* Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine @ 2021-09-08 14:35 brian 2021-09-08 21:39 ` Dave Chinner 0 siblings, 1 reply; 8+ messages in thread From: brian @ 2021-09-08 14:35 UTC (permalink / raw) To: linux-xfs I have CentOS 7 Virtualbox virtual machine (VM) that gives the following error on boot(removed parts replaced with …) ---------------------------------------------------------------------------- XFS (dm-2): Metadata corruption detected at xfs_inode_buf_verify... XFS (dm-2): unmount and run xfs_repair XFS (dm-2): First 128 bytes of corrupted metadata buffer: … XFS (dm-2): metadata I/O error in “xlog_recov_do..(read#2)” at daddr 0x1b32 error 117 ---------------------------------------------------------------------------- I tried entered emergency mode, entered password and tried: xfs_repair -L /dev/mapper/centos-root But I got the error: /dev/mapper/centos-root contains a mounted filesystem /dev/mapper/centos-root contains a mounted and writable filesystem Next I booted from the Centos ISO then went troubleshooting → Rescue a CentOS system → 1) Continue This fails. I get lines of = marks. When I left overnight, I had a blank screen. When I ALT+Tab to program-log and then back to main, I got a screen of scrolling errors. I nightly backup my data including this VM with rsync to a USB drive. My last backup was while the VM was running. In Virtualbox and on my computers drive, I deleted the VM using the virtualbox GUI. I thought this would move the VM to the trash but it permanently deleted the VM. The problem I'm having is with the backup VM from my USB drive. How should I troubleshoot and fix this problem? My main goal is to get my data off the VM. I also unsuccessful with mounting the .vdi file in another computer. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine 2021-09-08 14:35 Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine brian @ 2021-09-08 21:39 ` Dave Chinner 2021-09-08 23:08 ` brian 0 siblings, 1 reply; 8+ messages in thread From: Dave Chinner @ 2021-09-08 21:39 UTC (permalink / raw) To: brian; +Cc: linux-xfs On Wed, Sep 08, 2021 at 10:35:57AM -0400, brian wrote: > I have CentOS 7 Virtualbox virtual machine (VM) that gives the following > error on boot(removed parts replaced with …) > ---------------------------------------------------------------------------- > XFS (dm-2): Metadata corruption detected at xfs_inode_buf_verify... > XFS (dm-2): unmount and run xfs_repair > XFS (dm-2): First 128 bytes of corrupted metadata buffer: > … You cut out the part of the error report that actually means something to us. Can you post the error message in full? > XFS (dm-2): metadata I/O error in “xlog_recov_do..(read#2)” at daddr 0x1b32 > error 117 > ---------------------------------------------------------------------------- > > I tried entered emergency mode, entered password and tried: > xfs_repair -L /dev/mapper/centos-root > > But I got the error: > /dev/mapper/centos-root contains a mounted filesystem > /dev/mapper/centos-root contains a mounted and writable filesystem $ man xfs_repair .... -d Repair dangerously. Allow xfs_repair to repair an XFS filesystem mounted read only. This is typically done on a root filesystem from single user mode, immediately followed by a reboot. > Next I booted from the Centos ISO then went > troubleshooting → Rescue a CentOS system → 1) Continue > > This fails. I get lines of = marks. When I left overnight, I had a blank > screen. When I ALT+Tab to program-log and then back to main, I got a screen > of scrolling errors. Which typically happens if the filesystem full of inconsistencies and corruptions. > I nightly backup my data including this VM with rsync to a USB drive. My > last backup was while the VM was running. In Virtualbox and on my computers > drive, I deleted the VM using the virtualbox GUI. I thought this would move > the VM to the trash but it permanently deleted the VM. The problem I'm > having is with the backup VM from my USB drive. How did you do the backup? Was it an image file backup from the host, or a full filesystem rsync from inside the guest ? > How should I troubleshoot and fix this problem? My main goal is to get my > data off the VM. I also unsuccessful with mounting the .vdi file in another > computer. Sounds like you simply backed up the image file on the host? If so, did you freeze the filesystem in the guest (or the whole guest VM) while you did the backup? If the guest fs wasnt' frozen, and the VM was active while the backup was done as you