5 - name: client_cache_size
8 desc: soft maximum number of directory entries in client cache
9 fmt_desc: Set the number of inodes that the client keeps in the metadata cache.
14 - name: client_cache_mid
17 desc: mid-point of client cache LRU
18 fmt_desc: Set client cache midpoint. The midpoint splits the least recently used
19 lists into a hot and warm list.
24 - name: client_use_random_mds
27 desc: issue new requests to a random active MDS
28 fmt_desc: Choose random MDS for each request.
33 - name: client_mount_timeout
36 desc: timeout for mounting CephFS (seconds)
37 fmt_desc: Set the timeout for CephFS mount in seconds.
41 - name: client_tick_interval
44 desc: seconds between client upkeep ticks
45 fmt_desc: Set the interval in seconds between capability renewal and other upkeep.
52 desc: file containing trace of client operations
56 - name: client_readahead_min
59 desc: minimum bytes to readahead in a file
60 fmt_desc: Set the minimum number bytes that the client reads ahead.
65 - name: client_readahead_max_bytes
68 desc: maximum bytes to readahead in a file (zero is unlimited)
69 fmt_desc: Set the maximum number of bytes that the client reads ahead for
70 future read operations. Overridden by the ``client_readahead_max_periods``
76 # as multiple of file layout period (object size * num stripes)
77 - name: client_readahead_max_periods
80 desc: maximum stripe periods to readahead in a file
81 fmt_desc: Set the number of file layout periods (object size * number of
82 stripes) that the client reads ahead. Overrides the
83 ``client_readahead_max_bytes`` setting.
88 - name: client_reconnect_stale
91 desc: reconnect when the session becomes stale
95 - name: client_snapdir
98 desc: pseudo directory for snapshot access to a directory
99 fmt_desc: Set the snapshot directory name.
104 - name: client_mountpoint
107 desc: default mount-point
108 fmt_desc: Directory to mount on the CephFS file system. An alternative to the
109 ``-r`` option of the ``ceph-fuse`` command.
113 - name: client_mount_uid
116 desc: uid to mount as
120 fmt_desc: Set the user ID of CephFS mount.
122 - name: client_mount_gid
125 desc: gid to mount as
126 fmt_desc: Set the group ID of CephFS mount.
131 - name: client_notify_timeout
138 - name: osd_client_watch_timeout
145 - name: client_caps_release_delay
151 fmt_desc: Set the delay between capability releases in seconds. The delay
152 sets how many seconds a client waits to release capabilities that it no
153 longer needs in case the capabilities are needed for another user space
155 - name: client_quota_df
158 desc: show quota usage for statfs (df)
159 fmt_desc: Report root directory quota for the ``statfs`` operation.
167 desc: enable object caching
172 - name: client_oc_size
175 desc: maximum size of object cache
176 fmt_desc: Set how many bytes of data will the client cache.
183 # MB * n (dirty OR tx.. bigish)
184 - name: client_oc_max_dirty
187 desc: maximum size of dirty pages in object cache
188 fmt_desc: Set the maximum number of dirty bytes in the object cache.
195 # target dirty (keep this smallish)
196 - name: client_oc_target_dirty
199 desc: target size of dirty pages object cache
200 fmt_desc: Set the target size of dirty data. We recommend to keep this number low.
207 - name: client_oc_max_dirty_age
210 desc: maximum age of dirty pages in object cache (seconds)
211 fmt_desc: Set the maximum age in seconds of dirty data in the object cache
219 - name: client_oc_max_objects
222 desc: maximum number of objects in cache
223 fmt_desc: Set the maximum number of objects in the object cache.
230 # check if MDS reply contains wanted caps
231 - name: client_debug_getattr_caps
238 # always read synchronously (go to osds)
239 - name: client_debug_force_sync_read
245 fmt_desc: If set to ``true``, clients read data directly from OSDs instead
246 of using a local page cache.
248 - name: client_debug_inject_tick_delay
254 - name: client_max_inline_size
260 fmt_desc: Set the maximum size of inlined data stored in a file inode rather
261 than in a separate data object in RADOS. This setting only applies if the
262 ``inline_data`` flag is set on the MDS map.
264 # synthetic client bug for testing
265 - name: client_inject_release_failure
272 # synthetic client bug for testing
273 - name: client_inject_fixed_oldest_tid
280 - name: client_metadata
283 desc: metadata key=value comma-delimited pairs appended to session metadata
284 fmt_desc: Comma-delimited strings for client metadata sent to each MDS, in addition
285 to the automatically generated version, host name, and other metadata.
289 - name: client_acl_type
292 desc: ACL type to enforce (none or "posix_acl")
293 fmt_desc: Set the ACL type. Currently, only possible value is ``"posix_acl"`` to
294 enable POSIX ACL, or an empty string. This option only takes effect when the
295 ``fuse_default_permissions`` is set to ``false``.
