2 .\" Copyright (c) 2013 by Turbo Fredriksson <turbo@bayour.com>. All rights reserved.
3 .\" The contents of this file are subject to the terms of the Common Development
4 .\" and Distribution License (the "License"). You may not use this file except
5 .\" in compliance with the License. You can obtain a copy of the license at
6 .\" usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing.
8 .\" See the License for the specific language governing permissions and
9 .\" limitations under the License. When distributing Covered Code, include this
10 .\" CDDL HEADER in each file and include the License file at
11 .\" usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this
12 .\" CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your
13 .\" own identifying information:
14 .\" Portions Copyright [yyyy] [name of copyright owner]
15 .TH ZFS-MODULE-PARAMETERS 5 "Nov 16, 2013"
17 zfs\-module\-parameters \- ZFS module parameters
21 Description of the different parameters to the ZFS module.
23 .SS "Module parameters"
30 \fBl2arc_feed_again\fR (int)
33 Turbo L2ARC warm-up. When the L2ARC is cold the fill interval will be set as
36 Use \fB1\fR for yes (default) and \fB0\fR to disable.
42 \fBl2arc_feed_min_ms\fR (ulong)
45 Min feed interval in milliseconds. Requires \fBl2arc_feed_again=1\fR and only
46 applicable in related situations.
48 Default value: \fB200\fR.
54 \fBl2arc_feed_secs\fR (ulong)
57 Seconds between L2ARC writing
59 Default value: \fB1\fR.
65 \fBl2arc_headroom\fR (ulong)
68 How far through the ARC lists to search for L2ARC cacheable content, expressed
69 as a multiplier of \fBl2arc_write_max\fR
71 Default value: \fB2\fR.
77 \fBl2arc_headroom_boost\fR (ulong)
80 Scales \fBl2arc_headroom\fR by this percentage when L2ARC contents are being
81 successfully compressed before writing. A value of 100 disables this feature.
83 Default value: \fB200\fR.
89 \fBl2arc_max_block_size\fR (ulong)
92 The maximum block size which may be written to an L2ARC device, after
93 compression and other factors. This setting is used to prevent a small
94 number of large blocks from pushing a larger number of small blocks out
97 Default value: \fB16,777,216\fR.
103 \fBl2arc_nocompress\fR (int)
106 Skip compressing L2ARC buffers
108 Use \fB1\fR for yes and \fB0\fR for no (default).
114 \fBl2arc_noprefetch\fR (int)
117 Do not write buffers to L2ARC if they were prefetched but not used by
120 Use \fB1\fR for yes (default) and \fB0\fR to disable.
126 \fBl2arc_norw\fR (int)
129 No reads during writes
131 Use \fB1\fR for yes and \fB0\fR for no (default).
137 \fBl2arc_write_boost\fR (ulong)
140 Cold L2ARC devices will have \fBl2arc_write_nax\fR increased by this amount
141 while they remain cold.
143 Default value: \fB8,388,608\fR.
149 \fBl2arc_write_max\fR (ulong)
152 Max write bytes per interval
154 Default value: \fB8,388,608\fR.
160 \fBmetaslab_aliquot\fR (ulong)
163 Metaslab granularity, in bytes. This is roughly similar to what would be
164 referred to as the "stripe size" in traditional RAID arrays. In normal
165 operation, ZFS will try to write this amount of data to a top-level vdev
166 before moving on to the next one.
168 Default value: \fB524,288\fR.
174 \fBmetaslab_bias_enabled\fR (int)
177 Enable metaslab group biasing based on its vdev's over- or under-utilization
178 relative to the pool.
180 Use \fB1\fR for yes (default) and \fB0\fR for no.
186 \fBmetaslab_debug_load\fR (int)
189 Load all metaslabs during pool import.
191 Use \fB1\fR for yes and \fB0\fR for no (default).
197 \fBmetaslab_debug_unload\fR (int)
200 Prevent metaslabs from being unloaded.
202 Use \fB1\fR for yes and \fB0\fR for no (default).
208 \fBmetaslab_fragmentation_factor_enabled\fR (int)
211 Enable use of the fragmentation metric in computing metaslab weights.
213 Use \fB1\fR for yes (default) and \fB0\fR for no.
219 \fBmetaslabs_per_vdev\fR (int)
222 When a vdev is added, it will be divided into approximately (but no more than) this number of metaslabs.
224 Default value: \fB200\fR.
230 \fBmetaslab_preload_enabled\fR (int)
233 Enable metaslab group preloading.
235 Use \fB1\fR for yes (default) and \fB0\fR for no.
241 \fBmetaslab_lba_weighting_enabled\fR (int)
244 Give more weight to metaslabs with lower LBAs, assuming they have
245 greater bandwidth as is typically the case on a modern constant
246 angular velocity disk drive.
248 Use \fB1\fR for yes (default) and \fB0\fR for no.
254 \fBspa_config_path\fR (charp)
259 Default value: \fB/etc/zfs/zpool.cache\fR.
265 \fBspa_asize_inflation\fR (int)
268 Multiplication factor used to estimate actual disk consumption from the
269 size of data being written. The default value is a worst case estimate,
270 but lower values may be valid for a given pool depending on its
271 configuration. Pool administrators who understand the factors involved
272 may wish to specify a more realistic inflation factor, particularly if
273 they operate close to quota or capacity limits.
275 Default value: \fB24\fR.
281 \fBspa_load_verify_data\fR (int)
284 Whether to traverse data blocks during an "extreme rewind" (\fB-X\fR)
285 import. Use 0 to disable and 1 to enable.
287 An extreme rewind import normally performs a full traversal of all
288 blocks in the pool for verification. If this parameter is set to 0,
289 the traversal skips non-metadata blocks. It can be toggled once the
290 import has started to stop or start the traversal of non-metadata blocks.
292 Default value: \fB1\fR.
298 \fBspa_load_verify_metadata\fR (int)
301 Whether to traverse blocks during an "extreme rewind" (\fB-X\fR)
302 pool import. Use 0 to disable and 1 to enable.
