2 .\" Copyright (c) 2013 by Turbo Fredriksson <turbo@bayour.com>. All rights reserved.
3 .\" Copyright (c) 2018 by Delphix. All rights reserved.
4 .\" Copyright (c) 2019 Datto Inc.
5 .\" The contents of this file are subject to the terms of the Common Development
6 .\" and Distribution License (the "License"). You may not use this file except
7 .\" in compliance with the License. You can obtain a copy of the license at
8 .\" usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing.
10 .\" See the License for the specific language governing permissions and
11 .\" limitations under the License. When distributing Covered Code, include this
12 .\" CDDL HEADER in each file and include the License file at
13 .\" usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this
14 .\" CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your
15 .\" own identifying information:
16 .\" Portions Copyright [yyyy] [name of copyright owner]
17 .TH ZFS-MODULE-PARAMETERS 5 "Feb 8, 2019"
19 zfs\-module\-parameters \- ZFS module parameters
23 Description of the different parameters to the ZFS module.
25 .SS "Module parameters"
32 \fBdbuf_cache_max_bytes\fR (ulong)
35 Maximum size in bytes of the dbuf cache. When \fB0\fR this value will default
36 to \fB1/2^dbuf_cache_shift\fR (1/32) of the target ARC size, otherwise the
37 provided value in bytes will be used. The behavior of the dbuf cache and its
38 associated settings can be observed via the \fB/proc/spl/kstat/zfs/dbufstats\fR
41 Default value: \fB0\fR.
47 \fBdbuf_metadata_cache_max_bytes\fR (ulong)
50 Maximum size in bytes of the metadata dbuf cache. When \fB0\fR this value will
51 default to \fB1/2^dbuf_cache_shift\fR (1/16) of the target ARC size, otherwise
52 the provided value in bytes will be used. The behavior of the metadata dbuf
53 cache and its associated settings can be observed via the
54 \fB/proc/spl/kstat/zfs/dbufstats\fR kstat.
56 Default value: \fB0\fR.
62 \fBdbuf_cache_hiwater_pct\fR (uint)
65 The percentage over \fBdbuf_cache_max_bytes\fR when dbufs must be evicted
68 Default value: \fB10\fR%.
74 \fBdbuf_cache_lowater_pct\fR (uint)
77 The percentage below \fBdbuf_cache_max_bytes\fR when the evict thread stops
80 Default value: \fB10\fR%.
86 \fBdbuf_cache_shift\fR (int)
89 Set the size of the dbuf cache, \fBdbuf_cache_max_bytes\fR, to a log2 fraction
90 of the target arc size.
92 Default value: \fB5\fR.
98 \fBdbuf_metadata_cache_shift\fR (int)
101 Set the size of the dbuf metadata cache, \fBdbuf_metadata_cache_max_bytes\fR,
102 to a log2 fraction of the target arc size.
104 Default value: \fB6\fR.
110 \fBignore_hole_birth\fR (int)
113 When set, the hole_birth optimization will not be used, and all holes will
114 always be sent on zfs send. Useful if you suspect your datasets are affected
115 by a bug in hole_birth.
117 Use \fB1\fR for on (default) and \fB0\fR for off.
123 \fBl2arc_feed_again\fR (int)
126 Turbo L2ARC warm-up. When the L2ARC is cold the fill interval will be set as
129 Use \fB1\fR for yes (default) and \fB0\fR to disable.
135 \fBl2arc_feed_min_ms\fR (ulong)
138 Min feed interval in milliseconds. Requires \fBl2arc_feed_again=1\fR and only
139 applicable in related situations.
141 Default value: \fB200\fR.
147 \fBl2arc_feed_secs\fR (ulong)
150 Seconds between L2ARC writing
152 Default value: \fB1\fR.
158 \fBl2arc_headroom\fR (ulong)
161 How far through the ARC lists to search for L2ARC cacheable content, expressed
162 as a multiplier of \fBl2arc_write_max\fR
164 Default value: \fB2\fR.
170 \fBl2arc_headroom_boost\fR (ulong)
173 Scales \fBl2arc_headroom\fR by this percentage when L2ARC contents are being
174 successfully compressed before writing. A value of 100 disables this feature.
176 Default value: \fB200\fR%.
182 \fBl2arc_noprefetch\fR (int)
185 Do not write buffers to L2ARC if they were prefetched but not used by
188 Use \fB1\fR for yes (default) and \fB0\fR to disable.
194 \fBl2arc_norw\fR (int)
197 No reads during writes
199 Use \fB1\fR for yes and \fB0\fR for no (default).
205 \fBl2arc_write_boost\fR (ulong)
208 Cold L2ARC devices will have \fBl2arc_write_max\fR increased by this amount
209 while they remain cold.
211 Default value: \fB8,388,608\fR.
217 \fBl2arc_write_max\fR (ulong)
220 Max write bytes per interval
222 Default value: \fB8,388,608\fR.
228 \fBmetaslab_aliquot\fR (ulong)
231 Metaslab granularity, in bytes. This is roughly similar to what would be
232 referred to as the "stripe size" in traditional RAID arrays. In normal
233 operation, ZFS will try to write this amount of data to a top-level vdev
234 before moving on to the next one.
236 Default value: \fB524,288\fR.
242 \fBmetaslab_bias_enabled\fR (int)
245 Enable metaslab group biasing based on its vdev's over- or under-utilization
246 relative to the pool.
248 Use \fB1\fR for yes (default) and \fB0\fR for no.
254 \fBmetaslab_force_ganging\fR (ulong)
257 Make some blocks above a certain size be gang blocks. This option is used
258 by the test suite to facilitate testing.
260 Default value: \fB16,777,217\fR.
266 \fBzfs_metaslab_segment_weight_enabled\fR (int)
269 Enable/disable segment-based metaslab selection.
271 Use \fB1\fR for yes (default) and \fB0\fR for no.
277 \fBzfs_metaslab_switch_threshold\fR (int)
280 When using segment-based metaslab selection, continue allocating
281 from the active metaslab until \fBzfs_metaslab_switch_threshold\fR
282 worth of buckets have been exhausted.
284 Default value: \fB2\fR.
290 \fBmetaslab_debug_load\fR (int)
293 Load all metaslabs during pool import.
295 Use \fB1\fR for yes and \fB0\fR for no (default).
301 \fBmetaslab_debug_unload\fR (int)
304 Prevent metaslabs from being unloaded.
306 Use \fB1\fR for yes and \fB0\fR for no (default).
312 \fBmetaslab_fragmentation_factor_enabled\fR (int)
315 Enable use of the fragmentation metric in computing metaslab weights.
317 Use \fB1\fR for yes (default) and \fB0\fR for no.
323 \fBzfs_vdev_default_ms_count\fR (int)
326 When a vdev is added target this number of metaslabs per top-level vdev.
328 Default value: \fB200\fR.
334 \fBzfs_vdev_min_ms_count\fR (int)
337 Minimum number of metaslabs to create in a top-level vdev.
339 Default value: \fB16\fR.
345 \fBvdev_ms_count_limit\fR (int)
348 Practical upper limit of total metaslabs per top-level vdev.
350 Default value: \fB131,072\fR.
356 \fBmetaslab_preload_enabled\fR (int)
359 Enable metaslab group preloading.
361 Use \fB1\fR for yes (default) and \fB0\fR for no.
367 \fBmetaslab_lba_weighting_enabled\fR (int)
370 Give more weight to metaslabs with lower LBAs, assuming they have
371 greater bandwidth as is typically the case on a modern constant
372 angular velocity disk drive.
374 Use \fB1\fR for yes (default) and \fB0\fR for no.
380 \fBspa_config_path\fR (charp)
385 Default value: \fB/etc/zfs/zpool.cache\fR.
391 \fBspa_asize_inflation\fR (int)
394 Multiplication factor used to estimate actual disk consumption from the
395 size of data being written. The default value is a worst case estimate,
396 but lower values may be valid for a given pool depending on its
397 configuration. Pool administrators who understand the factors involved
398 may wish to specify a more realistic inflation factor, particularly if
399 they operate close to quota or capacity limits.
