On 2019/2/9 下午6:36, Jakob Schöttl wrote: > Hi, > > I've setup a RAID1 with two disks (disk1 and disk2) and I'm testing the > btrfs replace command. > > After replacing disk2 with disk3, I can only mount > (a) disk1 or disk3 (if both disk are plugged) and > (b) the original disk1 (degraded, if disk3 is unplugged). > > I cannot mount the replacement disk3 if disk1 is unplugged. Sounds like there is one single chunk on disk1, which caused the problem. > >> mount: /mnt: wrong fs type, bad option, bad superblock on /dev/loop3, > missing codepage or helper program, or other error. dmesg please. And btrfs-progs version please. Maybe mkfs is too old to leave SINGLE profile chunks on the original fs. And you could verify the chunk mapping by executing 'btrfs ins dump-tree -t chunk <device>' and paste the output. Thanks, Qu > > What I expect is that both disk1 and disk3 are fully valid and working > after a replace. > > Steps to reproduce: > > dd if=/dev/zero of=/vdisk1 bs=1024 count=300000 > losetup /dev/loop1 /vdisk1 > dd if=/dev/zero of=/vdisk2 bs=1024 count=300000 > losetup /dev/loop2 /vdisk2 > dd if=/dev/zero of=/vdisk3 bs=1024 count=300000 > # losetup /dev/loop3 /vdisk3 # don't plug this device yet > > Create RAID1 file system: > > mkfs.btrfs -L datavol -m raid1 -d raid1 /dev/loop1 /dev/loop2 > > Unplug device 2 to simulate a defect: > > losetup -d /dev/loop2 > > Plug device 3: > > losetup /dev/loop3 /vdisk3 > > Replace device 2 with device 3: > > mount -o degraded /dev/loop1 /mnt > btrfs filesystem show # to get devid of device 2 > btrfs replace start -Br 2 /dev/loop3 /mnt > btrfs replace status /mnt # check success > umount /mnt > > Unplug the original device 1 to see if device 3 has really replaced > device 2: > > losetup -d /dev/loop1 > mount -o degraded /dev/loop3 /mnt > > The mount fails with this error: > >> mount: /mnt: wrong fs type, bad option, bad superblock on /dev/loop3, > missing codepage or helper program, or other error. > > In this situation, btrfs device scan does not change anything and btrfs > filesystem show shows: > >> warning, device 1 is missing >> warning, device 1 is missing >> warning, device 1 is missing >> warning, device 1 is missing >> bad tree block 198180864, bytenr mismatch, want=198180864, have=0 >> ERROR: cannot read chunk root >> Label: 'datavol' uuid: 640e45d3-e741-4a78-a24e-2d8a41c6b8c3 >> Total devices 2 FS bytes used 128.00KiB >> devid 2 size 292.97MiB used 104.00MiB path /dev/loop3 >> *** Some devices missing > > Is this a known problem? Can you reproduce it? Am I doing something wrong? > > Regards, Jakob >
Attachment:
signature.asc
Description: OpenPGP digital signature
