5 ``mds_cache_memory_limit``
7 :Description: The memory limit the MDS should enforce for its cache.
8 :Type: 64-bit Integer Unsigned
11 ``mds_cache_reservation``
13 :Description: The cache reservation (memory or inodes) for the MDS cache to maintain.
14 Once the MDS begins dipping into its reservation, it will recall
15 client state until its cache size shrinks to restore the
23 :Description: The insertion point for new items in the cache LRU
30 ``mds_dir_commit_ratio``
32 :Description: The fraction of directory that is dirty before Ceph commits using
33 a full update (instead of partial update).
39 ``mds_dir_max_commit_size``
41 :Description: The maximum size of a directory update before Ceph breaks it into
42 smaller transactions (MB).
48 ``mds_decay_halflife``
50 :Description: The half-life of MDS cache temperature.
54 ``mds_beacon_interval``
56 :Description: The frequency (in seconds) of beacon messages sent
65 :Description: The interval without beacons before Ceph declares an MDS laggy
66 (and possibly replace it).
72 ``mds_blocklist_interval``
74 :Description: The blocklist duration for failed MDSs in the OSD map. Note,
75 this controls how long failed MDS daemons will stay in the
76 OSDMap blocklist. It has no effect on how long something is
77 blocklisted when the administrator blocklists it manually. For
78 example, ``ceph osd blocklist add`` will still use the default
81 :Default: ``24.0*60.0``
84 ``mds_reconnect_timeout``
86 :Description: The interval (in seconds) to wait for clients to reconnect
95 :Description: How frequently the MDS performs internal periodic tasks.
100 ``mds_dirstat_min_interval``
102 :Description: The minimum interval (in seconds) to try to avoid propagating
103 recursive stats up the tree.
108 ``mds_scatter_nudge_interval``
110 :Description: How quickly dirstat changes propagate up.
115 ``mds_client_prealloc_inos``
117 :Description: The number of inode numbers to preallocate per client session.
118 :Type: 32-bit Integer
124 :Description: Determines whether the MDS should allow clients to see request
125 results before they commit to the journal.
131 ``mds_default_dir_hash``
133 :Description: The function to use for hashing files across directory fragments.
134 :Type: 32-bit Integer
135 :Default: ``2`` (i.e., rjenkins)
138 ``mds_log_skip_corrupt_events``
140 :Description: Determines whether the MDS should try to skip corrupt journal
141 events during journal replay.
147 ``mds_log_max_events``
149 :Description: The maximum events in the journal before we initiate trimming.
150 Set to ``-1`` to disable limits.
152 :Type: 32-bit Integer
156 ``mds_log_max_segments``
158 :Description: The maximum number of segments (objects) in the journal before
159 we initiate trimming. Set to ``-1`` to disable limits.
161 :Type: 32-bit Integer
165 ``mds_bal_sample_interval``
167 :Description: Determines how frequently to sample directory temperature
168 (for fragmentation decisions).
174 ``mds_bal_replicate_threshold``
176 :Description: The maximum temperature before Ceph attempts to replicate
177 metadata to other nodes.
183 ``mds_bal_unreplicate_threshold``
185 :Description: The minimum temperature before Ceph stops replicating
186 metadata to other nodes.
192 ``mds_bal_split_size``
194 :Description: The maximum directory size before the MDS will split a directory
195 fragment into smaller bits.
197 :Type: 32-bit Integer
203 :Description: The maximum directory read temperature before Ceph splits
204 a directory fragment.
212 :Description: The maximum directory write temperature before Ceph splits
213 a directory fragment.
219 ``mds_bal_split_bits``
221 :Description: The number of bits by which to split a directory fragment.
222 :Type: 32-bit Integer
226 ``mds_bal_merge_size``
228 :Description: The minimum directory size before Ceph tries to merge
229 adjacent directory fragments.
231 :Type: 32-bit Integer
237 :Description: The frequency (in seconds) of workload exchanges between MDSs.
238 :Type: 32-bit Integer
242 ``mds_bal_fragment_interval``
244 :Description: The delay (in seconds) between a fragment being eligible for split
245 or merge and executing the fragmentation change.
246 :Type: 32-bit Integer
250 ``mds_bal_fragment_fast_factor``
252 :Description: The ratio by which frags may exceed the split size before
253 a split is executed immediately (skipping the fragment interval)
257 ``mds_bal_fragment_size_max``
259 :Description: The maximum size of a fragment before any new entries
260 are rejected with ENOSPC.