401 Default value: \fB24\fR.
407 \fBspa_load_print_vdev_tree\fR (int)
410 Whether to print the vdev tree in the debugging message buffer during pool import.
411 Use 0 to disable and 1 to enable.
413 Default value: \fB0\fR.
419 \fBspa_load_verify_data\fR (int)
422 Whether to traverse data blocks during an "extreme rewind" (\fB-X\fR)
423 import. Use 0 to disable and 1 to enable.
425 An extreme rewind import normally performs a full traversal of all
426 blocks in the pool for verification. If this parameter is set to 0,
427 the traversal skips non-metadata blocks. It can be toggled once the
428 import has started to stop or start the traversal of non-metadata blocks.
430 Default value: \fB1\fR.
436 \fBspa_load_verify_metadata\fR (int)
439 Whether to traverse blocks during an "extreme rewind" (\fB-X\fR)
440 pool import. Use 0 to disable and 1 to enable.
442 An extreme rewind import normally performs a full traversal of all
443 blocks in the pool for verification. If this parameter is set to 0,
444 the traversal is not performed. It can be toggled once the import has
445 started to stop or start the traversal.
447 Default value: \fB1\fR.
453 \fBspa_load_verify_maxinflight\fR (int)
456 Maximum concurrent I/Os during the traversal performed during an "extreme
457 rewind" (\fB-X\fR) pool import.
459 Default value: \fB10000\fR.
465 \fBspa_slop_shift\fR (int)
468 Normally, we don't allow the last 3.2% (1/(2^spa_slop_shift)) of space
469 in the pool to be consumed. This ensures that we don't run the pool
470 completely out of space, due to unaccounted changes (e.g. to the MOS).
471 It also limits the worst-case time to allocate space. If we have
472 less than this amount of free space, most ZPL operations (e.g. write,
473 create) will return ENOSPC.
475 Default value: \fB5\fR.
481 \fBvdev_removal_max_span\fR (int)
484 During top-level vdev removal, chunks of data are copied from the vdev
485 which may include free space in order to trade bandwidth for IOPS.
486 This parameter determines the maximum span of free space (in bytes)
487 which will be included as "unnecessary" data in a chunk of copied data.
489 The default value here was chosen to align with
490 \fBzfs_vdev_read_gap_limit\fR, which is a similar concept when doing
491 regular reads (but there's no reason it has to be the same).
493 Default value: \fB32,768\fR.
499 \fBzfetch_array_rd_sz\fR (ulong)
502 If prefetching is enabled, disable prefetching for reads larger than this size.
504 Default value: \fB1,048,576\fR.
510 \fBzfetch_max_distance\fR (uint)
513 Max bytes to prefetch per stream (default 8MB).
515 Default value: \fB8,388,608\fR.
521 \fBzfetch_max_streams\fR (uint)
524 Max number of streams per zfetch (prefetch streams per file).
526 Default value: \fB8\fR.
532 \fBzfetch_min_sec_reap\fR (uint)
535 Min time before an active prefetch stream can be reclaimed
537 Default value: \fB2\fR.
543 \fBzfs_arc_dnode_limit\fR (ulong)
546 When the number of bytes consumed by dnodes in the ARC exceeds this number of
547 bytes, try to unpin some of it in response to demand for non-metadata. This
548 value acts as a ceiling to the amount of dnode metadata, and defaults to 0 which
549 indicates that a percent which is based on \fBzfs_arc_dnode_limit_percent\fR of
550 the ARC meta buffers that may be used for dnodes.
552 See also \fBzfs_arc_meta_prune\fR which serves a similar purpose but is used
553 when the amount of metadata in the ARC exceeds \fBzfs_arc_meta_limit\fR rather
554 than in response to overall demand for non-metadata.
557 Default value: \fB0\fR.
563 \fBzfs_arc_dnode_limit_percent\fR (ulong)
566 Percentage that can be consumed by dnodes of ARC meta buffers.
568 See also \fBzfs_arc_dnode_limit\fR which serves a similar purpose but has a
569 higher priority if set to nonzero value.
571 Default value: \fB10\fR%.
577 \fBzfs_arc_dnode_reduce_percent\fR (ulong)
580 Percentage of ARC dnodes to try to scan in response to demand for non-metadata
581 when the number of bytes consumed by dnodes exceeds \fBzfs_arc_dnode_limit\fR.
584 Default value: \fB10\fR% of the number of dnodes in the ARC.
590 \fBzfs_arc_average_blocksize\fR (int)
593 The ARC's buffer hash table is sized based on the assumption of an average
594 block size of \fBzfs_arc_average_blocksize\fR (default 8K). This works out
595 to roughly 1MB of hash table per 1GB of physical memory with 8-byte pointers.
596 For configurations with a known larger average block size this value can be
597 increased to reduce the memory footprint.
600 Default value: \fB8192\fR.
606 \fBzfs_arc_evict_batch_limit\fR (int)
609 Number ARC headers to evict per sub-list before proceeding to another sub-list.
610 This batch-style operation prevents entire sub-lists from being evicted at once
611 but comes at a cost of additional unlocking and locking.
613 Default value: \fB10\fR.
619 \fBzfs_arc_grow_retry\fR (int)
622 If set to a non zero value, it will replace the arc_grow_retry value with this value.
623 The arc_grow_retry value (default 5) is the number of seconds the ARC will wait before
624 trying to resume growth after a memory pressure event.
626 Default value: \fB0\fR.
632 \fBzfs_arc_lotsfree_percent\fR (int)
635 Throttle I/O when free system memory drops below this percentage of total
636 system memory. Setting this value to 0 will disable the throttle.
638 Default value: \fB10\fR%.
644 \fBzfs_arc_max\fR (ulong)
647 Max arc size of ARC in bytes. If set to 0 then it will consume 1/2 of system
648 RAM. This value must be at least 67108864 (64 megabytes).
650 This value can be changed dynamically with some caveats. It cannot be set back
651 to 0 while running and reducing it below the current ARC size will not cause
652 the ARC to shrink without memory pressure to induce shrinking.
654 Default value: \fB0\fR.
660 \fBzfs_arc_meta_adjust_restarts\fR (ulong)
663 The number of restart passes to make while scanning the ARC attempting
664 the free buffers in order to stay below the \fBzfs_arc_meta_limit\fR.
665 This value should not need to be tuned but is available to facilitate
666 performance analysis.
668 Default value: \fB4096\fR.
674 \fBzfs_arc_meta_limit\fR (ulong)
677 The maximum allowed size in bytes that meta data buffers are allowed to
678 consume in the ARC. When this limit is reached meta data buffers will
679 be reclaimed even if the overall arc_c_max has not been reached. This
680 value defaults to 0 which indicates that a percent which is based on
681 \fBzfs_arc_meta_limit_percent\fR of the ARC may be used for meta data.
683 This value my be changed dynamically except that it cannot be set back to 0
684 for a specific percent of the ARC; it must be set to an explicit value.
686 Default value: \fB0\fR.
692 \fBzfs_arc_meta_limit_percent\fR (ulong)
695 Percentage of ARC buffers that can be used for meta data.
697 See also \fBzfs_arc_meta_limit\fR which serves a similar purpose but has a
698 higher priority if set to nonzero value.
701 Default value: \fB75\fR%.
707 \fBzfs_arc_meta_min\fR (ulong)
710 The minimum allowed size in bytes that meta data buffers may consume in
711 the ARC. This value defaults to 0 which disables a floor on the amount
712 of the ARC devoted meta data.
714 Default value: \fB0\fR.
720 \fBzfs_arc_meta_prune\fR (int)
723 The number of dentries and inodes to be scanned looking for entries
724 which can be dropped. This may be required when the ARC reaches the
725 \fBzfs_arc_meta_limit\fR because dentries and inodes can pin buffers
726 in the ARC. Increasing this value will cause to dentry and inode caches
727 to be pruned more aggressively. Setting this value to 0 will disable
728 pruning the inode and dentry caches.