304 An extreme rewind import normally performs a full traversal of all
305 blocks in the pool for verification. If this parameter is set to 0,
306 the traversal is not performed. It can be toggled once the import has
307 started to stop or start the traversal.
309 Default value: \fB1\fR.
315 \fBspa_load_verify_maxinflight\fR (int)
318 Maximum concurrent I/Os during the traversal performed during an "extreme
319 rewind" (\fB-X\fR) pool import.
321 Default value: \fB10000\fR.
327 \fBspa_slop_shift\fR (int)
330 Normally, we don't allow the last 3.2% (1/(2^spa_slop_shift)) of space
331 in the pool to be consumed. This ensures that we don't run the pool
332 completely out of space, due to unaccounted changes (e.g. to the MOS).
333 It also limits the worst-case time to allocate space. If we have
334 less than this amount of free space, most ZPL operations (e.g. write,
335 create) will return ENOSPC.
337 Default value: \fB5\fR.
343 \fBzfetch_array_rd_sz\fR (ulong)
346 If prefetching is enabled, disable prefetching for reads larger than this size.
348 Default value: \fB1,048,576\fR.
354 \fBzfetch_max_distance\fR (uint)
357 Max bytes to prefetch per stream (default 8MB).
359 Default value: \fB8,388,608\fR.
365 \fBzfetch_max_streams\fR (uint)
368 Max number of streams per zfetch (prefetch streams per file).
370 Default value: \fB8\fR.
376 \fBzfetch_min_sec_reap\fR (uint)
379 Min time before an active prefetch stream can be reclaimed
381 Default value: \fB2\fR.
387 \fBzfs_arc_dnode_limit\fR (ulong)
390 When the number of bytes consumed by dnodes in the ARC exceeds this number of
391 bytes, try to unpin some of it in response to demand for non-metadata. This
392 value acts as a floor to the amount of dnode metadata.
394 See also \fBzfs_arc_meta_prune\fR which serves a similar purpose but is used
395 when the amount of metadata in the ARC exceeds \fBzfs_arc_meta_limit\fR rather
396 than in response to overall demand for non-metadata.
399 Default value: \fB10% of zfs_arc_meta_limit\fR.
405 \fBzfs_arc_dnode_reduce_percent\fR (ulong)
408 Percentage of ARC dnodes to try to scan in response to demand for non-metadata
409 when the number of bytes consumed by dnodes exceeds \fBzfs_arc_dnode_limit\fB.
412 Default value: \fB10% of the number of dnodes in the ARC\fR.
418 \fBzfs_arc_average_blocksize\fR (int)
421 The ARC's buffer hash table is sized based on the assumption of an average
422 block size of \fBzfs_arc_average_blocksize\fR (default 8K). This works out
423 to roughly 1MB of hash table per 1GB of physical memory with 8-byte pointers.
424 For configurations with a known larger average block size this value can be
425 increased to reduce the memory footprint.
428 Default value: \fB8192\fR.
434 \fBzfs_arc_evict_batch_limit\fR (int)
437 Number ARC headers to evict per sub-list before proceeding to another sub-list.
438 This batch-style operation prevents entire sub-lists from being evicted at once
439 but comes at a cost of additional unlocking and locking.
441 Default value: \fB10\fR.
447 \fBzfs_arc_grow_retry\fR (int)
450 After a memory pressure event the ARC will wait this many seconds before trying
453 Default value: \fB5\fR.
459 \fBzfs_arc_lotsfree_percent\fR (int)
462 Throttle I/O when free system memory drops below this percentage of total
463 system memory. Setting this value to 0 will disable the throttle.
465 Default value: \fB10\fR.
471 \fBzfs_arc_max\fR (ulong)
474 Max arc size of ARC in bytes. If set to 0 then it will consume 1/2 of system
475 RAM. This value must be at least 67108864 (64 megabytes).
477 This value can be changed dynamically with some caveats. It cannot be set back
478 to 0 while running and reducing it below the current ARC size will not cause
479 the ARC to shrink without memory pressure to induce shrinking.
481 Default value: \fB0\fR.
487 \fBzfs_arc_meta_limit\fR (ulong)
490 The maximum allowed size in bytes that meta data buffers are allowed to
491 consume in the ARC. When this limit is reached meta data buffers will
492 be reclaimed even if the overall arc_c_max has not been reached. This
493 value defaults to 0 which indicates that 3/4 of the ARC may be used
496 This value my be changed dynamically except that it cannot be set back to 0
497 for 3/4 of the ARC; it must be set to an explicit value.
499 Default value: \fB0\fR.
505 \fBzfs_arc_meta_min\fR (ulong)
508 The minimum allowed size in bytes that meta data buffers may consume in
509 the ARC. This value defaults to 0 which disables a floor on the amount
510 of the ARC devoted meta data.
512 Default value: \fB0\fR.
518 \fBzfs_arc_meta_prune\fR (int)
521 The number of dentries and inodes to be scanned looking for entries
522 which can be dropped. This may be required when the ARC reaches the
523 \fBzfs_arc_meta_limit\fR because dentries and inodes can pin buffers
524 in the ARC. Increasing this value will cause to dentry and inode caches
525 to be pruned more aggressively. Setting this value to 0 will disable
526 pruning the inode and dentry caches.
528 Default value: \fB10,000\fR.
534 \fBzfs_arc_meta_adjust_restarts\fR (ulong)
537 The number of restart passes to make while scanning the ARC attempting
538 the free buffers in order to stay below the \fBzfs_arc_meta_limit\fR.
539 This value should not need to be tuned but is available to facilitate
540 performance analysis.
542 Default value: \fB4096\fR.
548 \fBzfs_arc_min\fR (ulong)
553 Default value: \fB100\fR.
559 \fBzfs_arc_min_prefetch_lifespan\fR (int)
562 Minimum time prefetched blocks are locked in the ARC, specified in jiffies.
563 A value of 0 will default to 1 second.
565 Default value: \fB0\fR.
571 \fBzfs_arc_num_sublists_per_state\fR (int)
574 To allow more fine-grained locking, each ARC state contains a series
575 of lists for both data and meta data objects. Locking is performed at
576 the level of these "sub-lists". This parameters controls the number of
577 sub-lists per ARC state.
