BTRFS filesystem

• Links: docker/docker-btrfs , linux/btrfs/crash201706 , https://wiki.debian.org/Btrfs

• Btrfs space calculator http://carfax.org.uk/btrfs-usage/

  • MODE: single=use all, raid0=smallest disk striped, raid1=2copy(smallest of biggest disk and rest)
• Create a raid1 mirror disk setup raid1 for both data and metadata

  •  mkfs.btrfs -m raid1 -d raid1 /dev/sdb /dev/sdc /dev/sdd /dev/sde 

  • Now find uuid with  sudo blkid /dev/sd* 

    • Add more hd's

      •        btrfs device scan
               btrfs device add /dev/sdc2 /
               btrfs filesystem balance /
               btrfs balance start -dconvert=raid1 -mconvert=raid1 /
               #Monitor progress with
               btrfs balance status /

• reflinks

  •  cp --reflink name1 name4 

  • reflinks are supported by BTRFS and OCFS2 and support transparent copy on write which is especially useful for snapshotting.
  • Note that since separate inodes are used, one can have different permissions to access the same data.
  • Reflinks have the same use as hardlinks, but are more space efficient and generally handle all subsequent operations on a file, not just unlink().
• If you have virtual images on a btrfs, cow(copy on write) will cause a lot of fragmentation, turn it of for the directory of file with
  • this disables cow(fragmentation) and also crc calculations on the files.

    •      chatter +C /path

• defrag with

  •    btrfs filesystem defragment /path/file

  • or faster for large files

    •      cp --reflink=never vbox.vdi vbox.new.vdi ; rm vbox.vdi ;      

• Create raid array, raid1 all data mirrored once.
  • # mkfs.btrfs -m raid1 -d raid1 -L root /dev/disk/by-label/root1 /dev/disk/by-label/root2 /dev/disk/by-lab

el/root3 /dev/disk/by-label/root4

• Mount options, add compress
• subvolumes
  • mount option subvol=homevol

Install on multi disk

• Will install on multiple disks

  •    sudo dpkg-reconfigure grub-pc 

• Use uuid , same for all disks.

  •    sudo blkid /dev/sd* 

BTRFS maintenace

• SCRUB disks in background, recovering for good copy in raid1 config

  •    sudo btrfs filesystem show
       sudo btrfs filesystem df /
       sudo btrfs scrub start /
       sudo btrfs scrub status -d /
       sudo btrfs device stats /

• Resize
  1. Increase the partition (lvm or fdisk)

  2. find the <devid> (disk number) # btrfs filesystem show

  3. btrfs filesystem resize <devid>:max /

• Error trying to convert to raid0

  •    # btrfs balance start -dconvert=raid0 /home
       ERROR: error during balancing '/home' - No space left on device
       In dmesg  btrfs: 262 enospc errors during balance

  • FIX

    •   sudo btrfs fi balance start -dusage=10 /
        btrfs balance start -dconvert=raid0 /home

• Remove a drive for a array. Can take long, as data is moved off.

  • btrfs fi show
    btrfs device delete /dev/sde1 /

• if single disk in raid1 fails, best course of action if you can replace the broken disk immediately is to convert to single drive.

  btrfs balance start -dconvert=single -mconvert=dup /whatever

Problem with filesystem full

• 2016-03 a big struggle with a full btrfs root fs and then corrupt grub2 on top of the problem.

  • grub rescue>

    • used >set prefix=hd(0,1)/@snapxxxxx/boot/grub

    • This loaded old non corrupt grub2 normal module and typing normal, loaded old bootloader yeah for btrfs.
    • then at the grub2 menu, edit the boot parameters to put root at /@ and load newer kernel from corrupt root fs.
  • Space Solved with removing snapshots and deleting some files, then re-balance.
    • To delete snapshot's you need to mount disk to get to subvolumes.
      • No space to create dir, mount over root with mount /dev/sda1 /root
      • Delete subvolumes with #btrfs subvolume delete /root/@snapshot*

HP Server Raid tools

• http://dmatwork.blogspot.com/2010/07/installation-of-hp-array-configuration.html

...