730 Default value: \fB10,000\fR.
736 \fBzfs_arc_meta_strategy\fR (int)
739 Define the strategy for ARC meta data buffer eviction (meta reclaim strategy).
740 A value of 0 (META_ONLY) will evict only the ARC meta data buffers.
741 A value of 1 (BALANCED) indicates that additional data buffers may be evicted if
742 that is required to in order to evict the required number of meta data buffers.
744 Default value: \fB1\fR.
750 \fBzfs_arc_min\fR (ulong)
753 Min arc size of ARC in bytes. If set to 0 then arc_c_min will default to
754 consuming the larger of 32M or 1/32 of total system memory.
756 Default value: \fB0\fR.
762 \fBzfs_arc_min_prefetch_ms\fR (int)
765 Minimum time prefetched blocks are locked in the ARC, specified in ms.
766 A value of \fB0\fR will default to 1000 ms.
768 Default value: \fB0\fR.
774 \fBzfs_arc_min_prescient_prefetch_ms\fR (int)
777 Minimum time "prescient prefetched" blocks are locked in the ARC, specified
778 in ms. These blocks are meant to be prefetched fairly aggresively ahead of
779 the code that may use them. A value of \fB0\fR will default to 6000 ms.
781 Default value: \fB0\fR.
787 \fBzfs_max_missing_tvds\fR (int)
790 Number of missing top-level vdevs which will be allowed during
791 pool import (only in read-only mode).
793 Default value: \fB0\fR
799 \fBzfs_multilist_num_sublists\fR (int)
802 To allow more fine-grained locking, each ARC state contains a series
803 of lists for both data and meta data objects. Locking is performed at
804 the level of these "sub-lists". This parameters controls the number of
805 sub-lists per ARC state, and also applies to other uses of the
806 multilist data structure.
808 Default value: \fB4\fR or the number of online CPUs, whichever is greater
814 \fBzfs_arc_overflow_shift\fR (int)
817 The ARC size is considered to be overflowing if it exceeds the current
818 ARC target size (arc_c) by a threshold determined by this parameter.
819 The threshold is calculated as a fraction of arc_c using the formula
820 "arc_c >> \fBzfs_arc_overflow_shift\fR".
822 The default value of 8 causes the ARC to be considered to be overflowing
823 if it exceeds the target size by 1/256th (0.3%) of the target size.
825 When the ARC is overflowing, new buffer allocations are stalled until
826 the reclaim thread catches up and the overflow condition no longer exists.
828 Default value: \fB8\fR.
835 \fBzfs_arc_p_min_shift\fR (int)
838 If set to a non zero value, this will update arc_p_min_shift (default 4)
840 arc_p_min_shift is used to shift of arc_c for calculating both min and max
843 Default value: \fB0\fR.
849 \fBzfs_arc_p_dampener_disable\fR (int)
852 Disable arc_p adapt dampener
854 Use \fB1\fR for yes (default) and \fB0\fR to disable.
860 \fBzfs_arc_shrink_shift\fR (int)
863 If set to a non zero value, this will update arc_shrink_shift (default 7)
866 Default value: \fB0\fR.
872 \fBzfs_arc_pc_percent\fR (uint)
875 Percent of pagecache to reclaim arc to
877 This tunable allows ZFS arc to play more nicely with the kernel's LRU
878 pagecache. It can guarantee that the arc size won't collapse under scanning
879 pressure on the pagecache, yet still allows arc to be reclaimed down to
880 zfs_arc_min if necessary. This value is specified as percent of pagecache
881 size (as measured by NR_FILE_PAGES) where that percent may exceed 100. This
882 only operates during memory pressure/reclaim.
884 Default value: \fB0\fR% (disabled).
890 \fBzfs_arc_sys_free\fR (ulong)
893 The target number of bytes the ARC should leave as free memory on the system.
894 Defaults to the larger of 1/64 of physical memory or 512K. Setting this
895 option to a non-zero value will override the default.
897 Default value: \fB0\fR.
903 \fBzfs_autoimport_disable\fR (int)
906 Disable pool import at module load by ignoring the cache file (typically \fB/etc/zfs/zpool.cache\fR).
908 Use \fB1\fR for yes (default) and \fB0\fR for no.
914 \fBzfs_checksums_per_second\fR (int)
917 Rate limit checksum events to this many per second. Note that this should
918 not be set below the zed thresholds (currently 10 checksums over 10 sec)
919 or else zed may not trigger any action.
927 \fBzfs_commit_timeout_pct\fR (int)
930 This controls the amount of time that a ZIL block (lwb) will remain "open"
931 when it isn't "full", and it has a thread waiting for it to be committed to
932 stable storage. The timeout is scaled based on a percentage of the last lwb
933 latency to avoid significantly impacting the latency of each individual
934 transaction record (itx).
936 Default value: \fB5\fR%.
942 \fBzfs_condense_indirect_vdevs_enable\fR (int)
945 Enable condensing indirect vdev mappings. When set to a non-zero value,
946 attempt to condense indirect vdev mappings if the mapping uses more than
947 \fBzfs_condense_min_mapping_bytes\fR bytes of memory and if the obsolete
948 space map object uses more than \fBzfs_condense_max_obsolete_bytes\fR
949 bytes on-disk. The condensing process is an attempt to save memory by
950 removing obsolete mappings.
952 Default value: \fB1\fR.
958 \fBzfs_condense_max_obsolete_bytes\fR (ulong)
961 Only attempt to condense indirect vdev mappings if the on-disk size
962 of the obsolete space map object is greater than this number of bytes
963 (see \fBfBzfs_condense_indirect_vdevs_enable\fR).
965 Default value: \fB1,073,741,824\fR.
971 \fBzfs_condense_min_mapping_bytes\fR (ulong)
974 Minimum size vdev mapping to attempt to condense (see
975 \fBzfs_condense_indirect_vdevs_enable\fR).
977 Default value: \fB131,072\fR.
983 \fBzfs_dbgmsg_enable\fR (int)
986 Internally ZFS keeps a small log to facilitate debugging. By default the log
987 is disabled, to enable it set this option to 1. The contents of the log can
988 be accessed by reading the /proc/spl/kstat/zfs/dbgmsg file. Writing 0 to
989 this proc file clears the log.
991 Default value: \fB0\fR.
997 \fBzfs_dbgmsg_maxsize\fR (int)
1000 The maximum size in bytes of the internal ZFS debug log.
1002 Default value: \fB4M\fR.
1008 \fBzfs_dbuf_state_index\fR (int)
1011 This feature is currently unused. It is normally used for controlling what
1012 reporting is available under /proc/spl/kstat/zfs.
1014 Default value: \fB0\fR.
1020 \fBzfs_deadman_enabled\fR (int)
1023 When a pool sync operation takes longer than \fBzfs_deadman_synctime_ms\fR
1024 milliseconds, or when an individual I/O takes longer than
1025 \fBzfs_deadman_ziotime_ms\fR milliseconds, then the operation is considered to
1026 be "hung". If \fBzfs_deadman_enabled\fR is set then the deadman behavior is
1027 invoked as described by the \fBzfs_deadman_failmode\fR module option.
1028 By default the deadman is enabled and configured to \fBwait\fR which results
1029 in "hung" I/Os only being logged. The deadman is automatically disabled
1030 when a pool gets suspended.
1032 Default value: \fB1\fR.
1038 \fBzfs_deadman_failmode\fR (charp)
1041 Controls the failure behavior when the deadman detects a "hung" I/O. Valid
1042 values are \fBwait\fR, \fBcontinue\fR, and \fBpanic\fR.
1044 \fBwait\fR - Wait for a "hung" I/O to complete. For each "hung" I/O a
1045 "deadman" event will be posted describing that I/O.
1047 \fBcontinue\fR - Attempt to recover from a "hung" I/O by re-dispatching it
1048 to the I/O pipeline if possible.
1050 \fBpanic\fR - Panic the system. This can be used to facilitate an automatic
1051 fail-over to a properly configured fail-over partner.
1053 Default value: \fBwait\fR.
1059 \fBzfs_deadman_checktime_ms\fR (int)
1062 Check time in milliseconds. This defines the frequency at which we check
1063 for hung I/O and potentially invoke the \fBzfs_deadman_failmode\fR behavior.