579 Default value: \fR1\fB or the number of online CPUs, whichever is greater
585 \fBzfs_arc_overflow_shift\fR (int)
588 The ARC size is considered to be overflowing if it exceeds the current
589 ARC target size (arc_c) by a threshold determined by this parameter.
590 The threshold is calculated as a fraction of arc_c using the formula
591 "arc_c >> \fBzfs_arc_overflow_shift\fR".
593 The default value of 8 causes the ARC to be considered to be overflowing
594 if it exceeds the target size by 1/256th (0.3%) of the target size.
596 When the ARC is overflowing, new buffer allocations are stalled until
597 the reclaim thread catches up and the overflow condition no longer exists.
599 Default value: \fB8\fR.
606 \fBzfs_arc_p_min_shift\fR (int)
609 arc_c shift to calc min/max arc_p
611 Default value: \fB4\fR.
617 \fBzfs_arc_p_aggressive_disable\fR (int)
620 Disable aggressive arc_p growth
622 Use \fB1\fR for yes (default) and \fB0\fR to disable.
628 \fBzfs_arc_p_dampener_disable\fR (int)
631 Disable arc_p adapt dampener
633 Use \fB1\fR for yes (default) and \fB0\fR to disable.
639 \fBzfs_arc_shrink_shift\fR (int)
642 log2(fraction of arc to reclaim)
644 Default value: \fB5\fR.
650 \fBzfs_arc_sys_free\fR (ulong)
653 The target number of bytes the ARC should leave as free memory on the system.
654 Defaults to the larger of 1/64 of physical memory or 512K. Setting this
655 option to a non-zero value will override the default.
657 Default value: \fB0\fR.
663 \fBzfs_autoimport_disable\fR (int)
666 Disable pool import at module load by ignoring the cache file (typically \fB/etc/zfs/zpool.cache\fR).
668 Use \fB1\fR for yes (default) and \fB0\fR for no.
674 \fBzfs_dbgmsg_enable\fR (int)
677 Internally ZFS keeps a small log to facilitate debugging. By default the log
678 is disabled, to enable it set this option to 1. The contents of the log can
679 be accessed by reading the /proc/spl/kstat/zfs/dbgmsg file. Writing 0 to
680 this proc file clears the log.
682 Default value: \fB0\fR.
688 \fBzfs_dbgmsg_maxsize\fR (int)
691 The maximum size in bytes of the internal ZFS debug log.
693 Default value: \fB4M\fR.
699 \fBzfs_dbuf_state_index\fR (int)
702 This feature is currently unused. It is normally used for controlling what
703 reporting is available under /proc/spl/kstat/zfs.
705 Default value: \fB0\fR.
711 \fBzfs_deadman_enabled\fR (int)
714 Enable deadman timer. See description below.
716 Use \fB1\fR for yes (default) and \fB0\fR to disable.
722 \fBzfs_deadman_synctime_ms\fR (ulong)
725 Expiration time in milliseconds. This value has two meanings. First it is
726 used to determine when the spa_deadman() logic should fire. By default the
727 spa_deadman() will fire if spa_sync() has not completed in 1000 seconds.
728 Secondly, the value determines if an I/O is considered "hung". Any I/O that
729 has not completed in zfs_deadman_synctime_ms is considered "hung" resulting
730 in a zevent being logged.
732 Default value: \fB1,000,000\fR.
738 \fBzfs_dedup_prefetch\fR (int)
741 Enable prefetching dedup-ed blks
743 Use \fB1\fR for yes and \fB0\fR to disable (default).
749 \fBzfs_delay_min_dirty_percent\fR (int)
752 Start to delay each transaction once there is this amount of dirty data,
753 expressed as a percentage of \fBzfs_dirty_data_max\fR.
754 This value should be >= zfs_vdev_async_write_active_max_dirty_percent.
755 See the section "ZFS TRANSACTION DELAY".
757 Default value: \fB60\fR.
763 \fBzfs_delay_scale\fR (int)
766 This controls how quickly the transaction delay approaches infinity.
767 Larger values cause longer delays for a given amount of dirty data.
769 For the smoothest delay, this value should be about 1 billion divided
770 by the maximum number of operations per second. This will smoothly
771 handle between 10x and 1/10th this number.
773 See the section "ZFS TRANSACTION DELAY".
775 Note: \fBzfs_delay_scale\fR * \fBzfs_dirty_data_max\fR must be < 2^64.
777 Default value: \fB500,000\fR.
783 \fBzfs_delete_blocks\fR (ulong)
786 This is the used to define a large file for the purposes of delete. Files
787 containing more than \fBzfs_delete_blocks\fR will be deleted asynchronously
788 while smaller files are deleted synchronously. Decreasing this value will
789 reduce the time spent in an unlink(2) system call at the expense of a longer
790 delay before the freed space is available.
792 Default value: \fB20,480\fR.
798 \fBzfs_dirty_data_max\fR (int)
801 Determines the dirty space limit in bytes. Once this limit is exceeded, new
802 writes are halted until space frees up. This parameter takes precedence
803 over \fBzfs_dirty_data_max_percent\fR.
804 See the section "ZFS TRANSACTION DELAY".
806 Default value: 10 percent of all memory, capped at \fBzfs_dirty_data_max_max\fR.
812 \fBzfs_dirty_data_max_max\fR (int)
815 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed in bytes.
816 This limit is only enforced at module load time, and will be ignored if
817 \fBzfs_dirty_data_max\fR is later changed. This parameter takes
818 precedence over \fBzfs_dirty_data_max_max_percent\fR. See the section
819 "ZFS TRANSACTION DELAY".
821 Default value: 25% of physical RAM.
827 \fBzfs_dirty_data_max_max_percent\fR (int)
830 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed as a
831 percentage of physical RAM. This limit is only enforced at module load
832 time, and will be ignored if \fBzfs_dirty_data_max\fR is later changed.
