/etc/fstab Options

This topic was published by and viewed 2244 times since "". The last page revision was "".

Viewing 1 post (of 1 total)
  • Author
    Posts
  • DevynCJohnson
    DevynCJohnson
    Keymaster
    • Topics - 444
    • @devyncjohnson

    The filesystem table (/etc/fstab) has a column that contains various parameters that are read by the mounting utility. Many options are applicable to most or all filesystems. Then, there are some filesystem-specific options. Also, most of these options (or flags) are recognized by most Unixoid systems However, some of these mount options may not be available on some systems. For instance, the "newinstance" flag for mounting devpts is only available on Linux kernels that were compiled with the "CONFIG_DEVPTS_MULTIPLE_INSTANCES" option enabled. Knowing most of these options can be quite beneficial to admins and general users.
    NOTE: Unixoid systems are Unix, Unix-like, and POSIX systems.
    In addition to the filesystem table (/etc/fstab), there is also a "Mounted Table" (/etc/mtab) which uses the same format as the filesystem table. However, /etc/mtab lists currently mounted filesystems and their active settings and flags.
    Index

    General Options

    • _netdev - Network filesystem; do not mount until the network is active/available
    • _rnetdev - Same as "_netdev", but the filesystem is also checked during initialization
    • async - Asynchronous read and write
    • atime - Record last access time
    • auto - Mount the filesystem on bootup
    • ctime - Record last status change time
    • defaults - The typical setting for the filesystem being edited; typically (on Linux systems) rw, suid, dev, exec, auto, nouser, async, and relatime
    • dev - Recognize special block and character device files
    • diratime - Record last directory access time
    • dirsync - Synchronous directory updates; used for syscalls like creat, link, unlink, symlink, mkdir, rmdir, mknod and rename
    • exec - Allow executables to be executed
    • group - Allow a user in the same group as the device to mount the filesystem; implies nosuid and nodev
    • iversion - Increment "i_version" each time an inode is accessed
    • lazytime - Causes atime, mtime, and ctime updates to be made to the in-memory version of the inode only; fsync() and other conditions cause the times to be written on-disk
    • mand - Allow mandatory locks
    • mtime - Record last modified time
    • noatime - Do not record last access time
    • noauto - Do not mount the filesystem on bootup; only mount explicitly
    • nodev - Do not recognize special block device files
    • nodiratime - Do not record last directory access time
    • noexec - Do not permit executables to be executed
    • nofail - Do not report errors if the device is not found
    • noiversion - Do not increment "i_version"
    • nomand - Do not allow mandatory locks
    • norelatime - Do not use "relatime"
    • nostrictatime - Do not use "strictatime"
    • nosuid - Forbid the use of SUID and SGID (set-user-identifier and set-group-identifier bit)
    • nouser - Do not permit regular users to mount the filesystem
    • owner - Allow the specified owner to mount the filesystem (owner=bob); implies nosuid and nodev
    • relatime - The last access time is updated if it is older than the last modified time
    • remount - Mount an already mounted filesystem, but change the flags
    • ro - Read-only
    • rw - Read+Write
    • strictatime - Record last access time; allow the userspace to override "relatime" and "noatime"
    • suid - Allow the use of SUID and SGID (set-user-identifier and set-group-identifier bit)
    • sync - Synchronous read and write
    • user - Allow a regular user to mount the filesystem; implies noexec, nosuid, and nodev
    • users - All users can mount and unmount the filesystem

    Filesystem Specific Options

    ADFS

    • gid - Set group by GID; default "0"
    • othmask - Set "other" permissions; default "0077"
    • ownmask - Set "owner" permissions; default "0700"
    • uid - Set owner by UID; default "0"