1065 Default value: \fB60,000\fR.
1071 \fBzfs_deadman_synctime_ms\fR (ulong)
1074 Interval in milliseconds after which the deadman is triggered and also
1075 the interval after which a pool sync operation is considered to be "hung".
1076 Once this limit is exceeded the deadman will be invoked every
1077 \fBzfs_deadman_checktime_ms\fR milliseconds until the pool sync completes.
1079 Default value: \fB600,000\fR.
1085 \fBzfs_deadman_ziotime_ms\fR (ulong)
1088 Interval in milliseconds after which the deadman is triggered and an
1089 individual I/O operation is considered to be "hung". As long as the I/O
1090 remains "hung" the deadman will be invoked every \fBzfs_deadman_checktime_ms\fR
1091 milliseconds until the I/O completes.
1093 Default value: \fB300,000\fR.
1099 \fBzfs_dedup_prefetch\fR (int)
1102 Enable prefetching dedup-ed blks
1104 Use \fB1\fR for yes and \fB0\fR to disable (default).
1110 \fBzfs_delay_min_dirty_percent\fR (int)
1113 Start to delay each transaction once there is this amount of dirty data,
1114 expressed as a percentage of \fBzfs_dirty_data_max\fR.
1115 This value should be >= zfs_vdev_async_write_active_max_dirty_percent.
1116 See the section "ZFS TRANSACTION DELAY".
1118 Default value: \fB60\fR%.
1124 \fBzfs_delay_scale\fR (int)
1127 This controls how quickly the transaction delay approaches infinity.
1128 Larger values cause longer delays for a given amount of dirty data.
1130 For the smoothest delay, this value should be about 1 billion divided
1131 by the maximum number of operations per second. This will smoothly
1132 handle between 10x and 1/10th this number.
1134 See the section "ZFS TRANSACTION DELAY".
1136 Note: \fBzfs_delay_scale\fR * \fBzfs_dirty_data_max\fR must be < 2^64.
1138 Default value: \fB500,000\fR.
1144 \fBzfs_slow_io_events_per_second\fR (int)
1147 Rate limit delay zevents (which report slow I/Os) to this many per second.
1155 \fBzfs_unlink_suspend_progress\fR (uint)
1158 When enabled, files will not be asynchronously removed from the list of pending
1159 unlinks and the space they consume will be leaked. Once this option has been
1160 disabled and the dataset is remounted, the pending unlinks will be processed
1161 and the freed space returned to the pool.
1162 This option is used by the test suite to facilitate testing.
1164 Uses \fB0\fR (default) to allow progress and \fB1\fR to pause progress.
1170 \fBzfs_delete_blocks\fR (ulong)
1173 This is the used to define a large file for the purposes of delete. Files
1174 containing more than \fBzfs_delete_blocks\fR will be deleted asynchronously
1175 while smaller files are deleted synchronously. Decreasing this value will
1176 reduce the time spent in an unlink(2) system call at the expense of a longer
1177 delay before the freed space is available.
1179 Default value: \fB20,480\fR.
1185 \fBzfs_dirty_data_max\fR (int)
1188 Determines the dirty space limit in bytes. Once this limit is exceeded, new
1189 writes are halted until space frees up. This parameter takes precedence
1190 over \fBzfs_dirty_data_max_percent\fR.
1191 See the section "ZFS TRANSACTION DELAY".
1193 Default value: \fB10\fR% of physical RAM, capped at \fBzfs_dirty_data_max_max\fR.
1199 \fBzfs_dirty_data_max_max\fR (int)
1202 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed in bytes.
1203 This limit is only enforced at module load time, and will be ignored if
1204 \fBzfs_dirty_data_max\fR is later changed. This parameter takes
1205 precedence over \fBzfs_dirty_data_max_max_percent\fR. See the section
1206 "ZFS TRANSACTION DELAY".
1208 Default value: \fB25\fR% of physical RAM.
1214 \fBzfs_dirty_data_max_max_percent\fR (int)
1217 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed as a
1218 percentage of physical RAM. This limit is only enforced at module load
1219 time, and will be ignored if \fBzfs_dirty_data_max\fR is later changed.
1220 The parameter \fBzfs_dirty_data_max_max\fR takes precedence over this
1221 one. See the section "ZFS TRANSACTION DELAY".
1223 Default value: \fB25\fR%.
1229 \fBzfs_dirty_data_max_percent\fR (int)
1232 Determines the dirty space limit, expressed as a percentage of all
1233 memory. Once this limit is exceeded, new writes are halted until space frees
1234 up. The parameter \fBzfs_dirty_data_max\fR takes precedence over this
1235 one. See the section "ZFS TRANSACTION DELAY".
1237 Default value: \fB10\fR%, subject to \fBzfs_dirty_data_max_max\fR.
1243 \fBzfs_dirty_data_sync_percent\fR (int)
1246 Start syncing out a transaction group if there's at least this much dirty data
1247 as a percentage of \fBzfs_dirty_data_max\fR. This should be less than
1248 \fBzfs_vdev_async_write_active_min_dirty_percent\fR.
1250 Default value: \fB20\fR% of \fBzfs_dirty_data_max\fR.
1256 \fBzfs_fletcher_4_impl\fR (string)
1259 Select a fletcher 4 implementation.
1261 Supported selectors are: \fBfastest\fR, \fBscalar\fR, \fBsse2\fR, \fBssse3\fR,
1262 \fBavx2\fR, \fBavx512f\fR, and \fBaarch64_neon\fR.
1263 All of the selectors except \fBfastest\fR and \fBscalar\fR require instruction
1264 set extensions to be available and will only appear if ZFS detects that they are
1265 present at runtime. If multiple implementations of fletcher 4 are available,
1266 the \fBfastest\fR will be chosen using a micro benchmark. Selecting \fBscalar\fR
1267 results in the original, CPU based calculation, being used. Selecting any option
1268 other than \fBfastest\fR and \fBscalar\fR results in vector instructions from
1269 the respective CPU instruction set being used.
1271 Default value: \fBfastest\fR.
1277 \fBzfs_free_bpobj_enabled\fR (int)
1280 Enable/disable the processing of the free_bpobj object.
1282 Default value: \fB1\fR.
1288 \fBzfs_async_block_max_blocks\fR (ulong)
1291 Maximum number of blocks freed in a single txg.
1293 Default value: \fB100,000\fR.
1299 \fBzfs_override_estimate_recordsize\fR (ulong)
1302 Record size calculation override for zfs send estimates.
1304 Default value: \fB0\fR.
1310 \fBzfs_vdev_async_read_max_active\fR (int)
1313 Maximum asynchronous read I/Os active to each device.
1314 See the section "ZFS I/O SCHEDULER".
1316 Default value: \fB3\fR.
1322 \fBzfs_vdev_async_read_min_active\fR (int)
1325 Minimum asynchronous read I/Os active to each device.
1326 See the section "ZFS I/O SCHEDULER".
1328 Default value: \fB1\fR.
1334 \fBzfs_vdev_async_write_active_max_dirty_percent\fR (int)
1337 When the pool has more than
1338 \fBzfs_vdev_async_write_active_max_dirty_percent\fR dirty data, use
1339 \fBzfs_vdev_async_write_max_active\fR to limit active async writes. If
1340 the dirty data is between min and max, the active I/O limit is linearly
1341 interpolated. See the section "ZFS I/O SCHEDULER".
1343 Default value: \fB60\fR%.
1349 \fBzfs_vdev_async_write_active_min_dirty_percent\fR (int)
1352 When the pool has less than
1353 \fBzfs_vdev_async_write_active_min_dirty_percent\fR dirty data, use
1354 \fBzfs_vdev_async_write_min_active\fR to limit active async writes. If
1355 the dirty data is between min and max, the active I/O limit is linearly
1356 interpolated. See the section "ZFS I/O SCHEDULER".
