Still going. This is ridiculous. I don’t know how the rebuilding process works and this may very well be the fastest possible method, but the drobo is pushing 3 days to replace one drive. For a product that bills data safety as the most important attribute, my data has been as risk for a single drive failure for 3 days now. And I still have 3 more drives to go.
I’m not doing this again for a week and a half. If the data is going to be in single drive risk anyway, I’m going to move it to the drives I removed and then copy it back after the drobo is stable.
Yeah. That is what I was thinking could be the case. I pulled 1.1 TB from thr drobo over night so I’m down to 2.4 TB used. Once the drobo stabilizes, I’ll swap another 2 TB for a 1.5TB and then copy another 1.4 TB or so over and then let the drobo convert with less than 1 TB of data used. That should really speed things up for the last two.
This is getting goofy. Last night, it was counting down from 17 hours. While I was copying files, the progress bar was at about 50% and the time was dropping. Was down to 10 hours when I stopped copying files. I opened dashboard an hour later and the app was at 3 hours. I wanted another 30 minutes and it was down to an hour. I waited another 30 minutes and it jumped to 4 hours. I went to bed and figured it would be done in the morning.
Nope. I opened dashboard 6 hours later and the progress bar is down to 10% and the time estimate is up to 8 hours now.
Manipulating data on the unit while relayout is in-progress will definitely make it take longer, because now it has to go back and re-do the fault-tolerance data there.
Even just reads slow it down, since access is prioritized over relayout (though in my experience both get horribly slow during relayout).
BeyondRAID is more complex than traditional “stripe across” RAID, therefore so is the fault-tolerance work required.
I think Docchris is right though - there may be “iffy” sectors on some of your drives that are slowing down the process.
During the data-shuffling of relayout, previously-unused areas of the disks are likely to be accessed, and in this access Drobo may encounter bad/iffy sectors that need to be remapped, either by it or the drive itself (I’m not sure which is the case, probably depends on the nature of the error). This can tremendously slow down the process.
That said, my gen 2 Drobo with 4 x 2TB WD20EADS drives took less than 72 hours to rebuild, at half-fill. Again, it’s not an exact science because BeyondRAID only worries about the actual blocks in use, rather than doing a blanket parity on everything.
Oh yeah. I was getting about 1 GB a minute when copying files from it. I realize that copying files from it will slow the rebuild, but what is unusual is the repeated countdown to “1 hour” then the jump back up. I stopped copying files last night and afterwards the drobo rebuild progress went to 1 hour then jumped back up to 6 hours. Earlier today, I connected to the dashboard (didn’t mount the drive) about every hour and watched the rebuild time drop from four to three to two and then to one hour left. I connected about an hour later and the rebuild jumped back up to 6 hours.
I guess there could be bad blocks on the drives, but my Drobo never gave me any warning and I did have it up to 94% full or so. I copied data off it until all lights were green, then I did the first drive swap.