299 - name: client_permissions
302 desc: client-enforced permission checking
303 fmt_desc: Check client permissions on all I/O operations.
308 - name: client_dirsize_rbytes
311 desc: set the directory size as the number of file bytes recursively used
312 long_desc: This option enables a CephFS feature that stores the recursive directory
313 size (the bytes used by files in the directory and its descendents) in the st_size
314 field of the stat structure.
319 - name: client_force_lazyio
325 - name: fuse_use_invalidate_cb
328 desc: use fuse 2.8+ invalidate callback to keep page cache consistent
332 - name: fuse_disable_pagecache
335 desc: disable page caching in the kernel for this FUSE mount
336 fmt_desc: If set to ``true``, kernel page cache is disabled for ``ceph-fuse``
337 mounts. When multiple clients read/write to a file at the same
338 time, readers may get stale data from page cache. Due to
339 limitations of FUSE, ``ceph-fuse`` can't disable page cache dynamically.
343 - name: fuse_allow_other
346 desc: pass allow_other to FUSE on mount
350 - name: fuse_default_permissions
353 desc: pass default_permisions to FUSE on mount
354 fmt_desc: When set to ``false``, ``ceph-fuse`` utility checks does its own
355 permissions checking, instead of relying on the permissions enforcement in
356 FUSE. Set to ``false`` together with the ``client acl type=posix_acl``
357 option to enable POSIX ACL.
363 - name: fuse_splice_read
366 desc: enable splice read to reduce the memory copies
370 - name: fuse_splice_write
373 desc: enable splice write to reduce the memory copies
377 - name: fuse_splice_move
380 desc: enable splice move to reduce the memory copies
384 - name: fuse_big_writes
387 desc: big_writes is deprecated in libfuse 3.0.0
391 - name: fuse_max_write
394 desc: set the maximum number of bytes in a single write operation
395 long_desc: Set the maximum number of bytes in a single write operation that may
396 pass atomically through FUSE. The FUSE default is 128kB and may be indicated by
397 setting this option to 0.
398 fmt_desc: Set the maximum number of bytes in a single write operation. A value of
399 0 indicates no change; the FUSE default of 128 kbytes remains in force.
403 - name: fuse_atomic_o_trunc
406 desc: pass atomic_o_trunc flag to FUSE on mount
413 desc: enable debugging for the libfuse
420 - name: fuse_multithreaded
423 desc: allow parallel processing through FUSE library
427 - name: fuse_require_active_mds
430 desc: require active MDSs in the file system when mounting
434 - name: fuse_syncfs_on_mksnap
437 desc: synchronize all local metadata/file changes after snapshot
441 - name: fuse_set_user_groups
444 desc: check for ceph-fuse to consider supplementary groups for permissions
448 # the client should try to use dentry invaldation instead of remounting, on kernels it believes that will work for
449 - name: client_try_dentry_invalidate
456 - name: client_die_on_failed_remount
462 - name: client_die_on_failed_dentry_invalidate
465 desc: kill the client when no dentry invalidation options are available
466 long_desc: The CephFS client requires a mechanism to invalidate dentries in the
467 caller (e.g. the kernel for ceph-fuse) when capabilities must be recalled. If
468 the client cannot do this then the MDS cache cannot shrink which can cause the
473 - name: client_check_pool_perm
476 desc: confirm access to inode's data pool/namespace described in file layout
481 - name: client_use_faked_inos
494 desc: CephFS file system name to mount
495 long_desc: Use this with ceph-fuse, or with any process that uses libcephfs. Programs
496 using libcephfs may also pass the filesystem name into mount(), which will override
497 this setting. If no filesystem name is given in mount() or this setting, the default
498 filesystem will be mounted (usually the first created).
503 - name: client_mds_namespace
510 - name: fake_statfs_for_testing
513 desc: Set a value for kb and compute kb_used from total of num_bytes
519 - name: debug_allow_any_pool_priority
522 desc: Allow any pool priority to be set to test conversion to new range
527 - name: client_asio_thread_count
530 desc: Size of thread pool for ASIO completions
537 - name: client_shutdown_timeout
540 desc: timeout for shutting down CephFS
541 long_desc: Timeout for shutting down CephFS via unmount or shutdown.
550 - name: client_collect_and_send_global_metrics
553 desc: to enable and force collecting and sending the global metrics to MDS
554 long_desc: To be careful for this, when connecting to some old ceph clusters
555 it may crash the MDS daemons while upgrading.
566 desc: Enable quota enforcement
567 long_desc: Enable quota_bytes and quota_files enforcement for the client.