Documentation/filesystems/xfs.txt

   1
   2 The SGI XFS Filesystem
   3 ======================
   4
   5 XFS is a high performance journaling filesystem which originated
   6 on the SGI IRIX platform.  It is completely multi-threaded, can
   7 support large files and large filesystems, extended attributes,
   8 variable block sizes, is extent based, and makes extensive use of
   9 Btrees (directories, extents, free space) to aid both performance
  10 and scalability.
  11
  12 Refer to the documentation at http://oss.sgi.com/projects/xfs/
  13 for further details.  This implementation is on-disk compatible
  14 with the IRIX version of XFS.
  15
  16
  17 Mount Options
  18 =============
  19
  20 When mounting an XFS filesystem, the following options are accepted.
  21
  22   allocsize=size
  23         Sets the buffered I/O end-of-file preallocation size when
  24         doing delayed allocation writeout (default size is 64KiB).
  25         Valid values for this option are page size (typically 4KiB)
  26         through to 1GiB, inclusive, in power-of-2 increments.
  27
  28   attr2/noattr2
  29         The options enable/disable (default is disabled for backward
  30         compatibility on-disk) an "opportunistic" improvement to be
  31         made in the way inline extended attributes are stored on-disk.
  32         When the new form is used for the first time (by setting or
  33         removing extended attributes) the on-disk superblock feature
  34         bit field will be updated to reflect this format being in use.
  35
  36         CRC enabled filesystems always use the attr2 format, and so
  37         will reject the noattr2 mount option if it is set.
  38
  39   barrier
  40         Enables the use of block layer write barriers for writes into
  41         the journal and unwritten extent conversion.  This allows for
  42         drive level write caching to be enabled, for devices that
  43         support write barriers.
  44
  45   discard
  46         Issue command to let the block device reclaim space freed by the
  47         filesystem.  This is useful for SSD devices, thinly provisioned
  48         LUNs and virtual machine images, but may have a performance
  49         impact.
  50
  51   dmapi
  52         Enable the DMAPI (Data Management API) event callouts.
  53         Use with the "mtpt" option.
  54
  55   grpid/bsdgroups and nogrpid/sysvgroups
  56         These options define what group ID a newly created file gets.
  57         When grpid is set, it takes the group ID of the directory in
  58         which it is created; otherwise (the default) it takes the fsgid
  59         of the current process, unless the directory has the setgid bit
  60         set, in which case it takes the gid from the parent directory,
  61         and also gets the setgid bit set if it is a directory itself.
  62
  63   ihashsize=value
  64         In memory inode hashes have been removed, so this option has
  65         no function as of August 2007. Option is deprecated.
  66
  67   ikeep/noikeep
  68         When ikeep is specified, XFS does not delete empty inode clusters
  69         and keeps them around on disk. ikeep is the traditional XFS
  70         behaviour. When noikeep is specified, empty inode clusters
  71         are returned to the free space pool. The default is noikeep for
  72         non-DMAPI mounts, while ikeep is the default when DMAPI is in use.
  73
  74   inode64
  75         Indicates that XFS is allowed to create inodes at any location
  76         in the filesystem, including those which will result in inode
  77         numbers occupying more than 32 bits of significance.  This is
  78         the default allocation option. Applications which do not handle
  79         inode numbers bigger than 32 bits, should use inode32 option.
  80
  81   inode32
  82         Indicates that XFS is limited to create inodes at locations which
  83         will not result in inode numbers with more than 32 bits of
  84         significance. This is provided for backwards compatibility, since
  85         64 bits inode numbers might cause problems for some applications
  86         that cannot handle large inode numbers.
  87
  88   largeio/nolargeio
  89         If "nolargeio" is specified, the optimal I/O reported in
  90         st_blksize by stat(2) will be as small as possible to allow user
  91         applications to avoid inefficient read/modify/write I/O.
  92         If "largeio" specified, a filesystem that has a "swidth" specified
  93         will return the "swidth" value (in bytes) in st_blksize. If the
  94         filesystem does not have a "swidth" specified but does specify
  95         an "allocsize" then "allocsize" (in bytes) will be returned
  96         instead.