    AFFS

    • bs - Blocksize (512, 1024, 2048, or 4096)
    • gid - Set group by GID; default "0"
    • grpquota - Allow group quotas; recognized by quota software
    • mode - Set the permissions of the files (i.e. "0777") and ignore previously set permissions
    • noquota - Disallow quotas; recognized by quota software
    • prefix - Prefix used in volume name
    • protect - Do not change the protection bits
    • quota - Allow quotas; recognized by quota software
    • reserved - The number of unused blocks at the beginning of the device (reserved=2)
    • root - Explicit location of the root-block
    • setgid - Set the group of the files (by GID)
    • setuid - Set the owner of the files (by UID)
    • uid - Set owner by UID; default "0"
    • usemp - Apply the UID and GID of the mountpoint to the filesystem
    • usrquota - Allow user quotas; recognized by quota software
    • verbose - Provide information about each mount
    • volume - Prefix used before "/" when following soft-links (symbolic links)

    CIFS

    • actimeo - Set the time (in seconds) the client will use its attribute cache before requesting updated info
    • backupgid - Only the specified group can access backup files
    • backupuid - Only the specified user can access backup files
    • cache - Set cache mode to none (no cache), loose (loose caching semantics), or strict (use the CIFS/SMB2 protocol)
    • cifsacl - Allow mapping between Linux permission bits and CIFS/NTFS ACLs, SID and UID&GID, and Security Descriptors
    • credentials - Location of the credentials file specifying the username, password, and domain; alternative to using "password" and "user"
    • cruid - Set the owner of the credentials cache
    • directio - Disallow inode caching; use cache=none instead
    • dir_mode - The directory-mode to use if Unix CIFS Extensions are not supported
    • domain - Set the workgroup of the user
    • dynperm - The server will store permissions in memory
    • file_mode - The file-mode to use if Unix CIFS Extensions are not supported
    • forcegid - Ignore the GID of the files specified by the server
    • forceuid - Ignore the UID of the files specified by the server
    • fsc - Use local-disk caching
    • gid - Set the GID of the files and directories if the server does not provide such info
    • guest - Passwords are not required to access the server share
    • hard - If the server crashes, programs accessing the filesystem will hang
    • ignorecase - Alias for "nocase"; case-insensitive paths
    • intr - UNUSED
    • iocharset - Set the I/O character set (iocharset=utf8)
    • ip - The IP address of the server containing the CIFS filesystem
    • mapchars - Re-map six of the seven reserved characters to the range above 0xF000
    • multiuser - Allow multiple credentials, one for each user
    • netbiosname - Set the client's NetBIOS name (when using port 139)
    • noacl - Disallow POSIX ACL operations
    • nobri - Do not send byte range lock requests to the server
    • nocase - Case-insensitive paths
    • nointr - UNUSED
    • nomapchars - Do not remap the seven reserved characters
    • noperm - Do not perform permission checks
    • noposixpaths - Disallow the use of POSIX-style pathnames
    • noserverino - Use client-generated inode numbers instead
    • nosetuids - Do not apply the UID and GID to files and directories
    • nounix - Disallow CIFS Unix Extensions
    • nouser_xattr - Disallow extended attributes
    • password - The password needed to access the share; if not set, then "$PASSWD" is used
    • perm - Perform permission checks
    • port - Use the specified port when connecting to the share
    • posixpaths - Allow the use of POSIX-style pathnames
    • prefixpath - Mount a subdirectory of a CIFS share
    • rsize - Network read-size; do not set larger than CIFSMaxBufSize (16kb)
    • rwpidforward - IO operations of a file inherit the PID of the process that opened the file
    • sec - Security Mode; many values are available
      • krb5 - Kerberos v5 authentication
      • krb5i - Kerberos v5 authentication with packet signing
      • none - connect as null user
      • ntlm - NTLM password hashing
      • ntlmi - NTLM password hashing with packet signing
      • ntlmssp - NTLMv2 password hashing encapsulated in a Raw NTLMSSP message
      • ntlmsspi - NTLMv2 password hashing encapsulated in a Raw NTLMSSP message with packet signing
      • ntlmv2 - NTLMv2 password hashing
      • ntlmv2i - NTLMv2 password hashing with packet signing
    • serverino - Use the server's inode numbers rather than temporary client inode numbers
    • servern - Alias for "servernetbiosname"
    • servernetbiosname - Set the server netbios name (RFC1001 name)
    • setuids - Try to apply the UID and GID to files and directories if Unix CIFS Extensions are not supported
    • sfu - create "Services for Unix (SFU)" compatible device files and fifos when the CIFS Unix Extensions are not negotiated
    • soft - If the server crashes, programs accessing the filesystem will return errors
    • strictcache - Use cache=strict instead
    • uid - Declare the owner (by UID) of the share
    • user (or username) - Connect as the specified user; if not set, then use "$USER"; alternate format "user%password"
    • wsize - Max size of negotiable write request (in bytes); do not set larger than 131007 bytes