833 The parameter \fBzfs_dirty_data_max_max\fR takes precedence over this
834 one. See the section "ZFS TRANSACTION DELAY".
836 Default value: \fN25\fR.
842 \fBzfs_dirty_data_max_percent\fR (int)
845 Determines the dirty space limit, expressed as a percentage of all
846 memory. Once this limit is exceeded, new writes are halted until space frees
847 up. The parameter \fBzfs_dirty_data_max\fR takes precedence over this
848 one. See the section "ZFS TRANSACTION DELAY".
850 Default value: 10%, subject to \fBzfs_dirty_data_max_max\fR.
856 \fBzfs_dirty_data_sync\fR (int)
859 Start syncing out a transaction group if there is at least this much dirty data.
861 Default value: \fB67,108,864\fR.
867 \fBzfs_fletcher_4_impl\fR (string)
870 Select a fletcher 4 implementation.
872 Supported selectors are: \fBfastest\fR, \fBscalar\fR, \fBsse2\fR, \fBssse3\fR,
873 and \fBavx2\fR. All of the selectors except \fBfastest\fR and \fBscalar\fR
874 require instruction set extensions to be available and will only appear if ZFS
875 detects that they are present at runtime. If multiple implementations of
876 fletcher 4 are available, the \fBfastest\fR will be chosen using a micro
877 benchmark. Selecting \fBscalar\fR results in the original CPU based calculation
878 being used. Selecting any option other than \fBfastest\fR and \fBscalar\fR
879 results in vector instructions from the respective CPU instruction set being
882 Default value: \fBfastest\fR.
888 \fBzfs_free_bpobj_enabled\fR (int)
891 Enable/disable the processing of the free_bpobj object.
893 Default value: \fB1\fR.
899 \fBzfs_free_max_blocks\fR (ulong)
902 Maximum number of blocks freed in a single txg.
904 Default value: \fB100,000\fR.
910 \fBzfs_vdev_async_read_max_active\fR (int)
913 Maximum asynchronous read I/Os active to each device.
914 See the section "ZFS I/O SCHEDULER".
916 Default value: \fB3\fR.
922 \fBzfs_vdev_async_read_min_active\fR (int)
925 Minimum asynchronous read I/Os active to each device.
926 See the section "ZFS I/O SCHEDULER".
928 Default value: \fB1\fR.
934 \fBzfs_vdev_async_write_active_max_dirty_percent\fR (int)
937 When the pool has more than
938 \fBzfs_vdev_async_write_active_max_dirty_percent\fR dirty data, use
939 \fBzfs_vdev_async_write_max_active\fR to limit active async writes. If
940 the dirty data is between min and max, the active I/O limit is linearly
941 interpolated. See the section "ZFS I/O SCHEDULER".
943 Default value: \fB60\fR.
949 \fBzfs_vdev_async_write_active_min_dirty_percent\fR (int)
952 When the pool has less than
953 \fBzfs_vdev_async_write_active_min_dirty_percent\fR dirty data, use
954 \fBzfs_vdev_async_write_min_active\fR to limit active async writes. If
955 the dirty data is between min and max, the active I/O limit is linearly
956 interpolated. See the section "ZFS I/O SCHEDULER".
958 Default value: \fB30\fR.
964 \fBzfs_vdev_async_write_max_active\fR (int)
967 Maximum asynchronous write I/Os active to each device.
968 See the section "ZFS I/O SCHEDULER".
970 Default value: \fB10\fR.
976 \fBzfs_vdev_async_write_min_active\fR (int)
979 Minimum asynchronous write I/Os active to each device.
980 See the section "ZFS I/O SCHEDULER".
982 Default value: \fB1\fR.
988 \fBzfs_vdev_max_active\fR (int)
991 The maximum number of I/Os active to each device. Ideally, this will be >=
992 the sum of each queue's max_active. It must be at least the sum of each
993 queue's min_active. See the section "ZFS I/O SCHEDULER".
995 Default value: \fB1,000\fR.
1001 \fBzfs_vdev_scrub_max_active\fR (int)
1004 Maximum scrub I/Os active to each device.
1005 See the section "ZFS I/O SCHEDULER".
1007 Default value: \fB2\fR.
1013 \fBzfs_vdev_scrub_min_active\fR (int)
1016 Minimum scrub I/Os active to each device.
1017 See the section "ZFS I/O SCHEDULER".
1019 Default value: \fB1\fR.
1025 \fBzfs_vdev_sync_read_max_active\fR (int)
1028 Maximum synchronous read I/Os active to each device.
1029 See the section "ZFS I/O SCHEDULER".
1031 Default value: \fB10\fR.
1037 \fBzfs_vdev_sync_read_min_active\fR (int)
1040 Minimum synchronous read I/Os active to each device.
1041 See the section "ZFS I/O SCHEDULER".
1043 Default value: \fB10\fR.
1049 \fBzfs_vdev_sync_write_max_active\fR (int)
1052 Maximum synchronous write I/Os active to each device.
1053 See the section "ZFS I/O SCHEDULER".
1055 Default value: \fB10\fR.
1061 \fBzfs_vdev_sync_write_min_active\fR (int)
1064 Minimum synchronous write I/Os active to each device.
1065 See the section "ZFS I/O SCHEDULER".
1067 Default value: \fB10\fR.
1073 \fBzfs_disable_dup_eviction\fR (int)
1076 Disable duplicate buffer eviction
1078 Use \fB1\fR for yes and \fB0\fR for no (default).
1084 \fBzfs_expire_snapshot\fR (int)
1087 Seconds to expire .zfs/snapshot
1089 Default value: \fB300\fR.
1095 \fBzfs_admin_snapshot\fR (int)
1098 Allow the creation, removal, or renaming of entries in the .zfs/snapshot
1099 directory to cause the creation, destruction, or renaming of snapshots.