  97         If neither of these two options are specified, then filesystem
  98         will behave as if "nolargeio" was specified.
  99
 100   logbufs=value
 101         Set the number of in-memory log buffers.  Valid numbers range
 102         from 2-8 inclusive.
 103         The default value is 8 buffers for filesystems with a
 104         blocksize of 64KiB, 4 buffers for filesystems with a blocksize
 105         of 32KiB, 3 buffers for filesystems with a blocksize of 16KiB
 106         and 2 buffers for all other configurations.  Increasing the
 107         number of buffers may increase performance on some workloads
 108         at the cost of the memory used for the additional log buffers
 109         and their associated control structures.
 110
 111   logbsize=value
 112         Set the size of each in-memory log buffer.
 113         Size may be specified in bytes, or in kilobytes with a "k" suffix.
 114         Valid sizes for version 1 and version 2 logs are 16384 (16k) and
 115         32768 (32k).  Valid sizes for version 2 logs also include
 116         65536 (64k), 131072 (128k) and 262144 (256k).
 117         The default value for machines with more than 32MiB of memory
 118         is 32768, machines with less memory use 16384 by default.
 119
 120   logdev=device and rtdev=device
 121         Use an external log (metadata journal) and/or real-time device.
 122         An XFS filesystem has up to three parts: a data section, a log
 123         section, and a real-time section.  The real-time section is
 124         optional, and the log section can be separate from the data
 125         section or contained within it.
 126
 127   mtpt=mountpoint
 128         Use with the "dmapi" option.  The value specified here will be
 129         included in the DMAPI mount event, and should be the path of
 130         the actual mountpoint that is used.
 131
 132   noalign
 133         Data allocations will not be aligned at stripe unit boundaries.
 134
 135   noatime
 136         Access timestamps are not updated when a file is read.
 137
 138   norecovery
 139         The filesystem will be mounted without running log recovery.
 140         If the filesystem was not cleanly unmounted, it is likely to
 141         be inconsistent when mounted in "norecovery" mode.
 142         Some files or directories may not be accessible because of this.
 143         Filesystems mounted "norecovery" must be mounted read-only or
 144         the mount will fail.
 145
 146   nouuid
 147         Don't check for double mounted file systems using the file system uuid.
 148         This is useful to mount LVM snapshot volumes.
 149
 150   uquota/usrquota/uqnoenforce/quota
 151         User disk quota accounting enabled, and limits (optionally)
 152         enforced.  Refer to xfs_quota(8) for further details.
 153
 154   gquota/grpquota/gqnoenforce
 155         Group disk quota accounting enabled and limits (optionally)
 156         enforced.  Refer to xfs_quota(8) for further details.
 157
 158   pquota/prjquota/pqnoenforce
 159         Project disk quota accounting enabled and limits (optionally)
 160         enforced.  Refer to xfs_quota(8) for further details.
 161
 162   sunit=value and swidth=value
 163         Used to specify the stripe unit and width for a RAID device or
 164         a stripe volume.  "value" must be specified in 512-byte block
 165         units.
 166         If this option is not specified and the filesystem was made on
 167         a stripe volume or the stripe width or unit were specified for
 168         the RAID device at mkfs time, then the mount system call will
 169         restore the value from the superblock.  For filesystems that
 170         are made directly on RAID devices, these options can be used
 171         to override the information in the superblock if the underlying
 172         disk layout changes after the filesystem has been created.
 173         The "swidth" option is required if the "sunit" option has been
 174         specified, and must be a multiple of the "sunit" value.
 175
 176   swalloc
 177         Data allocations will be rounded up to stripe width boundaries
 178         when the current end of file is being extended and the file
 179         size is larger than the stripe width size.
 180
 181
 182 sysctls
 183 =======
 184
 185 The following sysctls are available for the XFS filesystem:
 186
 187   fs.xfs.stats_clear            (Min: 0  Default: 0  Max: 1)
 188         Setting this to "1" clears accumulated XFS statistics
 189         in /proc/fs/xfs/stat.  It then immediately resets to "0".