    NOTE: "username" is typically used by smbfs while CIFS prefers "user".

    devpts

    • gid - Set group by GID; the default is the PID of the process that created it
    • newinstance - Create a private instance
    • ptmxmode - Set the mode for the new ptmx device
    • uid - Set owner by UID; the default is the PID of the process that created it

    EXT2

    • acl - Enable POSIX Access Control Lists
    • bsddf - The statfs syscall returns the total number of blocks on the filesystem minus the overhead blocks; conflicts with minixdf
    • bsdgroups - NO DOCUMENTATION FOUND
    • check - Set the fsck checking level (none, nocheck normal, or strict)
    • debug - Print debug info
    • errors - Set the desired behavior when an error occurs; errors=remount-ro
      • continue - ignore error and continue
      • panic - halt system
      • remount-ro - try to mount the filesystem as read-only
    • grpid - New files get the GID of the mountpoint's directory
    • grpquota - Allow group quotas; recognized by quota software
    • minixdf - The statfs syscall returns the total number of blocks on the filesystem; conflicts with bsddf
    • noacl - Disable POSIX Access Control Lists
    • nobh - Do not attach "buffer_heads" to the pagecache file
    • nogrpid - Do not set the GID of new files to the mountpoint's directory
    • noquota - Disallow quotas; recognized by quota software
    • nouid32 - Use 16-bit UIDs and GIDs instead of the 32-bit form
    • nouser_xattr - Disallow extended user attributes
    • oldalloc - Use the old allocator for new inodes
    • orlov - Use the Orlov allocator for new inodes
    • quota - Allow quotas; recognized by quota software
    • resgid - Reserve some blocks (5% by default) for the specified GID
    • resuid - Reserve some blocks (5% by default) for the specified UID
    • sb - Declare the location of the superblock by specifying a block (sb=5); default = 1
    • sysvgroups - NO DOCUMENTATION FOUND
    • user_xattr - Allow extended user attributes
    • usrquota - Allow user quotas; recognized by quota software

    EXT3

    NOTE: EXT3 supports all of the fstab flags that EXT2 uses

    • barrier - Enable (1) or disable (0) barriers; barriers enforce proper on-disk ordering of journal commits
    • commit - Synchronize data and metadata every X nanoseconds
    • data - Set journal mode
      • journal - data is written to the journal before being written to the filesystem
      • ordered - data is written to the filesystem before the journal
      • writeback - data may be written to either the journal or filesystem first
    • journal - Settings concerning the journal
      • inum - Place the journal on the specified inode (journal=100)
      • update - Use the latest journal format (journal=update)
    • journal_dev - If the journal is stored on an external device, this option can change/update the devnum major and minor numbers
    • noload - Disallow journal loading when mounting
    • norecovery - Do not permit the journal to recover data if an error occurred