1100 When enabled this functionality works both locally and over NFS exports
1101 which have the 'no_root_squash' option set. This functionality is disabled
1104 Use \fB1\fR for yes and \fB0\fR for no (default).
1110 \fBzfs_flags\fR (int)
1113 Set additional debugging flags. The following flags may be bitwise-or'd
1125 Enable dprintf entries in the debug log.
1127 2 ZFS_DEBUG_DBUF_VERIFY *
1128 Enable extra dbuf verifications.
1130 4 ZFS_DEBUG_DNODE_VERIFY *
1131 Enable extra dnode verifications.
1133 8 ZFS_DEBUG_SNAPNAMES
1134 Enable snapshot name verification.
1137 Check for illegally modified ARC buffers.
1140 Enable spa_dbgmsg entries in the debug log.
1142 64 ZFS_DEBUG_ZIO_FREE
1143 Enable verification of block frees.
1145 128 ZFS_DEBUG_HISTOGRAM_VERIFY
1146 Enable extra spacemap histogram verifications.
1149 * Requires debug build.
1151 Default value: \fB0\fR.
1157 \fBzfs_free_leak_on_eio\fR (int)
1160 If destroy encounters an EIO while reading metadata (e.g. indirect
1161 blocks), space referenced by the missing metadata can not be freed.
1162 Normally this causes the background destroy to become "stalled", as
1163 it is unable to make forward progress. While in this stalled state,
1164 all remaining space to free from the error-encountering filesystem is
1165 "temporarily leaked". Set this flag to cause it to ignore the EIO,
1166 permanently leak the space from indirect blocks that can not be read,
1167 and continue to free everything else that it can.
1169 The default, "stalling" behavior is useful if the storage partially
1170 fails (i.e. some but not all i/os fail), and then later recovers. In
1171 this case, we will be able to continue pool operations while it is
1172 partially failed, and when it recovers, we can continue to free the
1173 space, with no leaks. However, note that this case is actually
1176 Typically pools either (a) fail completely (but perhaps temporarily,
1177 e.g. a top-level vdev going offline), or (b) have localized,
1178 permanent errors (e.g. disk returns the wrong data due to bit flip or
1179 firmware bug). In case (a), this setting does not matter because the
1180 pool will be suspended and the sync thread will not be able to make
1181 forward progress regardless. In case (b), because the error is
1182 permanent, the best we can do is leak the minimum amount of space,
1183 which is what setting this flag will do. Therefore, it is reasonable
1184 for this flag to normally be set, but we chose the more conservative
1185 approach of not setting it, so that there is no possibility of
1186 leaking space in the "partial temporary" failure case.
1188 Default value: \fB0\fR.
1194 \fBzfs_free_min_time_ms\fR (int)
1197 During a \fRzfs destroy\fB operation using \fRfeature@async_destroy\fB a minimum
1198 of this much time will be spent working on freeing blocks per txg.
1200 Default value: \fB1,000\fR.
1206 \fBzfs_immediate_write_sz\fR (long)
1209 Largest data block to write to zil. Larger blocks will be treated as if the
1210 dataset being written to had the property setting \fRlogbias=throughput\fB.
1212 Default value: \fB32,768\fR.
1218 \fBzfs_max_recordsize\fR (int)
1221 We currently support block sizes from 512 bytes to 16MB. The benefits of
1222 larger blocks, and thus larger IO, need to be weighed against the cost of
1223 COWing a giant block to modify one byte. Additionally, very large blocks
1224 can have an impact on i/o latency, and also potentially on the memory
1225 allocator. Therefore, we do not allow the recordsize to be set larger than
1226 zfs_max_recordsize (default 1MB). Larger blocks can be created by changing
1227 this tunable, and pools with larger blocks can always be imported and used,
1228 regardless of this setting.
1230 Default value: \fB1,048,576\fR.
1236 \fBzfs_mdcomp_disable\fR (int)
1239 Disable meta data compression
1241 Use \fB1\fR for yes and \fB0\fR for no (default).
1247 \fBzfs_metaslab_fragmentation_threshold\fR (int)
1250 Allow metaslabs to keep their active state as long as their fragmentation
1251 percentage is less than or equal to this value. An active metaslab that
1252 exceeds this threshold will no longer keep its active status allowing
1253 better metaslabs to be selected.
1255 Default value: \fB70\fR.
1261 \fBzfs_mg_fragmentation_threshold\fR (int)
1264 Metaslab groups are considered eligible for allocations if their
1265 fragmentation metric (measured as a percentage) is less than or equal to
1266 this value. If a metaslab group exceeds this threshold then it will be
1267 skipped unless all metaslab groups within the metaslab class have also
1268 crossed this threshold.
1270 Default value: \fB85\fR.
1276 \fBzfs_mg_noalloc_threshold\fR (int)
1279 Defines a threshold at which metaslab groups should be eligible for
1280 allocations. The value is expressed as a percentage of free space
1281 beyond which a metaslab group is always eligible for allocations.
1282 If a metaslab group's free space is less than or equal to the
1283 threshold, the allocator will avoid allocating to that group
1284 unless all groups in the pool have reached the threshold. Once all
1285 groups have reached the threshold, all groups are allowed to accept
1286 allocations. The default value of 0 disables the feature and causes
1287 all metaslab groups to be eligible for allocations.
1289 This parameter allows to deal with pools having heavily imbalanced
1290 vdevs such as would be the case when a new vdev has been added.
1291 Setting the threshold to a non-zero percentage will stop allocations
1292 from being made to vdevs that aren't filled to the specified percentage
1293 and allow lesser filled vdevs to acquire more allocations than they
1294 otherwise would under the old \fBzfs_mg_alloc_failures\fR facility.
1296 Default value: \fB0\fR.
1302 \fBzfs_no_scrub_io\fR (int)
1305 Set for no scrub I/O. This results in scrubs not actually scrubbing data and
1306 simply doing a metadata crawl of the pool instead.
1308 Use \fB1\fR for yes and \fB0\fR for no (default).
1314 \fBzfs_no_scrub_prefetch\fR (int)
1317 Set to disable block prefetching for scrubs.
1319 Use \fB1\fR for yes and \fB0\fR for no (default).
1325 \fBzfs_nocacheflush\fR (int)
1328 Disable cache flush operations on disks when writing. Beware, this may cause
1329 corruption if disks re-order writes.