1358 Default value: \fB30\fR%.
1364 \fBzfs_vdev_async_write_max_active\fR (int)
1367 Maximum asynchronous write I/Os active to each device.
1368 See the section "ZFS I/O SCHEDULER".
1370 Default value: \fB10\fR.
1376 \fBzfs_vdev_async_write_min_active\fR (int)
1379 Minimum asynchronous write I/Os active to each device.
1380 See the section "ZFS I/O SCHEDULER".
1382 Lower values are associated with better latency on rotational media but poorer
1383 resilver performance. The default value of 2 was chosen as a compromise. A
1384 value of 3 has been shown to improve resilver performance further at a cost of
1385 further increasing latency.
1387 Default value: \fB2\fR.
1393 \fBzfs_vdev_initializing_max_active\fR (int)
1396 Maximum initializing I/Os active to each device.
1397 See the section "ZFS I/O SCHEDULER".
1399 Default value: \fB1\fR.
1405 \fBzfs_vdev_initializing_min_active\fR (int)
1408 Minimum initializing I/Os active to each device.
1409 See the section "ZFS I/O SCHEDULER".
1411 Default value: \fB1\fR.
1417 \fBzfs_vdev_max_active\fR (int)
1420 The maximum number of I/Os active to each device. Ideally, this will be >=
1421 the sum of each queue's max_active. It must be at least the sum of each
1422 queue's min_active. See the section "ZFS I/O SCHEDULER".
1424 Default value: \fB1,000\fR.
1430 \fBzfs_vdev_removal_max_active\fR (int)
1433 Maximum removal I/Os active to each device.
1434 See the section "ZFS I/O SCHEDULER".
1436 Default value: \fB2\fR.
1442 \fBzfs_vdev_removal_min_active\fR (int)
1445 Minimum removal I/Os active to each device.
1446 See the section "ZFS I/O SCHEDULER".
1448 Default value: \fB1\fR.
1454 \fBzfs_vdev_scrub_max_active\fR (int)
1457 Maximum scrub I/Os active to each device.
1458 See the section "ZFS I/O SCHEDULER".
1460 Default value: \fB2\fR.
1466 \fBzfs_vdev_scrub_min_active\fR (int)
1469 Minimum scrub I/Os active to each device.
1470 See the section "ZFS I/O SCHEDULER".
1472 Default value: \fB1\fR.
1478 \fBzfs_vdev_sync_read_max_active\fR (int)
1481 Maximum synchronous read I/Os active to each device.
1482 See the section "ZFS I/O SCHEDULER".
1484 Default value: \fB10\fR.
1490 \fBzfs_vdev_sync_read_min_active\fR (int)
1493 Minimum synchronous read I/Os active to each device.
1494 See the section "ZFS I/O SCHEDULER".
1496 Default value: \fB10\fR.
1502 \fBzfs_vdev_sync_write_max_active\fR (int)
1505 Maximum synchronous write I/Os active to each device.
1506 See the section "ZFS I/O SCHEDULER".
1508 Default value: \fB10\fR.
1514 \fBzfs_vdev_sync_write_min_active\fR (int)
1517 Minimum synchronous write I/Os active to each device.
1518 See the section "ZFS I/O SCHEDULER".
1520 Default value: \fB10\fR.
1526 \fBzfs_vdev_queue_depth_pct\fR (int)
1529 Maximum number of queued allocations per top-level vdev expressed as
1530 a percentage of \fBzfs_vdev_async_write_max_active\fR which allows the
1531 system to detect devices that are more capable of handling allocations
1532 and to allocate more blocks to those devices. It allows for dynamic
1533 allocation distribution when devices are imbalanced as fuller devices
1534 will tend to be slower than empty devices.
1536 See also \fBzio_dva_throttle_enabled\fR.
1538 Default value: \fB1000\fR%.
1544 \fBzfs_expire_snapshot\fR (int)
1547 Seconds to expire .zfs/snapshot
1549 Default value: \fB300\fR.
1555 \fBzfs_admin_snapshot\fR (int)
1558 Allow the creation, removal, or renaming of entries in the .zfs/snapshot
1559 directory to cause the creation, destruction, or renaming of snapshots.
1560 When enabled this functionality works both locally and over NFS exports
1561 which have the 'no_root_squash' option set. This functionality is disabled
1564 Use \fB1\fR for yes and \fB0\fR for no (default).
1570 \fBzfs_flags\fR (int)
1573 Set additional debugging flags. The following flags may be bitwise-or'd
1585 Enable dprintf entries in the debug log.
1587 2 ZFS_DEBUG_DBUF_VERIFY *
1588 Enable extra dbuf verifications.
1590 4 ZFS_DEBUG_DNODE_VERIFY *
1591 Enable extra dnode verifications.
1593 8 ZFS_DEBUG_SNAPNAMES
1594 Enable snapshot name verification.
1597 Check for illegally modified ARC buffers.
1599 64 ZFS_DEBUG_ZIO_FREE
1600 Enable verification of block frees.
1602 128 ZFS_DEBUG_HISTOGRAM_VERIFY
1603 Enable extra spacemap histogram verifications.
1605 256 ZFS_DEBUG_METASLAB_VERIFY
1606 Verify space accounting on disk matches in-core range_trees.
1608 512 ZFS_DEBUG_SET_ERROR
1609 Enable SET_ERROR and dprintf entries in the debug log.
1612 * Requires debug build.
1614 Default value: \fB0\fR.
1620 \fBzfs_free_leak_on_eio\fR (int)
1623 If destroy encounters an EIO while reading metadata (e.g. indirect
1624 blocks), space referenced by the missing metadata can not be freed.
1625 Normally this causes the background destroy to become "stalled", as
1626 it is unable to make forward progress. While in this stalled state,
1627 all remaining space to free from the error-encountering filesystem is
1628 "temporarily leaked". Set this flag to cause it to ignore the EIO,
1629 permanently leak the space from indirect blocks that can not be read,
1630 and continue to free everything else that it can.
1632 The default, "stalling" behavior is useful if the storage partially
1633 fails (i.e. some but not all i/os fail), and then later recovers. In
1634 this case, we will be able to continue pool operations while it is
1635 partially failed, and when it recovers, we can continue to free the
1636 space, with no leaks. However, note that this case is actually
1639 Typically pools either (a) fail completely (but perhaps temporarily,
1640 e.g. a top-level vdev going offline), or (b) have localized,
1641 permanent errors (e.g. disk returns the wrong data due to bit flip or
1642 firmware bug). In case (a), this setting does not matter because the
1643 pool will be suspended and the sync thread will not be able to make
1644 forward progress regardless. In case (b), because the error is
1645 permanent, the best we can do is leak the minimum amount of space,
1646 which is what setting this flag will do. Therefore, it is reasonable
1647 for this flag to normally be set, but we chose the more conservative
1648 approach of not setting it, so that there is no possibility of
1649 leaking space in the "partial temporary" failure case.
1651 Default value: \fB0\fR.
1657 \fBzfs_free_min_time_ms\fR (int)
1660 During a \fBzfs destroy\fR operation using \fBfeature@async_destroy\fR a minimum
1661 of this much time will be spent working on freeing blocks per txg.
1663 Default value: \fB1,000\fR.
1669 \fBzfs_immediate_write_sz\fR (long)
1672 Largest data block to write to zil. Larger blocks will be treated as if the
1673 dataset being written to had the property setting \fBlogbias=throughput\fR.
1675 Default value: \fB32,768\fR.
1681 \fBzfs_initialize_value\fR (ulong)
1684 Pattern written to vdev free space by \fBzpool initialize\fR.
1686 Default value: \fB16,045,690,984,833,335,022\fR (0xdeadbeefdeadbeee).
1692 \fBzfs_lua_max_instrlimit\fR (ulong)
1695 The maximum execution time limit that can be set for a ZFS channel program,
1696 specified as a number of Lua instructions.
1698 Default value: \fB100,000,000\fR.
1704 \fBzfs_lua_max_memlimit\fR (ulong)
1707 The maximum memory limit that can be set for a ZFS channel program, specified
1710 Default value: \fB104,857,600\fR.
1716 \fBzfs_max_dataset_nesting\fR (int)
1719 The maximum depth of nested datasets. This value can be tuned temporarily to
1720 fix existing datasets that exceed the predefined limit.