say, then it is guaranteed that the image file that was copied will contain an inconsistent image of the filesystem. If there was lots of modifications occurring at the time of the copy being run, it will be full of corruptions and may well be completely unrecoverable. DO you have any older backups you can try to recover from? Cheers, Dave. -- Dave Chinner david@fromorbit.com ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine 2021-09-08 21:39 ` Dave Chinner @ 2021-09-08 23:08 ` brian 2021-09-09 2:54 ` Dave Chinner 0 siblings, 1 reply; 8+ messages in thread From: brian @ 2021-09-08 23:08 UTC (permalink / raw) To: Dave Chinner; +Cc: linux-xfs Here is an update of the error. If something looks off, let me know and I'll double check I transcribed correct. Let me know if there is a way to post a screenshot. ---------------------------------------------------------------------------- XFS (dm-2): Metadata corruption detected at xfs_inode_buf_verify+0x14d/0x160 [xfs] xfs_inode block 0x1b5c1c0 xfs_inode_buf_verify XFS (dm-2): Unmount and run xfs_repair XFS (dm-2): First 128 bytes of corrupted metadata buffer: ffffae6842564000: 16 49 77 8a 32 7e 72 52 14 bb 51 98 7c a5 2c 9a .Iw.2~rR..Q.|.,. ffffae6842564010: dd 28 4d 94 88 03 2b 8c 99 33 67 ca 6a d5 aa c9 .(M...+..3g.j... ffffae6842564020: f8 f8 78 c7 90 fc f5 af 7f 95 03 07 0e 0c 4a 37 ..x...........J7 ffffae6842564030: 7c f8 70 c7 09 14 c7 81 a5 a7 a7 cb a8 a8 28 b9 |.p...........(. ffffae6842564040: 68 d1 a2 a0 73 f5 ae d5 73 94 23 3d 3d 5d 46 46 h...s...s.#--]FF ffffae6842564050: 46 ca f9 f3 e7 3b 69 4c d3 94 6d db b6 75 14 3d F....;iL..m..u.= ffffae6842564060: 30 68 77 ec d8 e1 a4 d9 b7 6f 9f 0c 0b 0b 0b 32 )hw......o.....2 ffffae6842564070: 8e 9d 29 aa 03 4b 4c 4c 0c 52 66 29 a5 7c f4 d1 ..)..KLL.Rf).|.. XFS (dm-2): metadata I/O error in “xlog_recov_do..(read#2)” at daddr 0x1b32 error 117 ---------------------------------------------------------------------------- My computer is an older CentOS 6 box (on my todo list to replace). I have an array of VirtualBox VMs. All my data including VMs is in /home/brian directory. I run rsync to backup /home/brian directory to a USB drive. Rsync treats the files of the VM just like any other file. I noticed that the VMs can get corrupted if I leave them running during back up with rsync so I normally shut them down. For this instance, I didn't shut down the VM when I launched rsync. @Dave Regarding your image file question, I don't believe the above is an image file. Rsync copies file by file. And I can navigate to the USB like my internal hdd. Regarding the freeze files system, no I did not freeze the guest. Regarding the older backups, I just have the rsync backup on a USB drive. I've been working with a copy from the USB to try at get this machine back, thus the unmodified USB rsync that was taken while the machine was running exists. While rsync was running, I had Selenium running web app tests in Chrome. All the data from my program was being written to a shared folder. Chrome has it's own internal cache that could be changing but overall, I wouldn't expect the file system to be changing too much especially the booting part. My ultimate goal is to get to the data on this VM. I've tried to mount this .vdi file in another instance of CentOS but have not been successful. I may spin up an Ubuntu instance to try to get to the data. The data on this VM can be recreated but I prefer not to have to redo the work. Thank you for your help. Brian On 09/08/2021 05:39 PM, Dave Chinner wrote: > On Wed, Sep 08, 2021 at 10:35:57AM -0400, brian wrote: >> I have CentOS 7 Virtualbox virtual machine (VM) that gives the following >> error on boot(removed parts replaced with …) >> ---------------------------------------------------------------------------- >> XFS (dm-2): Metadata corruption detected at xfs_inode_buf_verify... >> XFS (dm-2): unmount and run xfs_repair >> XFS (dm-2): First 128 bytes of corrupted metadata buffer: >> … > You cut out the part of the error report that actually means > something to us. Can you post the error message in full? > >> XFS (dm-2): metadata I/O error in “xlog_recov_do..