February 17, 201313 yr Out of the blue, I get a message in /tower/main that one of the disks is not installed and there's a magenta ball next to it which stands for "disk spun down" afaik. Under unmenu, the status of the disk shows as DISK_DSBL_NP. syslog attached. What now? syslog.zip
February 17, 201313 yr Check cabling of the drive... Did you change something recently ? Does your PSU have enough power ?
February 17, 201313 yr Author I had recently added the 4th drive and the array was working fine for the last 3 weeks now. Checked the cables, then made the post. Even interchanged a SATA cable that looked a bit bent from the missing drive to a working one, but its still the same. Recent changes: Up until yesterday I was trying to add more drives using a pci sata controller, but the drives never showed up in unRAID or the bios, so I've removed all that.
February 17, 201313 yr Can you get a SMART report on the drive ? Probably not though.. Try and exchange SATA and power connector with a drive that does work and reboot... if the problem travels with the drive its the drive, if the problems stays with the connection its the connection.. The fact that you have been working on the inside of the system makes a cabling issue pretty likely so best to double check.. If all else fails then remove drive and attach it to your pc or another system to see if it works there.. If the drive remains unreachable its obviously broken and you need to replace it..
February 17, 201313 yr Author I tried the SMART report command but that did not detect any such drive (sde in my case). Have swapped the sata once. Will swap power cable too and post back. In what machine should I connect the drive? Will a Linux machine (openelec) detect a resierfs filesystem disk?
February 17, 201313 yr In what machine should I connect the drive? Will a Linux machine (openelec) detect a resierfs filesystem disk? You do not mention what brand of disk it is? I would recommend that you download the diagnostic tools that all manufacturers seem to supply and use that to check out the disk. Such tools almost invariably run under Windows so that may dictate the machine you use. Diagnostic tools do not care about the file system format - they work at the physical level on the disk.
February 17, 201313 yr Author UPDATE: Removed the whole shebang from the cabinet and threw it back in. Now the drive shows in the disk drop box, though there's still a red dot with "Not installed" under it. BUT strangely it shows as "sdd" whereas it used to be sde (because it was the last drive assigned to the build). sde is now assigned to my flash. How do I change that? Or should I just assign as sdd and continue?
February 17, 201313 yr The dev/sd? names are not under your control. They are assigned dynamically by Linux as it recognizes the disks. It can vary between runs if any disks are slow to be recognized, although in practise this tends to be consistent so that you get consistent /dev/sd? device names. With the unRAID 5.x series the device names are not relevant - unRAID can now recognise a disk by its serial number regardless of where it is plugged into the system..
February 17, 201313 yr Author By that last post, do you mean I can assign the disk to the empty slot, and start the array? Edit: Uh oh. If I select the disk in sdd, without yet starting the array it shows a blue ball and the array status says : "Disabled Disk Replaced. Start will bring the array on-line, start Data-Rebuild, and then expand the file system (if possible)". Although it means that the data will be safe, that wouldnt exactly be the correct way to add a pre-existing drive to the array, I'm sure.
February 17, 201313 yr Out of the blue, I get a message in /tower/main that one of the disks is not installed and there's a magenta ball next to it which stands for "disk spun down" afaik. Under unmenu, the status of the disk shows as DISK_DSBL_NP. syslog attached. What now? That stands for Disk-Disabled-It-Is-NOT-PRESENT (DISK_DSBL_NP). Power down, re-seat the cables to it... and try not to disturb the cabling to the others. If you had mangled the cable to it previously, you might have damaged the cable, try a replacement cable.
February 17, 201313 yr Author That stands for Disk-Disabled-It-Is-NOT-PRESENT (DISK_DSBL_NP). Power down, re-seat the cables to it... and try not to disturb the cabling to the others. If you had mangled the cable to it previously, you might have damaged the cable, try a replacement cable. Did all that. Did you miss my post with the update right above yours?
February 17, 201313 yr By that last post, do you mean I can assign the disk to the empty slot, and start the array? No, but you can plug it into any disk controller port. It must stay assigned to the same logical slot in the array. Edit: Uh oh. If I select the disk in sdd, without yet starting the array it shows a blue ball and the array status says : "Disabled Disk Replaced. Start will bring the array on-line, start Data-Rebuild, and then expand the file system (if possible)". Although it means that the data will be safe, that wouldnt exactly be the correct way to add a pre-existing drive to the array, I'm sure. It is exactly the way if the array had been started with it missing.
February 17, 201313 yr Author I'm sorry but I'm not getting you Joe. Can I just continue with adding the disk to the slot (sdd) which was the slot for flash earlier, and start the array? I don't have any disk controllers. All are MoBo sata ports. Except that now flash is taking up sde which was earlier occupied by the "Not Installed" disk.