    EXT4

    NOTE: EXT4 supports all of the fstab flags that EXT2 and EXT3 use

    • abort - simulate ext4_abort(); used for debugging purposes
    • auto_da_alloc - Detect the replace-via-rename and replace-via-truncate patterns and write the new files data to the filesystem before the rename() operation is committed
    • block_validity - Track metadata within internal data structures; used for debugging
    • delalloc - Defer block allocation until write-out time
    • dioread_lock - Enable DIO read locking
    • dioread_nolock - Disable DIO read locking
    • discard - Allow discard (TRIM) commands when blocks are freed
    • inode_readahead_blks - The max amount of inode table blocks that can be pre-read to the buffer cache; must be a power of 2 (i.e. 8, 16, 32, 64, etc.)
    • i_version - Enable 64-bit inodes
    • journal_async_commit - The commit-block can be written without waiting for the descriptor-blocks
    • journal_checksum - Journal transactions have a checksum
    • journal_ioprio - kjournald2's I/O priority during a commit operation; 0-7; 0 is highest priority
    • max_batch_time - Max time to wait for more filesystem operations before beginning synchronous write operations
    • min_batch_time - Minimum time to wait for more filesystem operations before beginning synchronous write operations
    • noauto_da_alloc - Disable "auto_da_alloc"
    • nobarrier - Disable barriers
    • noblock_validity - Disallow metadata tracking within internal data structures
    • nodelalloc - Disable delayed allocation; Block allocation occurs when data is copied from user-cache to pagecache
    • nodiscard - Disallow discard (TRIM) commands
    • resize - Permit resizing to additional blocks; takes effect during "remount"
    • stripe - The number of blocks used by mballoc allocation size and alignment

    FAT

    • allow_utime - If set to "2", other users can change the access and modification times
    • blocksize - The number of bytes in a block (512, 1024, 2048)
    • check - Check filenames to ensure that the correct naming schemes are used; choices include "r" (relaxed), "n" (normal), and "s" (strict)
    • codepage - Sets the codepage for converting to shortname characters; default is "437"
    • conv - Perform DOS and Unix text-file conversions; choices include "b" (binary), "t" (text), and "a" (auto)
    • debug - Allow debugging features
    • dmask - Set the user directory creation mode mask
    • dots - NO DOCUMENTATION
    • fat - Declare the version of FAT (12, 16, 32)
    • flush - Flush the disk earlier than usual
    • fmask - Set the user file creation mode mask
    • gid - Set the GID given to the files
    • iocharset - The character-set to use when converting between 8-bit characters and 16-bit Unicode; default is "iso8859-1"
    • nodots - NO DOCUMENTATION
    • quiet - "chown" and "chmod" commands will fail, but not produce errors
    • showexec - The executable bit can be set on *.bat, *.com, and *.exe files
    • sys_immutable - The ATTR_SYS attribute will be treated as an IMMUTABLE flag
    • tz - Set the time-zone; tz=UTC
    • uid - Set the UID given to the files
    • umask - Set the user file/directory creation mode mask
    • usefree - Use the "free clusters" value stored on FSINFO
    • windows_names - Only permit characters accepted by Windows to be used in directory and file names

    HFS

    • conv - Perform text-file conversions; choices include "b" (binary), "t" (text), and "a" (auto)
    • creator - Set the creator value used by MacOS Finder
    • dir_umask - Set the user directory creation mode mask
    • file_umask - Set the user file creation mode mask
    • gid - Set the GID given to the files
    • part - Partition number
    • quiet - Do not report invalid mounting options
    • session - Select session to mount
    • type - Set the type value used by MacOS Finder
    • uid - Set the UID given to the files
    • umask - Set the user file/directory creation mode mask

    HPFS

    • conv - Convert filenames to lowercase (lower) or leave them unchanged (asis)
    • gid - Set the GID given to the files
    • nocheck - Do not perform consistency checks
    • uid - Set the UID given to the files
    • umask - Set the user file/directory creation mode mask

    ISO9660

    • block - Set blocksize (512, 1024, 2048)
    • check - Choices include relaxed (r - convert filenames to lowercase) and strict (s - default behavior)
    • cruft - Ignore the high-byte of the file-length if it contains useless data; files cannot be larger than 16MB
    • gid - Set the GID given to the files
    • iocharset - Character set to use when converting 16-bit Unicode characters to 8-bit; default is "iso8859-1"
    • map - Options include normal (n), off (o), and acorn (a)
      • normal - convert uppercase to lowercase, remove trailing ";1", and convert ";" to "."
      • off - no filename translation
      • acorn - same a "normal" but also adds Acorn extensions
    • mode - Set the file permissions for all files; mode=0444
    • nojoliet - Disable Microsoft Joliet extensions
    • norock - Disable Rock Ridge extensions
    • sbsector - Select the sector where the session begins
    • session - Select the session of a multi-session disc
    • uid - Set the UID given to the files
    • unhide - Show hidden files
    • utf8 - Convert 16-bit Unicode to UTF-8