1331 Use \fB1\fR for yes and \fB0\fR for no (default).
1337 \fBzfs_nopwrite_enabled\fR (int)
1342 Use \fB1\fR for yes (default) and \fB0\fR to disable.
1348 \fBzfs_pd_bytes_max\fR (int)
1351 The number of bytes which should be prefetched during a pool traversal
1352 (eg: \fRzfs send\fB or other data crawling operations)
1354 Default value: \fB52,428,800\fR.
1360 \fBzfs_prefetch_disable\fR (int)
1363 This tunable disables predictive prefetch. Note that it leaves "prescient"
1364 prefetch (e.g. prefetch for zfs send) intact. Unlike predictive prefetch,
1365 prescient prefetch never issues i/os that end up not being needed, so it
1366 can't hurt performance.
1368 Use \fB1\fR for yes and \fB0\fR for no (default).
1374 \fBzfs_read_chunk_size\fR (long)
1377 Bytes to read per chunk
1379 Default value: \fB1,048,576\fR.
1385 \fBzfs_read_history\fR (int)
1388 Historic statistics for the last N reads will be available in
1389 \fR/proc/spl/kstat/zfs/POOLNAME/reads\fB
1391 Default value: \fB0\fR (no data is kept).
1397 \fBzfs_read_history_hits\fR (int)
1400 Include cache hits in read history
1402 Use \fB1\fR for yes and \fB0\fR for no (default).
1408 \fBzfs_recover\fR (int)
1411 Set to attempt to recover from fatal errors. This should only be used as a
1412 last resort, as it typically results in leaked space, or worse.
1414 Use \fB1\fR for yes and \fB0\fR for no (default).
1420 \fBzfs_resilver_delay\fR (int)
1423 Number of ticks to delay prior to issuing a resilver I/O operation when
1424 a non-resilver or non-scrub I/O operation has occurred within the past
1425 \fBzfs_scan_idle\fR ticks.
1427 Default value: \fB2\fR.
1433 \fBzfs_resilver_min_time_ms\fR (int)
1436 Resilvers are processed by the sync thread. While resilvering it will spend
1437 at least this much time working on a resilver between txg flushes.
1439 Default value: \fB3,000\fR.
1445 \fBzfs_scan_idle\fR (int)
1448 Idle window in clock ticks. During a scrub or a resilver, if
1449 a non-scrub or non-resilver I/O operation has occurred during this
1450 window, the next scrub or resilver operation is delayed by, respectively
1451 \fBzfs_scrub_delay\fR or \fBzfs_resilver_delay\fR ticks.
1453 Default value: \fB50\fR.
1459 \fBzfs_scan_min_time_ms\fR (int)
1462 Scrubs are processed by the sync thread. While scrubbing it will spend
1463 at least this much time working on a scrub between txg flushes.
1465 Default value: \fB1,000\fR.
1471 \fBzfs_scrub_delay\fR (int)
1474 Number of ticks to delay prior to issuing a scrub I/O operation when
1475 a non-scrub or non-resilver I/O operation has occurred within the past
1476 \fBzfs_scan_idle\fR ticks.
1478 Default value: \fB4\fR.
1484 \fBzfs_send_corrupt_data\fR (int)
1487 Allow sending of corrupt data (ignore read/checksum errors when sending data)
1489 Use \fB1\fR for yes and \fB0\fR for no (default).
1495 \fBzfs_sync_pass_deferred_free\fR (int)
1498 Flushing of data to disk is done in passes. Defer frees starting in this pass
1500 Default value: \fB2\fR.
1506 \fBzfs_sync_pass_dont_compress\fR (int)
1509 Don't compress starting in this pass
1511 Default value: \fB5\fR.
1517 \fBzfs_sync_pass_rewrite\fR (int)
1520 Rewrite new block pointers starting in this pass
1522 Default value: \fB2\fR.
1528 \fBzfs_top_maxinflight\fR (int)
1531 Max concurrent I/Os per top-level vdev (mirrors or raidz arrays) allowed during
1532 scrub or resilver operations.
1534 Default value: \fB32\fR.
1540 \fBzfs_txg_history\fR (int)
1543 Historic statistics for the last N txgs will be available in
1544 \fR/proc/spl/kstat/zfs/POOLNAME/txgs\fB
1546 Default value: \fB0\fR.
1552 \fBzfs_txg_timeout\fR (int)
1555 Flush dirty data to disk at least every N seconds (maximum txg duration)
1557 Default value: \fB5\fR.
1563 \fBzfs_vdev_aggregation_limit\fR (int)
1566 Max vdev I/O aggregation size
1568 Default value: \fB131,072\fR.
1574 \fBzfs_vdev_cache_bshift\fR (int)
1577 Shift size to inflate reads too
1579 Default value: \fB16\fR (effectively 65536).
1585 \fBzfs_vdev_cache_max\fR (int)
1588 Inflate reads small than this value to meet the \fBzfs_vdev_cache_bshift\fR
1591 Default value: \fB16384\fR.
1597 \fBzfs_vdev_cache_size\fR (int)
1600 Total size of the per-disk cache in bytes.
1602 Currently this feature is disabled as it has been found to not be helpful
1603 for performance and in some cases harmful.
1605 Default value: \fB0\fR.
1611 \fBzfs_vdev_mirror_rotating_inc\fR (int)
1614 A number by which the balancing algorithm increments the load calculation for
1615 the purpose of selecting the least busy mirror member when an I/O immediately
1616 follows its predecessor on rotational vdevs for the purpose of making decisions
1619 Default value: \fB0\fR.
1625 \fBzfs_vdev_mirror_rotating_seek_inc\fR (int)
1628 A number by which the balancing algorithm increments the load calculation for
1629 the purpose of selecting the least busy mirror member when an I/O lacks
1630 locality as defined by the zfs_vdev_mirror_rotating_seek_offset. I/Os within
1631 this that are not immediately following the previous I/O are incremented by
1634 Default value: \fB5\fR.
1640 \fBzfs_vdev_mirror_rotating_seek_offset\fR (int)
1643 The maximum distance for the last queued I/O in which the balancing algorithm
1644 considers an I/O to have locality.