February 17, 201313 yr I'm sorry but I'm not getting you Joe. Can I just continue with adding the disk to the slot (sdd) which was the slot for flash earlier, and start the array? I don't have any disk controllers. All are MoBo sata ports. Except that now flash is taking up sde which was earlier occupied by the "Not Installed" disk. Your version of unRAID does not use the device names. When assigning drives to logical slots it uses the model/serial numbers of the disks. There is no "sdd" slot. Never was. There is a logical slot (disk1, disk2, disk3, parity) and a disk make/model number. On any given boot Linux assigns a /dev/sdX device number or a /dev/hdX device number. That three letter device name can change from one boot to the next. You need it only when preclearing the drive or when running "smart" reports on it. Again, since it can change from one boot to the next, it is important to look in the syslog, or on the web-management page at any given time to see its current assigned value. You most certainly DO have disk controller ports. It is just the ones you are using are built into the motherboard and not on a PCI or PCIe card. You can plug any disk into any port, and as long as the port is working unRAID will match up the model/serial number to the one you assigned to a logical slot in the array. It sounds as if you've had a disk failure, but the only way to know for sure is to look in the syslog. Even then, is you have a loose or defective SATA or POWER cable, or a defective or loose power splitter, you will not see a give drive available if the connection is not being made. Is the disk you see as "not present" listed in your syslog? (What is the make/model/serial number of the missing drive?) Joe L.
February 18, 201313 yr Author I ran a SMART test on the disk after it was rediscovered (post the cabling re-checking), and the report showed no errors. However I couldnt figure out why it would not be recognized by the array. Since "Data will be reconstructed" message was displayed, I went and added the disk to the array and the rebuilding process has started. Once it is complete, I will post back with the result. I was wondering if I should have used the option of "Trust my array" instead, to save the 6 hours it is taking to rebuild the drive. But I'm not sure it applies to this scenario so I rather not take that chance.
February 18, 201313 yr I ran a SMART test on the disk after it was rediscovered (post the cabling re-checking), and the report showed no errors. However I couldnt figure out why it would not be recognized by the array. Since "Data will be reconstructed" message was displayed, I went and added the disk to the array and the rebuilding process has started. Once it is complete, I will post back with the result. I was wondering if I should have used the option of "Trust my array" instead, to save the 6 hours it is taking to rebuild the drive. But I'm not sure it applies to this scenario so I rather not take that chance. For the drive to show a "red" indicator, a "write" to it had to have failed. Therefore, it is guaranteed to be out-of-date. If you "trust" it, you will be almost certain for it to be "out of date" Now, the difference might just be the "mount" date/time, or as significant as a critical file... in any case, it is out-of-sync with parity. Therefore, if you "trust" it, and do not follow up with a parity check, then parity may not be in sync with the un-written disk. My advice, never "trust" parity. At the very least, perform a non-correcting parity check, and then make an informed decision as to how to proceed.
February 18, 201313 yr Author Thanks, I understood the bit about the red ball. Good thing I didnt get hasty and try the Trust my array procedure. What I'm having difficulty understanding is - I suppose the disk is now being treated as some sort of fresh disk and is being rebuilt using parity data, right? Yet I still need to recheck parity once the rebuild process is done? I have no doubt you must be right in what you are saying, but I'm having trouble understanding the logic.
February 18, 201313 yr Thanks, I understood the bit about the red ball. Good thing I didnt get hasty and try the Trust my array procedure. What I'm having difficulty understanding is - I suppose the disk is now being treated as some sort of fresh disk and is being rebuilt using parity data, right? Yet I still need to recheck parity once the rebuild process is done? I have no doubt you must be right in what you are saying, but I'm having trouble understanding the logic. The reason is very simple. When you re-construct a disk-contents onto a replacement disk you are writing only. There is no reading of the new disk taking place. (you are reading all the other disks to generate the reconstructed contents) The only way to know for sure if what was written is actually readable is to perform a parity check immediately after the reconstruction. (It probably should be a non-correcting check) If there is a sector on the new disk that is un-readable, unRAID will re-construct the contents once more of that block and re-write it once more. If that occurs, and you see a new re-allocated sector on a smart report, you could perform another non-correcting parity check.... The goal is to have the non-correcting check find no errors, and to not see additional sectors pending re-allocation, or additional sectors being re-allocated. Joe L.
February 21, 201313 yr Author By following all the steps suggested in this thread, my data is back to its original state and the array is running error free! Thanks all!
Archived
This topic is now archived and is closed to further replies.