lfs-pool-unpin(1)

NAME

lfs-pool-unpin - unpin files, directories or FIDs from an OST pool

SYNOPSIS

DESCRIPTION

The lfs pool unpin subcommand removes the restriction on the named FILE(s), and/or files in the named DIRECTORY(/ies), so that there are no longer restrictions on migrating the file from the named POOL.

lfs pool unpin removes a pool entry for the selected POOL from the lustre.pin attribute. If no matching entry is present, the command succeeds without modifying the attribute. When the last entry is removed, the lustre.pin attribute itself may be deleted.

When the --pool option is omitted, POOL is inferred from the primary mirror layout of each file or directory. If the layout does not have an associated pool, the operation fails for that target and an error message is printed.

The --fid form of the command operates on files referenced by Lustre FIDs beneath the filesystem mounted at MOUNT, while the default and --file forms operate on filesystem paths.

Updating the lustre.pin attribute requires both appropriate write permissions on the target and sufficient privileges on the MDT. If these checks fail the operation will return EPERM, typically due to the mdt.enable_pin_gid policy; see mdt.enable_pin_gid(4) for details.

OPTIONS

--pool POOL

Name of the OST pool that should be unpinned from the specified files. The name must consist of only alphanumeric characters, underscores and hyphens, and may optionally be qualified with an fsname prefix (e.g. fsname.POOL), which is discarded before the pool name is used.

-f, --file

Interpret remaining arguments as filesystem paths (this is the default if neither --file nor --fid is given).

-F, --fid MOUNT

Interpret remaining arguments as Lustre FIDs within the filesystem mounted at MOUNT. This option is mutually exclusive with --file.

-q, --quiet

Suppress per-target error messages (such as a target that cannot be opened or an unrecognized FID); the exit status still reflects any errors encountered.

-h, --help

Display brief usage information for the lfs pool unpin subcommand.

EXIT STATUS

The lfs pool unpin command returns 0 if every target was pinned to the requested (or inferred) pool and is now unpinned, or 1 if one or more targets were not pinned there to begin with, could not be unpinned (for example due to insufficient privileges; see mdt.enable_pin_gid(4) for the MDT-side policy controlling which users may pin or unpin files), or otherwise did not end up unpinned; unpinning a target that had nothing to unpin is not treated as an error and does not modify lustre.pin. A non-zero exit status other than 1 is returned if any argument is invalid, a pool cannot be inferred when required, or a target could not be opened.

EXAMPLES

Unpin a directory tree and a single file from the pool scratch:

$ lfs pool unpin --pool scratch /mnt/testfs/proj /mnt/testfs/file

Unpin a file from the pool inferred from its primary mirror layout:

$ lfs pool unpin /mnt/testfs/file

Unpin a file referenced by FID from the pool scratch:

$ lfs pool unpin --pool scratch --fid /mnt/testfs 0x200000401:0x1:0x0

AVAILABILITY

The lfs pool unpin command is part of the lustre(7) filesystem package since release 2.18.0.

SEE ALSO

lfs(1), lfs-pool(1), lfs-pool-pin(1), lfs-pool-pin_info(1), llapi_pool_unpin_file(3), mdt.enable_pin_gid(4), lustre(7)