1645 See the section "ZFS I/O SCHEDULER".
1647 Default value: \fB1048576\fR.
1653 \fBzfs_vdev_mirror_non_rotating_inc\fR (int)
1656 A number by which the balancing algorithm increments the load calculation for
1657 the purpose of selecting the least busy mirror member on non-rotational vdevs
1658 when I/Os do not immediately follow one another.
1660 Default value: \fB0\fR.
1666 \fBzfs_vdev_mirror_non_rotating_seek_inc\fR (int)
1669 A number by which the balancing algorithm increments the load calculation for
1670 the purpose of selecting the least busy mirror member when an I/O lacks
1671 locality as defined by the zfs_vdev_mirror_rotating_seek_offset. I/Os within
1672 this that are not immediately following the previous I/O are incremented by
1675 Default value: \fB1\fR.
1681 \fBzfs_vdev_read_gap_limit\fR (int)
1684 Aggregate read I/O operations if the gap on-disk between them is within this
1687 Default value: \fB32,768\fR.
1693 \fBzfs_vdev_scheduler\fR (charp)
1696 Set the Linux I/O scheduler on whole disk vdevs to this scheduler
1698 Default value: \fBnoop\fR.
1704 \fBzfs_vdev_write_gap_limit\fR (int)
1707 Aggregate write I/O over gap
1709 Default value: \fB4,096\fR.
1715 \fBzfs_vdev_raidz_impl\fR (string)
1718 Parameter for selecting raidz parity implementation to use.
1720 Options marked (always) below may be selected on module load as they are
1721 supported on all systems.
1722 The remaining options may only be set after the module is loaded, as they
1723 are available only if the implementations are compiled in and supported
1724 on the running system.
1726 Once the module is loaded, the content of
1727 /sys/module/zfs/parameters/zfs_vdev_raidz_impl will show available options
1728 with the currently selected one enclosed in [].
1729 Possible options are:
1730 fastest - (always) implementation selected using built-in benchmark
1731 original - (always) original raidz implementation
1732 scalar - (always) scalar raidz implementation
1733 sse2 - implementation using SSE2 instruction set (64bit x86 only)
1734 ssse3 - implementation using SSSE3 instruction set (64bit x86 only)
1735 avx2 - implementation using AVX2 instruction set (64bit x86 only)
1737 Default value: \fBfastest\fR.
1743 \fBzfs_zevent_cols\fR (int)
1746 When zevents are logged to the console use this as the word wrap width.
1748 Default value: \fB80\fR.
1754 \fBzfs_zevent_console\fR (int)
1757 Log events to the console
1759 Use \fB1\fR for yes and \fB0\fR for no (default).
1765 \fBzfs_zevent_len_max\fR (int)
1768 Max event queue length. A value of 0 will result in a calculated value which
1769 increases with the number of CPUs in the system (minimum 64 events). Events
1770 in the queue can be viewed with the \fBzpool events\fR command.
1772 Default value: \fB0\fR.
1778 \fBzil_replay_disable\fR (int)
1781 Disable intent logging replay. Can be disabled for recovery from corrupted
1784 Use \fB1\fR for yes and \fB0\fR for no (default).
1790 \fBzil_slog_limit\fR (ulong)
1793 Max commit bytes to separate log device
1795 Default value: \fB1,048,576\fR.
1801 \fBzio_delay_max\fR (int)
1804 A zevent will be logged if a ZIO operation takes more than N milliseconds to
1805 complete. Note that this is only a logging facility, not a timeout on
1808 Default value: \fB30,000\fR.
1814 \fBzio_requeue_io_start_cut_in_line\fR (int)
1817 Prioritize requeued I/O
1819 Default value: \fB0\fR.
1825 \fBzio_taskq_batch_pct\fR (uint)
1828 Percentage of online CPUs (or CPU cores, etc) which will run a worker thread
1829 for IO. These workers are responsible for IO work such as compression and
1830 checksum calculations. Fractional number of CPUs will be rounded down.
1832 The default value of 75 was chosen to avoid using all CPUs which can result in
1833 latency issues and inconsistent application performance, especially when high
1834 compression is enabled.
1836 Default value: \fB75\fR.
1842 \fBzvol_inhibit_dev\fR (uint)
1845 Do not create zvol device nodes. This may slightly improve startup time on
1846 systems with a very large number of zvols.
1848 Use \fB1\fR for yes and \fB0\fR for no (default).
1854 \fBzvol_major\fR (uint)
1857 Major number for zvol block devices
1859 Default value: \fB230\fR.
1865 \fBzvol_max_discard_blocks\fR (ulong)
1868 Discard (aka TRIM) operations done on zvols will be done in batches of this
1869 many blocks, where block size is determined by the \fBvolblocksize\fR property
1872 Default value: \fB16,384\fR.
1878 \fBzvol_prefetch_bytes\fR (uint)
1881 When adding a zvol to the system prefetch \fBzvol_prefetch_bytes\fR
1882 from the start and end of the volume. Prefetching these regions
1883 of the volume is desirable because they are likely to be accessed
1884 immediately by \fBblkid(8)\fR or by the kernel scanning for a partition
1887 Default value: \fB131,072\fR.
1890 .SH ZFS I/O SCHEDULER
1891 ZFS issues I/O operations to leaf vdevs to satisfy and complete I/Os.
1892 The I/O scheduler determines when and in what order those operations are
1893 issued. The I/O scheduler divides operations into five I/O classes
1894 prioritized in the following order: sync read, sync write, async read,
1895 async write, and scrub/resilver. Each queue defines the minimum and
1896 maximum number of concurrent operations that may be issued to the
1897 device. In addition, the device has an aggregate maximum,
1898 \fBzfs_vdev_max_active\fR. Note that the sum of the per-queue minimums
1899 must not exceed the aggregate maximum. If the sum of the per-queue
1900 maximums exceeds the aggregate maximum, then the number of active I/Os
1901 may reach \fBzfs_vdev_max_active\fR, in which case no further I/Os will
1902 be issued regardless of whether all per-queue minimums have been met.
