6 Replies Latest reply on Nov 11, 2011 12:12 PM by Slowmo

    Lost data on RAID5 after adding more drives

    Slowmo

      I had to reinstall the operating system and at the same time decided to add more drives to existing 3 drive RAID 5 array. Because I have MSI motherboard, I went to the manufacturer's homepage and downloaded all the latest drivers for my particular motherboard. That included Rapid Storage Technology (v 9.5.0.1037). Verifying in help files that I will not lose data when expanding the array, I connected 3 more drives to existing ones (all of the drives are 1Tb) and added them to array using RST GUI application. Array started to expand.

      Then after couple of hours I had to restart the computer and for some reason Chkdsk war automatically run during startup. It started to fix some drive errors and took a while to finish. When I got back to Windows, most of the files were corrupt. All of the folder structure was still there but files were pointing to wrong parts of the disk (text files opened as binary files e.t.c.). I blamed Chkdsk for that, and decided to wait until rebuild is complete to see what I can recover. At that time I saw that many files where still usable (mostly those which were deeper in folder structure). Whole rebuild took a whole week and after the restart, to my disappointment, drive does not even have a valid file system anymore (Windows reports it as RAW). Disk manager shows 2 partitions (one partition in size which is equal to array size before I added extra disks, and other part is unallocated).

      Even data recovery software can’t help much. I tried to look for known file headers in raw data and although I had hundreds of such files, not one can be found. Recovery software reports that it finds 72 boot sectors. I’m not an expert in file systems, but I think NTFS should not have 72 boot sectors on one drive. It’s not like the whole array is randomized, because thousands of MFT entries could also be found and the only thing recoverable appears to be file names. There’s no way Chkdsk could do such damage to data so I think RST is to be blamed.

      Has anybody heard of something like this? I am still searching the drives for some valid data blocks, but so far haven’t had any luck.

       

      P.S.

      My motherboard has ICH10R chipset and I'm running Windows Server 2008 R2 (which is 64bit)

        • 1. Re: Lost data on RAID5 after adding more drives
          Arny006

          Hi,

           

          At first, I would suggest to install the Version 9.6!

          For second, are you sure that your operating system support HDD capacity over 2GB?

          At last, do you think that a RAID 10 is better then 5?

           

          Enjoy and BR

          • 2. Re: Lost data on RAID5 after adding more drives
            calamar

            Slowmo escribió:

            I connected 3 more drives to existing ones (all of the drives are 1Tb) and added them to array using RST GUI application. Array started to expand.

            Then after couple of hours I had to restart the computer and for some reason Chkdsk..

             

            Did you let the system finish the RAID5 build? "a couple of hours" sounds very short time for a 3TB RAID5 to rebuild. I think you killed your own RAID5 build before it finished, and it looks like your datta is scattered over the disks. RAID5 data distribution and parity distribution has to do with the controller, not the file system. Everything points to a definitive data loss.

            • 3. Re: Lost data on RAID5 after adding more drives
              Slowmo

              The rebuild process was not complete after couple of hours. I had to restart because I changed some settings which required a restart. I think that was a computer name which I changed. And at that point file system was already damaged because chkdsk was run and fixed a lot of errors. As I mentioned before, after the restart many of the files were still usable. I thought I will wait for rebuild to finish (which took exactly one week - 5Tb after all) and then deal with damaged files. But unfortunately after the rebuild, data on disks was so corrupt that even chkdsk could not repair it. It detected that file system was NTFS, but could not do much more. Windows reported partition as being RAW, so I could not access files which were good after the restart but before the completion of rebuild process. So I think that rebuild process instead of rearranging data on all of the disks, screwed data up slowely while "rebuilding".

               

              So when I understood that nothing could be recovered, I did a full format of an array (that took almost 24h) and now it works. Using 9.5 all of the drives were being marked with warning sign from time to time for no reason (some kind of SMART warning). But with 9.6 it does not happen anymore and array has been working fine since then.

              • 4. Re: Lost data on RAID5 after adding more drives
                Arny006

                Hello,

                Thanks for your feedback.

                At first, apparently the Operating-System support Hard-Disk-Drives bigger than 2 TB and that is enjoable because I would to enlarge my drive space.

                For second, again apparently the firmware 8.9 and 9.5 don't work properly, so 8.8 and 9.6 are running well.

                But some questions are still open:

                1. Why using the RAID 5 instead of RAID 10? Have you only 3 HDD?

                2. Why only each second firmware is working properly?

                 

                Thanks again and best regards

                • 5. Re: Lost data on RAID5 after adding more drives
                  Slowmo

                  Why Raid 5 instead of Raid 10? Because of high Raid 10 overhead. On Raid5 with 6 drives I get 5Tb space, but it would only be 3Tb on Raid 10. I don't need high performance, becuase the array is mainly used as network storage. By the way, I get about 300Mb/s average read spead on this array anyway, so it is more than I need. Write performance is not important to me.

                  • 6. Re: Lost data on RAID5 after adding more drives
                    Slowmo

                    It has been a long time since I posted my issue and now using 10.6.0.1002 drivers I see the following in Resolved Issues list:

                     

                    3234690

                    Reboot during capacity expansion of a RAID volume   migration (after having added a disk to the volume) causes the volume to go   failed

                     

                    Because I did reboot my system while expand process was still going, this bug might have been the cause of why I lost all the data.