1722 Default value: \fB50\fR.
1728 \fBzfs_max_recordsize\fR (int)
1731 We currently support block sizes from 512 bytes to 16MB. The benefits of
1732 larger blocks, and thus larger I/O, need to be weighed against the cost of
1733 COWing a giant block to modify one byte. Additionally, very large blocks
1734 can have an impact on i/o latency, and also potentially on the memory
1735 allocator. Therefore, we do not allow the recordsize to be set larger than
1736 zfs_max_recordsize (default 1MB). Larger blocks can be created by changing
1737 this tunable, and pools with larger blocks can always be imported and used,
1738 regardless of this setting.
1740 Default value: \fB1,048,576\fR.
1746 \fBzfs_metaslab_fragmentation_threshold\fR (int)
1749 Allow metaslabs to keep their active state as long as their fragmentation
1750 percentage is less than or equal to this value. An active metaslab that
1751 exceeds this threshold will no longer keep its active status allowing
1752 better metaslabs to be selected.
1754 Default value: \fB70\fR.
1760 \fBzfs_mg_fragmentation_threshold\fR (int)
1763 Metaslab groups are considered eligible for allocations if their
1764 fragmentation metric (measured as a percentage) is less than or equal to
1765 this value. If a metaslab group exceeds this threshold then it will be
1766 skipped unless all metaslab groups within the metaslab class have also
1767 crossed this threshold.
1769 Default value: \fB85\fR.
1775 \fBzfs_mg_noalloc_threshold\fR (int)
1778 Defines a threshold at which metaslab groups should be eligible for
1779 allocations. The value is expressed as a percentage of free space
1780 beyond which a metaslab group is always eligible for allocations.
1781 If a metaslab group's free space is less than or equal to the
1782 threshold, the allocator will avoid allocating to that group
1783 unless all groups in the pool have reached the threshold. Once all
1784 groups have reached the threshold, all groups are allowed to accept
1785 allocations. The default value of 0 disables the feature and causes
1786 all metaslab groups to be eligible for allocations.
1788 This parameter allows one to deal with pools having heavily imbalanced
1789 vdevs such as would be the case when a new vdev has been added.
1790 Setting the threshold to a non-zero percentage will stop allocations
1791 from being made to vdevs that aren't filled to the specified percentage
1792 and allow lesser filled vdevs to acquire more allocations than they
1793 otherwise would under the old \fBzfs_mg_alloc_failures\fR facility.
1795 Default value: \fB0\fR.
1801 \fBzfs_ddt_data_is_special\fR (int)
1804 If enabled, ZFS will place DDT data into the special allocation class.
1806 Default value: \fB1\fR.
1812 \fBzfs_user_indirect_is_special\fR (int)
1815 If enabled, ZFS will place user data (both file and zvol) indirect blocks
1816 into the special allocation class.
1818 Default value: \fB1\fR.
1824 \fBzfs_multihost_history\fR (int)
1827 Historical statistics for the last N multihost updates will be available in
1828 \fB/proc/spl/kstat/zfs/<pool>/multihost\fR
1830 Default value: \fB0\fR.
1836 \fBzfs_multihost_interval\fR (ulong)
1839 Used to control the frequency of multihost writes which are performed when the
1840 \fBmultihost\fR pool property is on. This is one factor used to determine
1841 the length of the activity check during import.
1843 The multihost write period is \fBzfs_multihost_interval / leaf-vdevs\fR milliseconds.
1844 This means that on average a multihost write will be issued for each leaf vdev every
1845 \fBzfs_multihost_interval\fR milliseconds. In practice, the observed period can
1846 vary with the I/O load and this observed value is the delay which is stored in
1849 On import the activity check waits a minimum amount of time determined by
1850 \fBzfs_multihost_interval * zfs_multihost_import_intervals\fR. The activity
1851 check time may be further extended if the value of mmp delay found in the best
1852 uberblock indicates actual multihost updates happened at longer intervals than
1853 \fBzfs_multihost_interval\fR. A minimum value of \fB100ms\fR is enforced.
1855 Default value: \fB1000\fR.
1861 \fBzfs_multihost_import_intervals\fR (uint)
1864 Used to control the duration of the activity test on import. Smaller values of
1865 \fBzfs_multihost_import_intervals\fR will reduce the import time but increase
1866 the risk of failing to detect an active pool. The total activity check time is
1867 never allowed to drop below one second. A value of 0 is ignored and treated as
1870 Default value: \fB10\fR.
1876 \fBzfs_multihost_fail_intervals\fR (uint)
1879 Controls the behavior of the pool when multihost write failures are detected.
1881 When \fBzfs_multihost_fail_intervals = 0\fR then multihost write failures are ignored.
1882 The failures will still be reported to the ZED which depending on its
1883 configuration may take action such as suspending the pool or offlining a device.
1885 When \fBzfs_multihost_fail_intervals > 0\fR then sequential multihost write failures
1886 will cause the pool to be suspended. This occurs when
1887 \fBzfs_multihost_fail_intervals * zfs_multihost_interval\fR milliseconds have
1888 passed since the last successful multihost write. This guarantees the activity test
1889 will see multihost writes if the pool is imported.
1891 Default value: \fB5\fR.
1897 \fBzfs_no_scrub_io\fR (int)
1900 Set for no scrub I/O. This results in scrubs not actually scrubbing data and
1901 simply doing a metadata crawl of the pool instead.
1903 Use \fB1\fR for yes and \fB0\fR for no (default).
1909 \fBzfs_no_scrub_prefetch\fR (int)
1912 Set to disable block prefetching for scrubs.
1914 Use \fB1\fR for yes and \fB0\fR for no (default).
1920 \fBzfs_nocacheflush\fR (int)
1923 Disable cache flush operations on disks when writing. Setting this will
1924 cause pool corruption on power loss if a volatile out-of-order write cache
1927 Use \fB1\fR for yes and \fB0\fR for no (default).
1933 \fBzfs_nopwrite_enabled\fR (int)
1938 Use \fB1\fR for yes (default) and \fB0\fR to disable.
1944 \fBzfs_dmu_offset_next_sync\fR (int)
1947 Enable forcing txg sync to find holes. When enabled forces ZFS to act
1948 like prior versions when SEEK_HOLE or SEEK_DATA flags are used, which
1949 when a dnode is dirty causes txg's to be synced so that this data can be
1952 Use \fB1\fR for yes and \fB0\fR to disable (default).
1958 \fBzfs_pd_bytes_max\fR (int)
1961 The number of bytes which should be prefetched during a pool traversal
1962 (eg: \fBzfs send\fR or other data crawling operations)
1964 Default value: \fB52,428,800\fR.
1970 \fBzfs_per_txg_dirty_frees_percent \fR (ulong)
1973 Tunable to control percentage of dirtied indirect blocks from frees allowed
1974 into one TXG. After this threshold is crossed, additional frees will wait until
1976 A value of zero will disable this throttle.
1978 Default value: \fB5\fR, set to \fB0\fR to disable.
1984 \fBzfs_prefetch_disable\fR (int)
1987 This tunable disables predictive prefetch. Note that it leaves "prescient"
1988 prefetch (e.g. prefetch for zfs send) intact. Unlike predictive prefetch,
1989 prescient prefetch never issues i/os that end up not being needed, so it
1990 can't hurt performance.
1992 Use \fB1\fR for yes and \fB0\fR for no (default).
1998 \fBzfs_read_chunk_size\fR (long)
2001 Bytes to read per chunk
2003 Default value: \fB1,048,576\fR.
2009 \fBzfs_read_history\fR (int)
2012 Historical statistics for the last N reads will be available in
2013 \fB/proc/spl/kstat/zfs/<pool>/reads\fR
2015 Default value: \fB0\fR (no data is kept).
2021 \fBzfs_read_history_hits\fR (int)
2024 Include cache hits in read history
2026 Use \fB1\fR for yes and \fB0\fR for no (default).
2032 \fBzfs_reconstruct_indirect_combinations_max\fR (int)
2035 If an indirect split block contains more than this many possible unique
2036 combinations when being reconstructed, consider it too computationally
2037 expensive to check them all. Instead, try at most
2038 \fBzfs_reconstruct_indirect_combinations_max\fR randomly-selected
2039 combinations each time the block is accessed. This allows all segment
2040 copies to participate fairly in the reconstruction when all combinations
2041 cannot be checked and prevents repeated use of one bad copy.