(read#2)” at daddr 0x1b32 >> error 117 >> ---------------------------------------------------------------------------- >> >> I tried entered emergency mode, entered password and tried: >> xfs_repair -L /dev/mapper/centos-root >> >> But I got the error: >> /dev/mapper/centos-root contains a mounted filesystem >> /dev/mapper/centos-root contains a mounted and writable filesystem > $ man xfs_repair > .... > -d Repair dangerously. Allow xfs_repair to repair an XFS > filesystem mounted read only. This is typically done on a > root filesystem from single user mode, immediately followed > by a reboot. > >> Next I booted from the Centos ISO then went >> troubleshooting → Rescue a CentOS system → 1) Continue >> >> This fails. I get lines of = marks. When I left overnight, I had a blank >> screen. When I ALT+Tab to program-log and then back to main, I got a screen >> of scrolling errors. > Which typically happens if the filesystem full of inconsistencies > and corruptions. > >> I nightly backup my data including this VM with rsync to a USB drive. My >> last backup was while the VM was running. In Virtualbox and on my computers >> drive, I deleted the VM using the virtualbox GUI. I thought this would move >> the VM to the trash but it permanently deleted the VM. The problem I'm >> having is with the backup VM from my USB drive. > How did you do the backup? Was it an image file backup from the > host, or a full filesystem rsync from inside the guest ? > >> How should I troubleshoot and fix this problem? My main goal is to get my >> data off the VM. I also unsuccessful with mounting the .vdi file in another >> computer. > Sounds like you simply backed up the image file on the host? > > If so, did you freeze the filesystem in the guest (or the whole > guest VM) while you did the backup? If the guest fs wasnt' frozen, > and the VM was active while the backup was done as you say, then it > is guaranteed that the image file that was copied will contain an > inconsistent image of the filesystem. If there was lots of > modifications occurring at the time of the copy being run, it will > be full of corruptions and may well be completely unrecoverable. > > DO you have any older backups you can try to recover from? > > Cheers, > > Dave. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine 2021-09-08 23:08 ` brian @ 2021-09-09 2:54 ` Dave Chinner 2021-09-09 3:39 ` brian 2021-09-09 19:11 ` brian 0 siblings, 2 replies; 8+ messages in thread From: Dave Chinner @ 2021-09-09 2:54 UTC (permalink / raw) To: brian; +Cc: linux-xfs On Wed, Sep 08, 2021 at 07:08:54PM -0400, brian wrote: > Here is an update of the error. If something looks off, let me know and > I'll double check I transcribed correct. Let me know if there is a way to > post a screenshot. > > ---------------------------------------------------------------------------- > > XFS (dm-2): Metadata corruption detected at xfs_inode_buf_verify+0x14d/0x160 > [xfs] xfs_inode block 0x1b5c1c0 xfs_inode_buf_verify > XFS (dm-2): Unmount and run xfs_repair > XFS (dm-2): First 128 bytes of corrupted metadata buffer: > ffffae6842564000: 16 49 77 8a 32 7e 72 52 14 bb 51 98 7c a5 2c 9a > .Iw.2~rR..Q.|.,. > ffffae6842564010: dd 28 4d 94 88 03 2b 8c 99 33 67 ca 6a d5 aa c9 > .(M...+..3g.j... > ffffae6842564020: f8 f8 78 c7 90 fc f5 af 7f 95 03 07 0e 0c 4a 37 > ..x...........J7 > ffffae6842564030: 7c f8 70 c7 09 14 c7 81 a5 a7 a7 cb a8 a8 28 b9 > |.p...........(. > ffffae6842564040: 68 d1 a2 a0 73 f5 ae d5 73 94 23 3d 3d 5d 46 46 > h...s...s.#--]FF > ffffae6842564050: 46 ca f9 f3 e7 3b 69 4c d3 94 6d db b6 75 14 3d > F....;iL..m..u.= > ffffae6842564060: 30 68 77 ec d8 e1 a4 d9 b7 6f 9f 0c 0b 0b 0b 32 > )hw......o.....2 > ffffae6842564070: 8e 9d 29 aa 03 4b 4c 4c 0c 52 66 29 a5 7c f4 d1 > ..)..KLL.Rf).|.. > XFS (dm-2): metadata I/O error in “xlog_recov_do..(read#2)” at daddr 0x1b32 > error 117 Ok, that doesn't contain inodes. It's not even recognisable as filesystem metadata at all. > ---------------------------------------------------------------------------- > > My computer is an older CentOS 6 box (on my todo list to replace). I have > an array of VirtualBox VMs. All my data including VMs is in /home/brian > directory. I run rsync to backup /home/brian directory to a USB drive. > Rsync treats the files of the VM just like any other file. I noticed that > the VMs can get corrupted if I leave them running during back up with rsync > so I normally shut them down. For this instance, I didn't shut down the VM > when I launched rsync. > > @Dave > Regarding your image file question, I don't believe the above is an image > file. Rsync copies file by file. And I can navigate to the USB like my > internal hdd. > > Regarding the freeze files system, no I did not freeze the guest. > > Regarding the older backups, I just have the rsync backup on a USB drive. > I've been working with a copy from the