ANNOUNCE-3.0

   1 Subject:  ANNOUNCE: mdadm 3.0 - A tool for managing Soft RAID under Linux
   2
   3 I am pleased to (finally) announce the availability of
   4    mdadm version 3.0
   5
   6 It is available at the usual places:
   7    countrycode=xx.
   8    http://www.${countrycode}kernel.org/pub/linux/utils/raid/mdadm/
   9 and via git at
  10    git://neil.brown.name/mdadm
  11    http://neil.brown.name/git?p=mdadm
  12
  13
  14 This is a major new version and as such should be treated with some
  15 caution.  However it has seen substantial testing and is considerred
  16 to be ready for wide use.
  17
  18
  19 The significant change which justifies the new major version number is
  20 that mdadm can now handle metadata updates entirely in userspace.
  21 This allows mdadm to support metadata formats that the kernel knows
  22 nothing about.
  23
  24 Currently two such metadata formats are supported:
  25   - DDF  - The SNIA standard format
  26   - Intel Matrix - The metadata used by recent Intel ICH controlers.
  27
  28 Also the approach to device names has changed significantly.
  29
  30 If udev is installed on the system, mdadm will not create any devices
  31 in /dev.  Rather it allows udev to manage those devices.  For this to work
  32 as expected, the included udev rules file should be installed.
  33
  34 If udev is not installed, mdadm will still create devices and symlinks
  35 as required, and will also remove them when the array is stopped.
  36
  37 mdadm now requires all devices which do not have a standard name (mdX
  38 or md_dX) to live in the directory /dev/md/.  Names in this directory
  39 will always be created as symlinks back to the standard name in /dev.
  40
  41 The man pages contain some information about the new externally managed
  42 metadata.  However see below for a more condensed overview.
  43
  44 Externally managed metadata introduces the concept of a 'container'.
  45 A container is a collection of (normally) physical devices which have
  46 a common set of metadata.  A container is assembled as an md array, but
  47 is left 'inactive'.
  48
  49 A container can contain one or more data arrays.  These are composed from
  50 slices (partitions?) of various devices in the container.
  51
  52 For example, a 5 devices DDF set can container a RAID1 using the first
  53 half of two devices, a RAID0 using the first half of the remain 3 devices,
  54 and a RAID5 over thte second half of all 5 devices.
  55
  56 A container can be created with
  57
  58    mdadm --create /dev/md0 -e ddf -n5 /dev/sd[abcde]
  59
  60 or "-e imsm" to use the Intel Matrix Storage Manager.
  61
  62 An array can be created within a container either by giving the
  63 container name and the only member:
  64
  65    mdadm -C /dev/md1 --level raid1 -n 2 /dev/md0
  66
  67 or by listing the component devices
  68
  69    mdadm -C /dev/md2 --level raid0 -n 3 /dev/sd[cde]
  70
  71 To assemble a container, it is easiest just to pass each device in turn to
  72 mdadm -I
  73
  74   for i in /dev/sd[abcde]
  75   do mdadm -I $i
  76   done
  77
  78 This will assemble the container and the components.
  79
  80 Alternately the container can be assembled explicitly
  81
  82    mdadm -A /dev/md0 /dev/sd[abcde]
  83
  84 Then the components can all be assembled with
  85
  86    mdadm -I /dev/md0
  87
  88 For each container, mdadm will start a program called "mdmon" which will
  89 monitor the array and effect any metadata updates needed.  The array is
  90 initially assembled readonly. It is up to "mdmon" to mark the metadata
  91 as 'dirty' and which the array to 'read-write'.
  92
  93 The version 0.90 and 1.x metadata formats supported by previous
  94 versions for mdadm are still supported and the kernel still performs
  95 the same updates it use to.  The new 'mdmon' approach is only used for
  96 newly introduced metadata types.
  97
  98 NeilBrown 2nd June 2009