 190
 191   fs.xfs.xfssyncd_centisecs     (Min: 100  Default: 3000  Max: 720000)
 192         The interval at which the xfssyncd thread flushes metadata
 193         out to disk.  This thread will flush log activity out, and
 194         do some processing on unlinked inodes.
 195
 196   fs.xfs.xfsbufd_centisecs      (Min: 50  Default: 100  Max: 3000)
 197         The interval at which xfsbufd scans the dirty metadata buffers list.
 198
 199   fs.xfs.age_buffer_centisecs   (Min: 100  Default: 1500  Max: 720000)
 200         The age at which xfsbufd flushes dirty metadata buffers to disk.
 201
 202   fs.xfs.error_level            (Min: 0  Default: 3  Max: 11)
 203         A volume knob for error reporting when internal errors occur.
 204         This will generate detailed messages & backtraces for filesystem
 205         shutdowns, for example.  Current threshold values are:
 206
 207                 XFS_ERRLEVEL_OFF:       0
 208                 XFS_ERRLEVEL_LOW:       1
 209                 XFS_ERRLEVEL_HIGH:      5
 210
 211   fs.xfs.panic_mask             (Min: 0  Default: 0  Max: 127)
 212         Causes certain error conditions to call BUG(). Value is a bitmask;
 213         AND together the tags which represent errors which should cause panics:
 214
 215                 XFS_NO_PTAG                     0
 216                 XFS_PTAG_IFLUSH                 0x00000001
 217                 XFS_PTAG_LOGRES                 0x00000002
 218                 XFS_PTAG_AILDELETE              0x00000004
 219                 XFS_PTAG_ERROR_REPORT           0x00000008
 220                 XFS_PTAG_SHUTDOWN_CORRUPT       0x00000010
 221                 XFS_PTAG_SHUTDOWN_IOERROR       0x00000020
 222                 XFS_PTAG_SHUTDOWN_LOGERROR      0x00000040
 223
 224         This option is intended for debugging only.
 225
 226   fs.xfs.irix_symlink_mode      (Min: 0  Default: 0  Max: 1)
 227         Controls whether symlinks are created with mode 0777 (default)
 228         or whether their mode is affected by the umask (irix mode).
 229
 230   fs.xfs.irix_sgid_inherit      (Min: 0  Default: 0  Max: 1)
 231         Controls files created in SGID directories.
 232         If the group ID of the new file does not match the effective group
 233         ID or one of the supplementary group IDs of the parent dir, the
 234         ISGID bit is cleared if the irix_sgid_inherit compatibility sysctl
 235         is set.
 236
 237   fs.xfs.inherit_sync           (Min: 0  Default: 1  Max: 1)
 238         Setting this to "1" will cause the "sync" flag set
 239         by the xfs_io(8) chattr command on a directory to be
 240         inherited by files in that directory.
 241
 242   fs.xfs.inherit_nodump         (Min: 0  Default: 1  Max: 1)
 243         Setting this to "1" will cause the "nodump" flag set
 244         by the xfs_io(8) chattr command on a directory to be
 245         inherited by files in that directory.
 246
 247   fs.xfs.inherit_noatime        (Min: 0  Default: 1  Max: 1)
 248         Setting this to "1" will cause the "noatime" flag set
 249         by the xfs_io(8) chattr command on a directory to be
 250         inherited by files in that directory.
 251
 252   fs.xfs.inherit_nosymlinks     (Min: 0  Default: 1  Max: 1)
 253         Setting this to "1" will cause the "nosymlinks" flag set
 254         by the xfs_io(8) chattr command on a directory to be
 255         inherited by files in that directory.
 256
 257   fs.xfs.rotorstep              (Min: 1  Default: 1  Max: 256)
 258         In "inode32" allocation mode, this option determines how many
 259         files the allocator attempts to allocate in the same allocation
 260         group before moving to the next allocation group.  The intent
 261         is to control the rate at which the allocator moves between
 262         allocation groups when allocating extents for new files.