Documentation/filesystems/ocfs2.rst

   1 .. SPDX-License-Identifier: GPL-2.0
   2
   3 ================
   4 OCFS2 filesystem
   5 ================
   6
   7 OCFS2 is a general purpose extent based shared disk cluster file
   8 system with many similarities to ext3. It supports 64 bit inode
   9 numbers, and has automatically extending metadata groups which may
  10 also make it attractive for non-clustered use.
  11
  12 You'll want to install the ocfs2-tools package in order to at least
  13 get "mount.ocfs2" and "ocfs2_hb_ctl".
  14
  15 Project web page:    http://ocfs2.wiki.kernel.org
  16 Tools git tree:      https://github.com/markfasheh/ocfs2-tools
  17 OCFS2 mailing lists: http://oss.oracle.com/projects/ocfs2/mailman/
  18
  19 All code copyright 2005 Oracle except when otherwise noted.
  20
  21 Credits
  22 =======
  23
  24 Lots of code taken from ext3 and other projects.
  25
  26 Authors in alphabetical order:
  27
  28 - Joel Becker   <joel.becker@oracle.com>
  29 - Zach Brown    <zach.brown@oracle.com>
  30 - Mark Fasheh   <mfasheh@suse.com>
  31 - Kurt Hackel   <kurt.hackel@oracle.com>
  32 - Tao Ma        <tao.ma@oracle.com>
  33 - Sunil Mushran <sunil.mushran@oracle.com>
  34 - Manish Singh  <manish.singh@oracle.com>
  35 - Tiger Yang    <tiger.yang@oracle.com>
  36
  37 Caveats
  38 =======
  39 Features which OCFS2 does not support yet:
  40
  41         - Directory change notification (F_NOTIFY)
  42         - Distributed Caching (F_SETLEASE/F_GETLEASE/break_lease)
  43
  44 Mount options
  45 =============
  46
  47 OCFS2 supports the following mount options:
  48
  49 (*) == default
  50
  51 ======================= ========================================================
  52 barrier=1               This enables/disables barriers. barrier=0 disables it,
  53                         barrier=1 enables it.
  54 errors=remount-ro(*)    Remount the filesystem read-only on an error.
  55 errors=panic            Panic and halt the machine if an error occurs.
  56 intr            (*)     Allow signals to interrupt cluster operations.
  57 nointr                  Do not allow signals to interrupt cluster
  58                         operations.
  59 noatime                 Do not update access time.
  60 relatime(*)             Update atime if the previous atime is older than
  61                         mtime or ctime
  62 strictatime             Always update atime, but the minimum update interval
  63                         is specified by atime_quantum.
  64 atime_quantum=60(*)     OCFS2 will not update atime unless this number
  65                         of seconds has passed since the last update.
  66                         Set to zero to always update atime. This option need
  67                         work with strictatime.
  68 data=ordered    (*)     All data are forced directly out to the main file
  69                         system prior to its metadata being committed to the
  70                         journal.
  71 data=writeback          Data ordering is not preserved, data may be written
  72                         into the main file system after its metadata has been
  73                         committed to the journal.
  74 preferred_slot=0(*)     During mount, try to use this filesystem slot first. If
  75                         it is in use by another node, the first empty one found
  76                         will be chosen. Invalid values will be ignored.
  77 commit=nrsec    (*)     Ocfs2 can be told to sync all its data and metadata
  78                         every 'nrsec' seconds. The default value is 5 seconds.
  79                         This means that if you lose your power, you will lose
  80                         as much as the latest 5 seconds of work (your
  81                         filesystem will not be damaged though, thanks to the
  82                         journaling).  This default value (or any low value)
  83                         will hurt performance, but it's good for data-safety.
  84                         Setting it to 0 will have the same effect as leaving
  85                         it at the default (5 seconds).
  86                         Setting it to very large values will improve
  87                         performance.
  88 localalloc=8(*)         Allows custom localalloc size in MB. If the value is too
  89                         large, the fs will silently revert it to the default.
  90 localflocks             This disables cluster aware flock.
  91 inode64                 Indicates that Ocfs2 is allowed to create inodes at
  92                         any location in the filesystem, including those which
  93                         will result in inode numbers occupying more than 32
  94                         bits of significance.
  95 user_xattr      (*)     Enables Extended User Attributes.
  96 nouser_xattr            Disables Extended User Attributes.
  97 acl                     Enables POSIX Access Control Lists support.
  98 noacl           (*)     Disables POSIX Access Control Lists support.
  99 resv_level=2    (*)     Set how aggressive allocation reservations will be.
 100                         Valid values are between 0 (reservations off) to 8
 101                         (maximum space for reservations).
 102 dir_resv_level= (*)     By default, directory reservations will scale with file
 103                         reservations - users should rarely need to change this
 104                         value. If allocation reservations are turned off, this
 105                         option will have no effect.
 106 coherency=full  (*)     Disallow concurrent O_DIRECT writes, cluster inode
 107                         lock will be taken to force other nodes drop cache,
 108                         therefore full cluster coherency is guaranteed even
 109                         for O_DIRECT writes.
 110 coherency=buffered      Allow concurrent O_DIRECT writes without EX lock among
 111                         nodes, which gains high performance at risk of getting
 112                         stale data on other nodes.
 113 journal_async_commit    Commit block can be written to disk without waiting
 114                         for descriptor blocks. If enabled older kernels cannot
 115                         mount the device. This will enable 'journal_checksum'
 116                         internally.
 117 ======================= ========================================================