2043 Default value: \fB4096\fR.
2049 \fBzfs_recover\fR (int)
2052 Set to attempt to recover from fatal errors. This should only be used as a
2053 last resort, as it typically results in leaked space, or worse.
2055 Use \fB1\fR for yes and \fB0\fR for no (default).
2061 \fBzfs_removal_ignore_errors\fR (int)
2065 Ignore hard IO errors during device removal. When set, if a device encounters
2066 a hard IO error during the removal process the removal will not be cancelled.
2067 This can result in a normally recoverable block becoming permanently damaged
2068 and is not recommended. This should only be used as a last resort when the
2069 pool cannot be returned to a healthy state prior to removing the device.
2071 Default value: \fB0\fR.
2077 \fBzfs_resilver_min_time_ms\fR (int)
2080 Resilvers are processed by the sync thread. While resilvering it will spend
2081 at least this much time working on a resilver between txg flushes.
2083 Default value: \fB3,000\fR.
2089 \fBzfs_scan_ignore_errors\fR (int)
2092 If set to a nonzero value, remove the DTL (dirty time list) upon
2093 completion of a pool scan (scrub) even if there were unrepairable
2094 errors. It is intended to be used during pool repair or recovery to
2095 stop resilvering when the pool is next imported.
2097 Default value: \fB0\fR.
2103 \fBzfs_scrub_min_time_ms\fR (int)
2106 Scrubs are processed by the sync thread. While scrubbing it will spend
2107 at least this much time working on a scrub between txg flushes.
2109 Default value: \fB1,000\fR.
2115 \fBzfs_scan_checkpoint_intval\fR (int)
2118 To preserve progress across reboots the sequential scan algorithm periodically
2119 needs to stop metadata scanning and issue all the verifications I/Os to disk.
2120 The frequency of this flushing is determined by the
2121 \fBzfs_scan_checkpoint_intval\fR tunable.
2123 Default value: \fB7200\fR seconds (every 2 hours).
2129 \fBzfs_scan_fill_weight\fR (int)
2132 This tunable affects how scrub and resilver I/O segments are ordered. A higher
2133 number indicates that we care more about how filled in a segment is, while a
2134 lower number indicates we care more about the size of the extent without
2135 considering the gaps within a segment. This value is only tunable upon module
2136 insertion. Changing the value afterwards will have no affect on scrub or
2137 resilver performance.
2139 Default value: \fB3\fR.
2145 \fBzfs_scan_issue_strategy\fR (int)
2148 Determines the order that data will be verified while scrubbing or resilvering.
2149 If set to \fB1\fR, data will be verified as sequentially as possible, given the
2150 amount of memory reserved for scrubbing (see \fBzfs_scan_mem_lim_fact\fR). This
2151 may improve scrub performance if the pool's data is very fragmented. If set to
2152 \fB2\fR, the largest mostly-contiguous chunk of found data will be verified
2153 first. By deferring scrubbing of small segments, we may later find adjacent data
2154 to coalesce and increase the segment size. If set to \fB0\fR, zfs will use
2155 strategy \fB1\fR during normal verification and strategy \fB2\fR while taking a
2158 Default value: \fB0\fR.
2164 \fBzfs_scan_legacy\fR (int)
2167 A value of 0 indicates that scrubs and resilvers will gather metadata in
2168 memory before issuing sequential I/O. A value of 1 indicates that the legacy
2169 algorithm will be used where I/O is initiated as soon as it is discovered.
2170 Changing this value to 0 will not affect scrubs or resilvers that are already
2173 Default value: \fB0\fR.
2179 \fBzfs_scan_max_ext_gap\fR (int)
2182 Indicates the largest gap in bytes between scrub / resilver I/Os that will still
2183 be considered sequential for sorting purposes. Changing this value will not
2184 affect scrubs or resilvers that are already in progress.
2186 Default value: \fB2097152 (2 MB)\fR.
2192 \fBzfs_scan_mem_lim_fact\fR (int)
2195 Maximum fraction of RAM used for I/O sorting by sequential scan algorithm.
2196 This tunable determines the hard limit for I/O sorting memory usage.
2197 When the hard limit is reached we stop scanning metadata and start issuing
2198 data verification I/O. This is done until we get below the soft limit.
2200 Default value: \fB20\fR which is 5% of RAM (1/20).
2206 \fBzfs_scan_mem_lim_soft_fact\fR (int)
2209 The fraction of the hard limit used to determined the soft limit for I/O sorting
2210 by the sequential scan algorithm. When we cross this limit from bellow no action
2211 is taken. When we cross this limit from above it is because we are issuing
2212 verification I/O. In this case (unless the metadata scan is done) we stop
2213 issuing verification I/O and start scanning metadata again until we get to the
2216 Default value: \fB20\fR which is 5% of the hard limit (1/20).
2222 \fBzfs_scan_vdev_limit\fR (int)
2225 Maximum amount of data that can be concurrently issued at once for scrubs and
2226 resilvers per leaf device, given in bytes.
2228 Default value: \fB41943040\fR.
2234 \fBzfs_send_corrupt_data\fR (int)
2237 Allow sending of corrupt data (ignore read/checksum errors when sending data)
2239 Use \fB1\fR for yes and \fB0\fR for no (default).
2245 \fBzfs_send_queue_length\fR (int)
2248 The maximum number of bytes allowed in the \fBzfs send\fR queue. This value
2249 must be at least twice the maximum block size in use.
2251 Default value: \fB16,777,216\fR.
2257 \fBzfs_recv_queue_length\fR (int)
2261 The maximum number of bytes allowed in the \fBzfs receive\fR queue. This value
2262 must be at least twice the maximum block size in use.
2264 Default value: \fB16,777,216\fR.
2270 \fBzfs_sync_pass_deferred_free\fR (int)
2273 Flushing of data to disk is done in passes. Defer frees starting in this pass
2275 Default value: \fB2\fR.
2281 \fBzfs_spa_discard_memory_limit\fR (int)
2284 Maximum memory used for prefetching a checkpoint's space map on each
2285 vdev while discarding the checkpoint.
2287 Default value: \fB16,777,216\fR.
2293 \fBzfs_sync_pass_dont_compress\fR (int)
2296 Don't compress starting in this pass
2298 Default value: \fB5\fR.
2304 \fBzfs_sync_pass_rewrite\fR (int)
2307 Rewrite new block pointers starting in this pass
2309 Default value: \fB2\fR.
2315 \fBzfs_sync_taskq_batch_pct\fR (int)
2318 This controls the number of threads used by the dp_sync_taskq. The default
2319 value of 75% will create a maximum of one thread per cpu.
2321 Default value: \fB75\fR%.
2327 \fBzfs_txg_history\fR (int)
2330 Historical statistics for the last N txgs will be available in
2331 \fB/proc/spl/kstat/zfs/<pool>/txgs\fR
2333 Default value: \fB0\fR.
2339 \fBzfs_txg_timeout\fR (int)
2342 Flush dirty data to disk at least every N seconds (maximum txg duration)
2344 Default value: \fB5\fR.
2350 \fBzfs_vdev_aggregation_limit\fR (int)
2353 Max vdev I/O aggregation size
2355 Default value: \fB131,072\fR.
2361 \fBzfs_vdev_cache_bshift\fR (int)
2364 Shift size to inflate reads too
2366 Default value: \fB16\fR (effectively 65536).
2372 \fBzfs_vdev_cache_max\fR (int)
2375 Inflate reads smaller than this value to meet the \fBzfs_vdev_cache_bshift\fR
2378 Default value: \fB16384\fR.
2384 \fBzfs_vdev_cache_size\fR (int)
2387 Total size of the per-disk cache in bytes.
2389 Currently this feature is disabled as it has been found to not be helpful
2390 for performance and in some cases harmful.