USB to try at get this machine back, > thus the unmodified USB rsync that was taken while the machine was running > exists. > > While rsync was running, I had Selenium running web app tests in Chrome. > All the data from my program was being written to a shared folder. Chrome > has it's own internal cache that could be changing but overall, I wouldn't > expect the file system to be changing too much especially the booting part. Log recovery covers anything that changed in the filesystem - it has nothing to do with what is being accessed by boot operations. > My ultimate goal is to get to the data on this VM. I've tried to mount this > .vdi file in another instance of CentOS but have not been successful. I may > spin up an Ubuntu instance to try to get to the data. The data on this VM > can be recreated but I prefer not to have to redo the work. 'mount -o ro,norecovery <dev>' will allow you to skip log recovery and hopefully let you navigate the filesystem. It will not be error free, though, because log recovery is needed for that to occur. If the filesystem looks mostly intact, copy all the data off it you can. Then unmount it is and run xfs_repair -L <dev> on it to blow away the log and repair any corruption found. This will cause data loss, so don't do this unless you are prepared for it of have a complete copy of the broken filesystem. I'd highly recommend doing this with a recent xfs_repair, not the tools that came with CentOS 6. You could also do dry-run testing by taking a metadump of the broken filesystem and restoring that to an image file and running reapir on that image file. That will at least tell you how much of the directory structure will be trashed and wht you'll be able to access after repair. It may be that mounting w/ norecovery is your best bet of recovering the contents of the bad filesystems, so do that first... Cheers, Dave. -- Dave Chinner david@fromorbit.com ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine 2021-09-09 2:54 ` Dave Chinner @ 2021-09-09 3:39 ` brian 2021-09-09 19:11 ` brian 1 sibling, 0 replies; 8+ messages in thread From: brian @ 2021-09-09 3:39 UTC (permalink / raw) To: Dave Chinner; +Cc: linux-xfs @Dave, I would consider myself a beginner with Linux file systems. I'm not sure how to implement your suggestions. If you have any suggested places to dig deeper, let me know. When you say 'mount -o ro,norecovery <dev>', do you mean to boot the VM I'm having problems with and add this from the prompt? I'm not sure what <dev> is. I assume you mean /dev/.... Also, I thought I had to add what is being mounted (e.g., image file). So far, I've been using the xfs_repair that is build into the VM (CentOS 7). I've only accessed xfs_repair while in the emergency mode. I'm not sure how to use another version; I didn't think the emergency mode had internet. Can you give me more direction in how to execute on your suggestions? Thank you again for your help. I appreciate all the help. Brian On 09/08/2021 10:54 PM, Dave Chinner wrote: > On Wed, Sep 08, 2021 at 07:08:54PM -0400, brian wrote: >> Here is an update of the error. If something looks off, let me know and >> I'll double check I transcribed correct. Let me know if there is a way to >> post a screenshot. >> >> ---------------------------------------------------------------------------- >> >> XFS (dm-2): Metadata corruption detected at xfs_inode_buf_verify+0x14d/0x160 >> [xfs] xfs_inode block 0x1b5c1c0 xfs_inode_buf_verify >> XFS (dm-2): Unmount and run xfs_repair >> XFS (dm-2): First 128 bytes of corrupted metadata buffer: >> ffffae6842564000: 16 49 77 8a 32 7e 72 52 14 bb 51 98 7c a5 2c 9a >> .Iw.2~rR..Q.|.,. >> ffffae6842564010: dd 28 4d 94 88 03 2b 8c 99 33 67 ca 6a d5 aa c9 >> .(M...+..3g.j... >> ffffae6842564020: f8 f8 78 c7 90 fc f5 af 7f 95 03 07 0e 0c 4a 37 >> ..x...........J7 >> ffffae6842564030: 7c f8 70 c7 09 14 c7 81 a5 a7 a7 cb a8 a8 28 b9 >> |.p...........(. >> ffffae6842564040: 68 d1 a2 a0 73 f5 ae d5 73 94 23 3d 3d 5d 46 46 >> h...s...s.#--]FF >> ffffae6842564050: 46 ca f9 f3 e7 3b 69 4c d3 94 6d db b6 75 14 3d >> F....;iL..m..u.= >> ffffae6842564060: 30 68 77 ec d8 e1 a4 d9 b7 6f 9f 0c 0b 0b 0b 32 >> )hw......o.....2 >> ffffae6842564070: 8e 9d 29 aa 03 4b 4c 4c 0c 52 66 29 a5 7c f4 d1 >> ..)..KLL.Rf).|.. >> XFS (dm-2): metadata I/O error in “xlog_recov_do..