1904 For many physical devices, throughput increases with the number of
1905 concurrent operations, but latency typically suffers. Further, physical
1906 devices typically have a limit at which more concurrent operations have no
1907 effect on throughput or can actually cause it to decrease.
1909 The scheduler selects the next operation to issue by first looking for an
1910 I/O class whose minimum has not been satisfied. Once all are satisfied and
1911 the aggregate maximum has not been hit, the scheduler looks for classes
1912 whose maximum has not been satisfied. Iteration through the I/O classes is
1913 done in the order specified above. No further operations are issued if the
1914 aggregate maximum number of concurrent operations has been hit or if there
1915 are no operations queued for an I/O class that has not hit its maximum.
1916 Every time an I/O is queued or an operation completes, the I/O scheduler
1917 looks for new operations to issue.
1919 In general, smaller max_active's will lead to lower latency of synchronous
1920 operations. Larger max_active's may lead to higher overall throughput,
1921 depending on underlying storage.
1923 The ratio of the queues' max_actives determines the balance of performance
1924 between reads, writes, and scrubs. E.g., increasing
1925 \fBzfs_vdev_scrub_max_active\fR will cause the scrub or resilver to complete
1926 more quickly, but reads and writes to have higher latency and lower throughput.
1928 All I/O classes have a fixed maximum number of outstanding operations
1929 except for the async write class. Asynchronous writes represent the data
1930 that is committed to stable storage during the syncing stage for
1931 transaction groups. Transaction groups enter the syncing state
1932 periodically so the number of queued async writes will quickly burst up
1933 and then bleed down to zero. Rather than servicing them as quickly as
1934 possible, the I/O scheduler changes the maximum number of active async
1935 write I/Os according to the amount of dirty data in the pool. Since
1936 both throughput and latency typically increase with the number of
1937 concurrent operations issued to physical devices, reducing the
1938 burstiness in the number of concurrent operations also stabilizes the
1939 response time of operations from other -- and in particular synchronous
1940 -- queues. In broad strokes, the I/O scheduler will issue more
1941 concurrent operations from the async write queue as there's more dirty
1946 The number of concurrent operations issued for the async write I/O class
1947 follows a piece-wise linear function defined by a few adjustable points.
1950 | o---------| <-- zfs_vdev_async_write_max_active
1957 |-------o | | <-- zfs_vdev_async_write_min_active
1958 0|_______^______|_________|
1959 0% | | 100% of zfs_dirty_data_max
1961 | `-- zfs_vdev_async_write_active_max_dirty_percent
1962 `--------- zfs_vdev_async_write_active_min_dirty_percent
1965 Until the amount of dirty data exceeds a minimum percentage of the dirty
1966 data allowed in the pool, the I/O scheduler will limit the number of
1967 concurrent operations to the minimum. As that threshold is crossed, the
1968 number of concurrent operations issued increases linearly to the maximum at
1969 the specified maximum percentage of the dirty data allowed in the pool.
1971 Ideally, the amount of dirty data on a busy pool will stay in the sloped
1972 part of the function between \fBzfs_vdev_async_write_active_min_dirty_percent\fR
1973 and \fBzfs_vdev_async_write_active_max_dirty_percent\fR. If it exceeds the
1974 maximum percentage, this indicates that the rate of incoming data is
1975 greater than the rate that the backend storage can handle. In this case, we
1976 must further throttle incoming writes, as described in the next section.
1978 .SH ZFS TRANSACTION DELAY
1979 We delay transactions when we've determined that the backend storage
1980 isn't able to accommodate the rate of incoming writes.
1982 If there is already a transaction waiting, we delay relative to when
1983 that transaction will finish waiting. This way the calculated delay time
1984 is independent of the number of threads concurrently executing
1987 If we are the only waiter, wait relative to when the transaction
1988 started, rather than the current time. This credits the transaction for
1989 "time already served", e.g. reading indirect blocks.
1991 The minimum time for a transaction to take is calculated as:
1993 min_time = zfs_delay_scale * (dirty - min) / (max - dirty)
1994 min_time is then capped at 100 milliseconds.
1997 The delay has two degrees of freedom that can be adjusted via tunables. The
1998 percentage of dirty data at which we start to delay is defined by
1999 \fBzfs_delay_min_dirty_percent\fR. This should typically be at or above
2000 \fBzfs_vdev_async_write_active_max_dirty_percent\fR so that we only start to
2001 delay after writing at full speed has failed to keep up with the incoming write
2002 rate. The scale of the curve is defined by \fBzfs_delay_scale\fR. Roughly speaking,
2003 this variable determines the amount of delay at the midpoint of the curve.
2007 10ms +-------------------------------------------------------------*+
2023 2ms + (midpoint) * +
2026 | zfs_delay_scale ----------> ******** |
2027 0 +-------------------------------------*********----------------+
2028 0% <- zfs_dirty_data_max -> 100%
2031 Note that since the delay is added to the outstanding time remaining on the
2032 most recent transaction, the delay is effectively the inverse of IOPS.
2033 Here the midpoint of 500us translates to 2000 IOPS. The shape of the curve
2034 was chosen such that small changes in the amount of accumulated dirty data
2035 in the first 3/4 of the curve yield relatively small differences in the
2038 The effects can be easier to understand when the amount of delay is
2039 represented on a log scale:
2043 100ms +-------------------------------------------------------------++
2052 + zfs_delay_scale ----------> ***** +
2063 +--------------------------------------------------------------+
2064 0% <- zfs_dirty_data_max -> 100%
2067 Note here that only as the amount of dirty data approaches its limit does
2068 the delay start to increase rapidly. The goal of a properly tuned system
2069 should be to keep the amount of dirty data out of that range by first
2070 ensuring that the appropriate limits are set for the I/O scheduler to reach
2071 optimal throughput on the backend storage, and then by changing the value
2072 of \fBzfs_delay_scale\fR to increase the steepness of the curve.