    JFS

    • errors - Set the desired behavior when an error occurs; errors=remount-ro
      • continue - ignore error and continue
      • panic - halt system
      • remount-ro - try to mount the filesystem as read-only
    • grpquota - Allow group quotas; recognized by quota software
    • integrity - Commit metadata to the journal
    • iocharset - Character set to use when converting Unicode characters to ASCII
    • nointegrity - Do not write to the journal
    • noquota - Disallow quotas; recognized by quota software
    • quota - Allow quotas; recognized by quota software
    • resize - Number of blocks of the new volume size; takes effect during "remount"
    • usrquota - Allow user quotas; recognized by quota software

    NFS

    • addr - Set the IP address

    NTFS

    • dmask - Set the user directory creation mode mask
    • fmask - Set the user file creation mode mask
    • gid - Set the GID given to the files
    • uid - Set the UID given to the files
    • umask - Set the user file/directory creation mode mask
    • uni_xlate - Special characters
      • 0 - do not escape unknown characters
      • 1 - use VFAT-style 4-byte-escapes starting with ":"; byte-swapped big-endian encoding
      • 2 - use VFAT-style 4-byte-escapes starting with ":"; little-endian encoding
    • utf8 - Convert filenames to UTF8
    • windows_names - Only permit characters accepted by Windows to be used in directory and file names

    PROC

    • gid - Set group by GID; default "0"
    • uid - Set owner by UID; default "0"

    ReiserFS

    • acl - Enable POSIX Access Control Lists
    • barrier - Disable (none) or enable (flush) barriers; enforces proper on-disk ordering of journal commits
    • conv - Use the v3.6 software and format on ReiserFS v3.5
    • hash - Select the hash used to find files in directories
      • detect - auto-detect the hash used
      • r5 - modified rupasov hash
      • rupasov - preserves locality and maps lexicographically similar filenames to similar hashes
      • tea - random hashes
    • hashed_relocation - Tunes the block allocator
    • noborder - Disable the border allocator algorithm
    • nolog - Disable journalling
    • notail - Disable tail-packing
    • no_unhashed_relocation - NO DOCUMENTATION
    • replayonly - Replay the transactions in the journal
    • resize - Expand the filesystem to the specified number of blocks
    • user_xattr - Enable extended user attributes

    tmpfs

    • gid - Set group by GID; default "0"
    • mode - File permissions of the root directory
    • mpol - Set the NUMA memory allocation policy
      • bind:NODE_LIST - allocate memory from nodes in NODE_LIST; mpol=bind:0-2,5,7,11-13
      • default - allocate memory from the local node
      • interleave - allocate from each node in turn
      • interleave:NODE_LIST - allocate from each node of NODE_LIST in turn
      • prefer:NODE - allocate memory from the given NODE; mpol=prefer:2
    • nr_blocks - Override filesystem size; set size by blocks of PAGE_CACHE_SIZE
    • nr_inodes - Overrides the filesystem size by the amount of inodes
    • size - Override the default filesystem size in bytes or percentage of RAM; default = size=50%
    • uid - Set owner by UID; default "0"

    UBIFS

    • bulk_read - Allow bulk-reading; VFS read-ahead is disabled
    • chk_data_crc - Check CRC-32 checksums
    • compr - Select the desired compression to use for new files (none, lzo, zlib); this does not change pre-existing compressions
    • no_bulk_read - Do not bulk-read
    • no_chk_data_crc - Disable checksum checking

    UDF

    • anchor - Override standard anchor location; default = anchor=256
    • bs - Set blocksize
    • gid - Set group by GID; default "0"
    • iocharset - Set the character set
    • lastblock - Set the last block of the filesystem
    • nostrict - Disable strict conformance
    • novrs - Skip volume sequence recognition
    • session - Set the CDROM session counting
    • uid - Set owner by UID; default "0"
    • umask - Set the default umask (file permissions)
    • undelete - Show deleted files
    • unhide - Show hidden files