(read#2)” at daddr 0x1b32 >> error 117 > Ok, that doesn't contain inodes. It's not even recognisable > as filesystem metadata at all. > >> ---------------------------------------------------------------------------- >> >> My computer is an older CentOS 6 box (on my todo list to replace). I have >> an array of VirtualBox VMs. All my data including VMs is in /home/brian >> directory. I run rsync to backup /home/brian directory to a USB drive. >> Rsync treats the files of the VM just like any other file. I noticed that >> the VMs can get corrupted if I leave them running during back up with rsync >> so I normally shut them down. For this instance, I didn't shut down the VM >> when I launched rsync. >> >> @Dave >> Regarding your image file question, I don't believe the above is an image >> file. Rsync copies file by file. And I can navigate to the USB like my >> internal hdd. >> >> Regarding the freeze files system, no I did not freeze the guest. >> >> Regarding the older backups, I just have the rsync backup on a USB drive. >> I've been working with a copy from the USB to try at get this machine back, >> thus the unmodified USB rsync that was taken while the machine was running >> exists. >> >> While rsync was running, I had Selenium running web app tests in Chrome. >> All the data from my program was being written to a shared folder. Chrome >> has it's own internal cache that could be changing but overall, I wouldn't >> expect the file system to be changing too much especially the booting part. > Log recovery covers anything that changed in the filesystem - it has > nothing to do with what is being accessed by boot operations. > >> My ultimate goal is to get to the data on this VM. I've tried to mount this >> .vdi file in another instance of CentOS but have not been successful. I may >> spin up an Ubuntu instance to try to get to the data. The data on this VM >> can be recreated but I prefer not to have to redo the work. > 'mount -o ro,norecovery <dev>' will allow you to skip log recovery > and hopefully let you navigate the filesystem. It will not be error > free, though, because log recovery is needed for that to occur. > > If the filesystem looks mostly intact, copy all the data off it you > can. Then unmount it is and run xfs_repair -L <dev> on it to blow > away the log and repair any corruption found. This will cause data > loss, so don't do this unless you are prepared for it of have a > complete copy of the broken filesystem. > > I'd highly recommend doing this with a recent xfs_repair, not the > tools that came with CentOS 6. > > You could also do dry-run testing by taking a metadump of the broken > filesystem and restoring that to an image file and running reapir on > that image file. That will at least tell you how much of the > directory structure will be trashed and wht you'll be able to access > after repair. > > It may be that mounting w/ norecovery is your best bet of recovering > the contents of the bad filesystems, so do that first... > > Cheers, > > Dave. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine 2021-09-09 2:54 ` Dave Chinner 2021-09-09 3:39 ` brian @ 2021-09-09 19:11 ` brian 2021-09-09 21:09 ` Dave Chinner 1 sibling, 1 reply; 8+ messages in thread From: brian @ 2021-09-09 19:11 UTC (permalink / raw) To: Dave Chinner; +Cc: linux-xfs I switched over to ubuntu and got the error "Device or resource busy". How do I get around this error? Here is what I did: --------------------------------------------------------- sudo apt-get install qemu sudo apt install qemu-utils sudo apt-get install xfsprogs sudo modprobe nbd max_part=8 sudo qemu-nbd --connect=/dev/nbd0 /media/sf_virtual_machine_share/centoOS7Python3p9_tmp-disk1.vdi sudo xfs_repair /dev/nbd0p2 --------------------------------------------------------- I got the error: --------------------------------------------------------- xfs_repair: cannot open /dev/nbd0p2: Device or resource busy --------------------------------------------------------- When I tried xfs_repair with the -n flag: --------------------------------------------------------- sudo xfs_repair -n /dev/nbd0p2 --------------------------------------------------------- I got (I replaced some content with ... Most of the content was ...): --------------------------------------------------------- Phase 1 - find and verify superblock... bad primary superblock - bad magic number !!! ... attempting to find secondary superblock... Sorry, could not find valid secondary superblock --------------------------------------------------------- Here are some things I ran while trying to debug. cmd: --------------------------------------------------------- sudo partprobe /dev/nbd0p2 --------------------------------------------------------- Result: --------------------------------------------------------- Error: Partition(s) 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64 on /dev/nbd0p2 have been written, but we have been unable to inform the kernel of the change, probably because it/they are in use. As a result, the old partition(s) will remain in use. You should reboot now before making further changes. --------------------------------------------------------- cmd: --------------------------------------------------------- sudo fdisk -l /dev/nbd0p2 --------------------------------------------------------- Result: --------------------------------------------------------- Disk /dev/nbd0p2: 101.72 GiB, 109207093248 bytes, 213295104 sectors Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes brian@brian-VirtualBox:~/Desktop$ sudo lvmdiskscan /dev/loop0 [ 219.00 MiB] /dev/loop1 [ <55.44 MiB] /dev/sda1 [ 512.00 MiB] /dev/nbd0p1 [ 1.00 GiB] /dev/loop2 [ <65.10 MiB] /dev/nbd0p2 [ <101.71 GiB] LVM physical volume /dev/loop3 [ <50.96 MiB] /dev/loop4 [ <32.30 MiB] /dev/loop5 [ 32.30 MiB] /dev/sda5 [ <2.00 TiB] 0 disks 9 partitions 0 LVM physical volume whole disks 1 LVM physical volume --------------------------------------------------------- cmd: --------------------------------------------------------- sudo lvscan --------------------------------------------------------- Result: --------------------------------------------------------- ACTIVE '/dev/centos/swap' [2.00 GiB] inherit ACTIVE '/dev/centos/home' [<49.70 GiB] inherit ACTIVE '/dev/centos/root' [50.00 GiB] inherit --------------------------------------------------------- cmd: --------------------------------------------------------- sudo lvm pvscan --------------------------------------------------------- Result: --------------------------------------------------------- PV /dev/nbd0p2 VG centos lvm2 [101.70 GiB / 4.00 MiB free] Total: 1 [101.70 GiB] / in use: 1 [101.70 GiB] / in no VG: 0 [0 ] --------------------------------------------------------- cmd: --------------------------------------------------------- sudo lvm lvs --------------------------------------------------------- result: --------------------------------------------------------- LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert home centos -wi-a----- <49.70g root centos -wi-a----- 50.00g swap centos -wi-a----- 2.00g --------------------------------------------------------- cmd: --------------------------------------------------------- sudo e2fsck /dev/nbd0p2 --------------------------------------------------------- result: --------------------------------------------------------- e2fsck 1.45.5 (07-Jan-2020) /dev/nbd0p2 is in use. e2fsck: Cannot continue, aborting. --------------------------------------------------------- cmd: --------------------------------------------------------- fsck -N /dev/nbd0p2 --------------------------------------------------------- result: --------------------------------------------------------- fsck from util-linux 2.34 [/usr/sbin/fsck.ext2 (1) -- /dev/nbd0p2] fsck.ext2 /dev/nbd0p2 --------------------------------------------------------- cmd: --------------------------------------------------------- sudo stat -f /dev/nbd0p2 --------------------------------------------------------- result: --------------------------------------------------------- File: "/dev/nbd0p2" ID: 0 Namelen: 255 Type: tmpfs Block size: 4096 Fundamental block size: 4096 Blocks: Total: 495622 Free: 495622 Available: 495622 Inodes: Total: 495622 Free: 495101 --------------------------------------------------------- On 09/08/2021 10:54 PM, Dave Chinner wrote: > On Wed, Sep 08, 2021 at 07:08:54PM -0400, brian wrote: >> Here is an update of the error. If something looks off, let me know and >> I'll double check I transcribed correct. Let me know if there is a way to >> post a screenshot. >> >> ---------------------------------------------------------------------------- >> >> XFS (dm-2): Metadata corruption detected at xfs_inode_buf_verify+0x14d/0x160 >> [xfs] xfs_inode block 0x1b5c1c0 xfs_inode_buf_verify >> XFS (dm-2): Unmount and run xfs_repair >> XFS (dm-2): First 128 bytes of corrupted metadata buffer: >> ffffae6842564000: 16 49 77 8a 32 7e 72 52 14 bb 51 98 7c a5 2c 9a >> .Iw.2~rR..Q.|.,. >> ffffae6842564010: dd 28 4d 94 88 03 2b 8c 99 33 67 ca 6a d5 aa c9 >> .(M...+..3g.j... >> ffffae6842564020: f8 f8 78 c7 90 fc f5 af 7f 95 03 07 0e 0c 4a 37 >> ..x...........J7 >> ffffae6842564030: 7c f8 70 c7 09 14 c7 81 a5 a7 a7 cb a8 a8 28 b9 >> |.p...........