261 :Type: 32-bit Integer
264 ``mds_bal_idle_threshold``
266 :Description: The minimum temperature before Ceph migrates a subtree
275 :Description: The number of iterations to run balancer before Ceph stops.
276 (used for testing purposes only)
278 :Type: 32-bit Integer
282 ``mds_bal_max_until``
284 :Description: The number of seconds to run balancer before Ceph stops.
285 (used for testing purposes only)
287 :Type: 32-bit Integer
293 :Description: The method for calculating MDS load.
296 - ``1`` = Request rate and latency.
299 :Type: 32-bit Integer
303 ``mds_bal_min_rebalance``
305 :Description: The minimum subtree temperature before Ceph migrates.
310 ``mds_bal_min_start``
312 :Description: The minimum subtree temperature before Ceph searches a subtree.
319 :Description: The minimum fraction of target subtree size to accept.
326 :Description: The maximum fraction of target subtree size to accept.
333 :Description: Ceph will migrate any subtree that is larger than this fraction
334 of the target subtree size.
342 :Description: Ceph will ignore any subtree that is smaller than this fraction
343 of the target subtree size.
349 ``mds_bal_target_removal_min``
351 :Description: The minimum number of balancer iterations before Ceph removes
352 an old MDS target from the MDS map.
354 :Type: 32-bit Integer
358 ``mds_bal_target_removal_max``
360 :Description: The maximum number of balancer iterations before Ceph removes
361 an old MDS target from the MDS map.
363 :Type: 32-bit Integer
367 ``mds_replay_interval``
369 :Description: The journal poll interval when in standby-replay mode.
376 ``mds_shutdown_check``
378 :Description: The interval for polling the cache during MDS shutdown.
379 :Type: 32-bit Integer
383 ``mds_thrash_exports``
385 :Description: Ceph will randomly export subtrees between nodes (testing only).
386 :Type: 32-bit Integer
390 ``mds_thrash_fragments``
392 :Description: Ceph will randomly fragment or merge directories.
393 :Type: 32-bit Integer
397 ``mds_dump_cache_on_map``
399 :Description: Ceph will dump the MDS cache contents to a file on each MDSMap.
404 ``mds_dump_cache_after_rejoin``
406 :Description: Ceph will dump MDS cache contents to a file after
407 rejoining the cache (during recovery).
413 ``mds_verify_scatter``
415 :Description: Ceph will assert that various scatter/gather invariants
416 are ``true`` (developers only).
422 ``mds_debug_scatterstat``
424 :Description: Ceph will assert that various recursive stat invariants
425 are ``true`` (for developers only).
433 :Description: Ceph will verify directory fragmentation invariants
434 when convenient (developers only).
440 ``mds_debug_auth_pins``
442 :Description: The debug auth pin invariants (for developers only).
447 ``mds_debug_subtrees``
449 :Description: The debug subtree invariants (for developers only).
454 ``mds_kill_mdstable_at``
456 :Description: Ceph will inject MDS failure in MDSTable code
457 (for developers only).
459 :Type: 32-bit Integer
463 ``mds_kill_export_at``
465 :Description: Ceph will inject MDS failure in the subtree export code
466 (for developers only).
468 :Type: 32-bit Integer
472 ``mds_kill_import_at``
474 :Description: Ceph will inject MDS failure in the subtree import code
475 (for developers only).
477 :Type: 32-bit Integer
483 :Description: Ceph will inject MDS failure in hard link code
484 (for developers only).
486 :Type: 32-bit Integer
490 ``mds_kill_rename_at``
492 :Description: Ceph will inject MDS failure in the rename code
493 (for developers only).
495 :Type: 32-bit Integer
499 ``mds_wipe_sessions``
501 :Description: Ceph will delete all client sessions on startup
508 ``mds_wipe_ino_prealloc``
510 :Description: Ceph will delete ino preallocation metadata on startup
519 :Description: The number of inode numbers to skip on startup
522 :Type: 32-bit Integer
526 ``mds_min_caps_per_client``
528 :Description: Set the minimum number of capabilities a client may hold.
533 ``mds_max_ratio_caps_per_client``
535 :Description: Set the maximum ratio of current caps that may be recalled during MDS cache pressure.