    UFS

    • onerror - Set error behavior (panic)
    • ufstype - Specify UFS implementation
      • 44bsd - UFS created by BSD systems
      • hp - UFS created by HP-UX; read-only
      • nextstep - UFS created by NeXTStep; read-only
      • nextstep-cd - NextStep CDROMs; read-only; block-size=2048
      • old - Old UFS; read-only
      • openstep - UFS created by OpenStep and OSX
      • sun - UFS created by SunOS or Solaris on Sparc
      • sunx86 - UFS created by Solaris on x86

    VFAT

    • shortname - Control how long and short (less than 8.3 characters; i.e. three-character extensions) filenames are displayed and created
      • lower - short filenames use lower-case letters; store the long filename
      • mixed - do not change short filenames; store long filenames when the short filename is not all uppercase
      • win95 - short filenames use upper-case letters; store the long filename
      • winnt - do not change short filenames; store long filenames
    • nonumtail - Try to use short filenames before using number sequences (i.e. "FILE~1.txt")
    • uni_xlate - Translate Unicode characters to special escaped sequences
    • utf8 - Use UTF-8 for filenames

    USBFS

    • busgid - Set the group of the bus directories by GID
    • busmode - Set the permission mode of the bus directories
    • busuid - Set the owner of the bus directories by UID
    • devgid - Set the group of the device files by GID
    • devmode - Set the permission mode of the device files
    • devuid - Set the owner of the device files by UID
    • listgid - Set group by GID
    • listmode - Set permission mode
    • listuid - Set owner by UID

    XFS

    • allocsize - Sets the buffered I/O end-of-file preallocation size; 4KiB-1GiB by power-of-2
    • attr2 - Enable improvements for in-line extended attributes
    • barrier - Enable block-layer write-barriers
    • bsdgroups - NO DOCUMENTATION
    • dmapi - Enable Data Management API event callouts
    • gqnoenforce - Disable group quotas
    • gquota - Enable group quotas
    • grpid - NO DOCUMENTATION
    • grpquota - Enable group quotas
    • ihashsize - Number of hash-buckets used to hash the in-memory inodes of the mount-point
    • ikeep - Emptied inode clusters are not released to the free space pool
    • inode64 - XFS is permitted to store inodes anywhere on the filesystem even beyond 32-bit inode numbers
    • largeio - "swidth" or "allocsize" will be used as the blocksize
    • logbsize - The size of each in-memory log buffer (in bytes)
    • logbufs - Set the number of in-memory log buffers; valid values are 2-8, inclusive
    • logdev - Specify the device file containing the external metadata journal
    • mtpt - The mount-point included in DMAPI mount events; requires the dmapi option
    • noalign - Data allocations will not be aligned to stripe unit boundaries
    • noatime - Do not update file access timestamps
    • noattr2 - Disable improvements for in-line extended attributes
    • nogrpid - NO DOCUMENTATION
    • noikeep - Return emptied inode clusters to the free space pool
    • nolargeio - The smallest optimal I/O blocksize will be used
    • norecovery - Mount without running log recovery
    • pqnoenforce - Disable project quotas
    • pquota - Enable project quotas
    • prjquota - Enable project quotas
    • quota - Enable user quotas
    • rtdev - Specify the device file containing the real-time
    • sunit - Stripe unit for RAID or stripe volumes; specified in 512-byte block units
    • swalloc - Data allocations will rounded-up to stripe width boundaries when the current end of file is being extended and the file size is larger than the stripe width size
    • swidth - Stripe width for RAID or stripe volumes; specified in 512-byte block units
    • sysvgroups - NO DOCUMENTATION
    • uqnoenforce - Disable user quotas
    • uquota - Enable user quotas
    • usrquota - Enable user quotas

    Further Reading

Viewing 1 post (of 1 total)