(. >> ffffae6842564040: 68 d1 a2 a0 73 f5 ae d5 73 94 23 3d 3d 5d 46 46 >> h...s...s.#--]FF >> ffffae6842564050: 46 ca f9 f3 e7 3b 69 4c d3 94 6d db b6 75 14 3d >> F....;iL..m..u.= >> ffffae6842564060: 30 68 77 ec d8 e1 a4 d9 b7 6f 9f 0c 0b 0b 0b 32 >> )hw......o.....2 >> ffffae6842564070: 8e 9d 29 aa 03 4b 4c 4c 0c 52 66 29 a5 7c f4 d1 >> ..)..KLL.Rf).|.. >> XFS (dm-2): metadata I/O error in “xlog_recov_do..(read#2)” at daddr 0x1b32 >> error 117 > Ok, that doesn't contain inodes. It's not even recognisable > as filesystem metadata at all. > >> ---------------------------------------------------------------------------- >> >> My computer is an older CentOS 6 box (on my todo list to replace). I have >> an array of VirtualBox VMs. All my data including VMs is in /home/brian >> directory. I run rsync to backup /home/brian directory to a USB drive. >> Rsync treats the files of the VM just like any other file. I noticed that >> the VMs can get corrupted if I leave them running during back up with rsync >> so I normally shut them down. For this instance, I didn't shut down the VM >> when I launched rsync. >> >> @Dave >> Regarding your image file question, I don't believe the above is an image >> file. Rsync copies file by file. And I can navigate to the USB like my >> internal hdd. >> >> Regarding the freeze files system, no I did not freeze the guest. >> >> Regarding the older backups, I just have the rsync backup on a USB drive. >> I've been working with a copy from the USB to try at get this machine back, >> thus the unmodified USB rsync that was taken while the machine was running >> exists. >> >> While rsync was running, I had Selenium running web app tests in Chrome. >> All the data from my program was being written to a shared folder. Chrome >> has it's own internal cache that could be changing but overall, I wouldn't >> expect the file system to be changing too much especially the booting part. > Log recovery covers anything that changed in the filesystem - it has > nothing to do with what is being accessed by boot operations. > >> My ultimate goal is to get to the data on this VM. I've tried to mount this >> .vdi file in another instance of CentOS but have not been successful. I may >> spin up an Ubuntu instance to try to get to the data. The data on this VM >> can be recreated but I prefer not to have to redo the work. > 'mount -o ro,norecovery <dev>' will allow you to skip log recovery > and hopefully let you navigate the filesystem. It will not be error > free, though, because log recovery is needed for that to occur. > > If the filesystem looks mostly intact, copy all the data off it you > can. Then unmount it is and run xfs_repair -L <dev> on it to blow > away the log and repair any corruption found. This will cause data > loss, so don't do this unless you are prepared for it of have a > complete copy of the broken filesystem. > > I'd highly recommend doing this with a recent xfs_repair, not the > tools that came with CentOS 6. > > You could also do dry-run testing by taking a metadump of the broken > filesystem and restoring that to an image file and running reapir on > that image file. That will at least tell you how much of the > directory structure will be trashed and wht you'll be able to access > after repair. > > It may be that mounting w/ norecovery is your best bet of recovering > the contents of the bad filesystems, so do that first... > > Cheers, > > Dave. ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine 2021-09-09 19:11 ` brian @ 2021-09-09 21:09 ` Dave Chinner 2021-09-10 3:26 ` brian 0 siblings, 1 reply; 8+ messages in thread From: Dave Chinner @ 2021-09-09 21:09 UTC (permalink / raw) To: brian; +Cc: linux-xfs On Thu, Sep 09, 2021 at 03:11:45PM -0400, brian wrote: > I switched over to ubuntu and got the error "Device or resource busy". How > do I get around this error? > > Here is what I did: > --------------------------------------------------------- > sudo apt-get install qemu > sudo apt install qemu-utils > sudo apt-get install xfsprogs > > sudo modprobe nbd max_part=8 > > sudo qemu-nbd --connect=/dev/nbd0 > /media/sf_virtual_machine_share/centoOS7Python3p9_tmp-disk1.vdi > > sudo xfs_repair /dev/nbd0p2 > --------------------------------------------------------- > > I got the error: > --------------------------------------------------------- > xfs_repair: cannot open /dev/nbd0p2: Device or resource busy > --------------------------------------------------------- Because /dev/nbd0p2 is not the device the filesystem is on. The filesystem is on a lvm volume: > brian@brian-VirtualBox:~/Desktop$ sudo lvmdiskscan > /dev/loop0 [ 219.00 MiB] > /dev/loop1 [ <55.44 MiB] > /dev/sda1 [ 512.00 MiB] > /dev/nbd0p1 [ 1.00 GiB] > /dev/loop2 [ <65.10 MiB] > /dev/nbd0p2 [ <101.71 GiB] LVM physical volume As noted here. > cmd: > --------------------------------------------------------- > sudo lvscan > --------------------------------------------------------- > > Result: > --------------------------------------------------------- > ACTIVE '/dev/centos/swap' [2.00 GiB] inherit > ACTIVE '/dev/centos/home' [<49.70 GiB] inherit > ACTIVE '/dev/centos/root' [50.00 GiB] inherit > --------------------------------------------------------- And these are the devices inside the LVM volume that contain filesystems/data. Likely the one you are having trouble with is /dev/centos/root, but there may be issues with /dev/centos/home, too. And to answer your other question, "<dev>" is just shorthand for "<insert whatever device your filesystem is on here>". i.e. /dev/centos/root in this case... Cheers, Dave. -- Dave Chinner david@fromorbit.com ^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine 2021-09-09 21:09 ` Dave Chinner @ 2021-09-10 3:26 ` brian 0 siblings, 0 replies; 8+ messages in thread From: brian @ 2021-09-10 3:26 UTC (permalink / raw) To: Dave Chinner; +Cc: linux-xfs I was able to get /home mounted and copy my data off. Here is what I did --------------------------------------------------------- sudo apt-get install qemu sudo apt install qemu-utils sudo apt-get install xfsprogs sudo modprobe nbd max_part=8 sudo qemu-nbd --connect=/dev/nbd0 /media/sf_virtual_machine_share/centoOS7Python3p9_tmp-disk1.vdi sudo mkdir /mnt/dirNbd0p2 sudo mount -o ro,norecovery /dev/mapper/centos-home /mnt/dirNbd0p2 --------------------------------------------------------- I couldn't get the vdi corruption fixed but I did get my data. @Dave Thank you for your help. Brian On 09/09/2021 05:09 PM, Dave Chinner wrote: > On Thu, Sep 09, 2021 at 03:11:45PM -0400, brian wrote: >> I switched over to ubuntu and got the error "Device or resource busy". How >> do I get around this error? >> >> Here is what I did: >> --------------------------------------------------------- >> sudo apt-get install qemu >> sudo apt install qemu-utils >> sudo apt-get install xfsprogs >> >> sudo modprobe nbd max_part=8 >> >> sudo qemu-nbd --connect=/dev/nbd0 >> /media/sf_virtual_machine_share/centoOS7Python3p9_tmp-disk1.vdi >> >> sudo xfs_repair /dev/nbd0p2 >> --------------------------------------------------------- >> >> I got the error: >> --------------------------------------------------------- >> xfs_repair: cannot open /dev/nbd0p2: Device or resource busy >> --------------------------------------------------------- > Because /dev/nbd0p2 is not the device the filesystem is on. The > filesystem is on a lvm volume: > >> brian@brian-VirtualBox:~/Desktop$ sudo lvmdiskscan >> /dev/loop0 [ 219.00 MiB] >> /dev/loop1 [ <55.44 MiB] >> /dev/sda1 [ 512.00 MiB] >> /dev/nbd0p1 [ 1.00 GiB] >> /dev/loop2 [ <65.10 MiB] >> /dev/nbd0p2 [ <101.71 GiB] LVM physical volume > As noted here. > >> cmd: >> --------------------------------------------------------- >> sudo lvscan >> --------------------------------------------------------- >> >> Result: >> --------------------------------------------------------- >> ACTIVE '/dev/centos/swap' [2.00 GiB] inherit >> ACTIVE '/dev/centos/home' [<49.70 GiB] inherit >> ACTIVE '/dev/centos/root' [50.00 GiB] inherit >> --------------------------------------------------------- > And these are the devices inside the LVM volume that contain > filesystems/data. > > Likely the one you are having trouble with is /dev/centos/root, > but there may be issues with /dev/centos/home, too. > > And to answer your other question, "<dev>" is just shorthand for > "<insert whatever device your filesystem is on here>". i.e. > /dev/centos/root in this case... > > Cheers, > > Dave. > ^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2021-09-10 3:26 UTC | newest] Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2021-09-08 14:35 Repair Metadata corruption detected at xfs_inode_buf_verify on CentOS 7 virtual machine brian 2021-09-08 21:39 ` Dave Chinner 2021-09-08 23:08 ` brian 2021-09-09 2:54 ` Dave Chinner 2021-09-09 3:39 ` brian 2021-09-09 19:11 ` brian 2021-09-09 21:09 ` Dave Chinner 2021-09-10 3:26 ` brian
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.