware, to make it recognise if you eject and reinsert a drive from the same disk pack that it wouldnt just blank it and redo layouts - Must be a way. (even by writing a small set of data like last inserted, last written to, disck pack volume info etc)
(sorry cronjob, this rambling doesnt help you much - just thought of it and wanted to post for dri as related)
like bhiga said, dont be tempted to risk anything.
i dont know what would happen in the event of a powercut so it one of your friends comes round with a new electirc guitar and power amp and asks to set it up, say “no”
just wondering… even with the dashboard not finding the unit… can windows actually still see the drive and data? (am assuming so) in which case if theres a critical set of text files/folder you probably wont lose anything by trying to copy it to another drive letter etc to play safe.
do you have any details about what drive you replaced with which one? eg models makes etc.
for example, it might be possible to work out the “inefficiency” of the relayout, based on 4ksector sizes or drive makes, speeds etc to find out how long your data amount might have needed to do a full relayout, and then you could see how much more time you might need to wait etc.
(also, i seem to remember someone saying “not” to run any diagnostic attempts or chkdsks while blinkings etc are on.)
If you don’t have the drive out long enough for Drobo to “notice” then no harm, no foul, no change.
For quickest relayout, don’t have Drobo connected to the computer at all while it’s doing relayout. Indexing, virus scans, any type of access will slow the relayout process. Of course it’s also dependent on the amount of data you have stored on Drobo too.
I checked for this about a week ago and could not tell with all certainty whether they were spinning. Primarily because they share the same surface as two desktop machines and another Drobo - so there’s already a little vibration, there. The drives were very warm, however - about the temperature you’d expect if they had been actively running for awhile - so I suspect they were.
The dashboard appeared exactly as you describe it, though. In fact, I used the same language to describe the progress bar in another post. I had completely reset the device based on the document at DRI, to no avail.
I still haven’t heard back from their support group, but that appears to be my fault. I filed a ticket on Friday and when I checked Tuesday, it didn’t show up in my account. No idea way. So I filed another one, yesterday. Only today did I realize that it responded immediately with an auto-generated list of documents I should (and already had, of course) review. So I had to attach a diagnostics and resubmit my question again. Hopefully I’ll hear back in another day or so.[hr]
[quote=“Paul, post:6, topic:2219”]
Just a thought on this from a future standpoint - and maybe DRI can confirm…
but there must be a way to make a failsafe feature so that if you pull a drive out which was being relaid out, (or powercuts etc) that it should be able to recover when you put it back in (and not perform a simply wipe etc)[/quote]
It would certainly seem to be more efficient that way, but I would even be okay if I simply knew that in such an event the data was certain to be recovered. For example, what if the newly inserted drive went bad during the process or something else happened? You should still be able to recover from the other three existing drives, in my mind. There shouldn’t be any point where adding/replacing the additional drive would cause complete failure in the event that the new hardware or the whole process itself failed. (Maybe this is the case, but I have yet to confirm such).
just wondering… even with the dashboard not finding the unit… can windows actually still see the drive and data? (am assuming so) in which case if theres a critical set of text files/folder you probably wont lose anything by trying to copy it to another drive letter etc to play safe.[/quote]
This is connected to OSX, in which the drive isn’t actually mounted . . . even though the dashboard says that “data is not protected” while it’s undergoing the process and that “you can still access your data”. Of course, can’t access your data if the drive isn’t mounted.
This drobo is just four 1.5tb WD15EARS which were going to be slowly replaced with four 2tb WD20EARS, like the other Drobo is using. Removed the old 1.5tb that the dashboard instructed and replaced it with the 2tb. Waited 72hrs for the process to complete; then it rebooted itself and went into this “in progress” mode with no ETA and no mounting for the last five days.
I have had miserable experiences at least 60% of the time with the new 4k sector 2tb WD drives - even on systems that are supposed to support it out of the box (Windows 7, OSX 10.6.5, etc). However, these are the drives that DRI advises (and sells) through their own site, so they should be fine.
I’m almost starting to wonder if there is some sort of damage to the drive that is causing all of this, but since there is no user-oriented diagnostics available for the Drobo, I have no idea. It would be fantastic if it could offer some feedback during the process, like “hey, this drive seems to be failing, so I’m going to stop activities until you replace it”.
The reason I am starting to suspect this is that the other WD20EARS that came in the pack doesn’t seem to mount on anything. While I’ve usually had problems with WD20EARS before, they almost always can at least mount. On the other hand, I’ve seen enough problems with them that I can’t be certain if a failure to mount them is due to the drive or just incompatibilities with the OS.
On OSX, attempting to erase/format the drive via Disk Utility GUI or the diskutil CLI produces variations on the error “Disk Erase failed with the error: POSIX reports: The operation couldn’t be completed. Cannot allocate memory”.
All reports that I’ve seen regarding this assign the blame to issues with the USB docks, but it occurs for me both via USB dock (which is known to work successfully with 2tb drives) as well as internally, via SATA.
Then, on Windows 7, the BIOS sees the drive but the OS doesn’t. Not even via Computer Management->Storage->Disk Management. Both via USB and SATA.
I haven’t tested on my Solaris box, because it’s SCSI.[hr]
In my specific case, it was nearly 4tb of data, but it’s on a headless OSX desktop with no activity occurring. The process took 72hrs, but like I say, it rebooted itself at the end and went into . . . whatever mode it’s in now for the past five or six days.
If a drive goes bad during the process of a relayout/expansion, does the Drobo provide any feedback? I mean, if it turns out the drive went bad after being inserted, will it just continue to attempt to layout indefinitely or would it eventually throw up an error on the dashboard that instructs me of issues with the drive so I can deal with it? (I ask, because a drive going bad during the 72hr process is the only explanation I can think of outside of the Drobo itself having failed).