2392 Default value: \fB0\fR.
2398 \fBzfs_vdev_mirror_rotating_inc\fR (int)
2401 A number by which the balancing algorithm increments the load calculation for
2402 the purpose of selecting the least busy mirror member when an I/O immediately
2403 follows its predecessor on rotational vdevs for the purpose of making decisions
2406 Default value: \fB0\fR.
2412 \fBzfs_vdev_mirror_rotating_seek_inc\fR (int)
2415 A number by which the balancing algorithm increments the load calculation for
2416 the purpose of selecting the least busy mirror member when an I/O lacks
2417 locality as defined by the zfs_vdev_mirror_rotating_seek_offset. I/Os within
2418 this that are not immediately following the previous I/O are incremented by
2421 Default value: \fB5\fR.
2427 \fBzfs_vdev_mirror_rotating_seek_offset\fR (int)
2430 The maximum distance for the last queued I/O in which the balancing algorithm
2431 considers an I/O to have locality.
2432 See the section "ZFS I/O SCHEDULER".
2434 Default value: \fB1048576\fR.
2440 \fBzfs_vdev_mirror_non_rotating_inc\fR (int)
2443 A number by which the balancing algorithm increments the load calculation for
2444 the purpose of selecting the least busy mirror member on non-rotational vdevs
2445 when I/Os do not immediately follow one another.
2447 Default value: \fB0\fR.
2453 \fBzfs_vdev_mirror_non_rotating_seek_inc\fR (int)
2456 A number by which the balancing algorithm increments the load calculation for
2457 the purpose of selecting the least busy mirror member when an I/O lacks
2458 locality as defined by the zfs_vdev_mirror_rotating_seek_offset. I/Os within
2459 this that are not immediately following the previous I/O are incremented by
2462 Default value: \fB1\fR.
2468 \fBzfs_vdev_read_gap_limit\fR (int)
2471 Aggregate read I/O operations if the gap on-disk between them is within this
2474 Default value: \fB32,768\fR.
2480 \fBzfs_vdev_scheduler\fR (charp)
2483 Set the Linux I/O scheduler on whole disk vdevs to this scheduler. Valid options
2484 are noop, cfq, bfq & deadline
2486 Default value: \fBnoop\fR.
2492 \fBzfs_vdev_write_gap_limit\fR (int)
2495 Aggregate write I/O over gap
2497 Default value: \fB4,096\fR.
2503 \fBzfs_vdev_raidz_impl\fR (string)
2506 Parameter for selecting raidz parity implementation to use.
2508 Options marked (always) below may be selected on module load as they are
2509 supported on all systems.
2510 The remaining options may only be set after the module is loaded, as they
2511 are available only if the implementations are compiled in and supported
2512 on the running system.
2514 Once the module is loaded, the content of
2515 /sys/module/zfs/parameters/zfs_vdev_raidz_impl will show available options
2516 with the currently selected one enclosed in [].
2517 Possible options are:
2518 fastest - (always) implementation selected using built-in benchmark
2519 original - (always) original raidz implementation
2520 scalar - (always) scalar raidz implementation
2521 sse2 - implementation using SSE2 instruction set (64bit x86 only)
2522 ssse3 - implementation using SSSE3 instruction set (64bit x86 only)
2523 avx2 - implementation using AVX2 instruction set (64bit x86 only)
2524 avx512f - implementation using AVX512F instruction set (64bit x86 only)
2525 avx512bw - implementation using AVX512F & AVX512BW instruction sets (64bit x86 only)
2526 aarch64_neon - implementation using NEON (Aarch64/64 bit ARMv8 only)
2527 aarch64_neonx2 - implementation using NEON with more unrolling (Aarch64/64 bit ARMv8 only)
2529 Default value: \fBfastest\fR.
2535 \fBzfs_zevent_cols\fR (int)
2538 When zevents are logged to the console use this as the word wrap width.
2540 Default value: \fB80\fR.
2546 \fBzfs_zevent_console\fR (int)
2549 Log events to the console
2551 Use \fB1\fR for yes and \fB0\fR for no (default).
2557 \fBzfs_zevent_len_max\fR (int)
2560 Max event queue length. A value of 0 will result in a calculated value which
2561 increases with the number of CPUs in the system (minimum 64 events). Events
2562 in the queue can be viewed with the \fBzpool events\fR command.
2564 Default value: \fB0\fR.
2570 \fBzfs_zil_clean_taskq_maxalloc\fR (int)
2573 The maximum number of taskq entries that are allowed to be cached. When this
2574 limit is exceeded transaction records (itxs) will be cleaned synchronously.
2576 Default value: \fB1048576\fR.
2582 \fBzfs_zil_clean_taskq_minalloc\fR (int)
2585 The number of taskq entries that are pre-populated when the taskq is first
2586 created and are immediately available for use.
2588 Default value: \fB1024\fR.
2594 \fBzfs_zil_clean_taskq_nthr_pct\fR (int)
2597 This controls the number of threads used by the dp_zil_clean_taskq. The default
2598 value of 100% will create a maximum of one thread per cpu.
2600 Default value: \fB100\fR%.
2606 \fBzil_nocacheflush\fR (int)
2609 Disable the cache flush commands that are normally sent to the disk(s) by
2610 the ZIL after an LWB write has completed. Setting this will cause ZIL
2611 corruption on power loss if a volatile out-of-order write cache is enabled.
2613 Use \fB1\fR for yes and \fB0\fR for no (default).
2619 \fBzil_replay_disable\fR (int)
2622 Disable intent logging replay. Can be disabled for recovery from corrupted
2625 Use \fB1\fR for yes and \fB0\fR for no (default).
2631 \fBzil_slog_bulk\fR (ulong)
2634 Limit SLOG write size per commit executed with synchronous priority.
2635 Any writes above that will be executed with lower (asynchronous) priority
2636 to limit potential SLOG device abuse by single active ZIL writer.
2638 Default value: \fB786,432\fR.
2644 \fBzio_decompress_fail_fraction\fR (int)
2647 If non-zero, this value represents the denominator of the probability that zfs
2648 should induce a decompression failure. For instance, for a 5% decompression
2649 failure rate, this value should be set to 20.
2651 Default value: \fB0\fR.
2657 \fBzio_slow_io_ms\fR (int)
2660 When an I/O operation takes more than \fBzio_slow_io_ms\fR milliseconds to
2661 complete is marked as a slow I/O. Each slow I/O causes a delay zevent. Slow
2662 I/O counters can be seen with "zpool status -s".
2665 Default value: \fB30,000\fR.
2671 \fBzio_dva_throttle_enabled\fR (int)
2674 Throttle block allocations in the I/O pipeline. This allows for
2675 dynamic allocation distribution when devices are imbalanced.
2676 When enabled, the maximum number of pending allocations per top-level vdev
2677 is limited by \fBzfs_vdev_queue_depth_pct\fR.
2679 Default value: \fB1\fR.
2685 \fBzio_requeue_io_start_cut_in_line\fR (int)
2688 Prioritize requeued I/O
2690 Default value: \fB0\fR.
2696 \fBzio_taskq_batch_pct\fR (uint)
2699 Percentage of online CPUs (or CPU cores, etc) which will run a worker thread
2700 for I/O. These workers are responsible for I/O work such as compression and
2701 checksum calculations. Fractional number of CPUs will be rounded down.
2703 The default value of 75 was chosen to avoid using all CPUs which can result in
2704 latency issues and inconsistent application performance, especially when high
2705 compression is enabled.
2707 Default value: \fB75\fR.
2713 \fBzvol_inhibit_dev\fR (uint)
2716 Do not create zvol device nodes. This may slightly improve startup time on
2717 systems with a very large number of zvols.
2719 Use \fB1\fR for yes and \fB0\fR for no (default).
2725 \fBzvol_major\fR (uint)
2728 Major number for zvol block devices
2730 Default value: \fB230\fR.
2736 \fBzvol_max_discard_blocks\fR (ulong)
2739 Discard (aka TRIM) operations done on zvols will be done in batches of this
2740 many blocks, where block size is determined by the \fBvolblocksize\fR property
2743 Default value: \fB16,384\fR.
2749 \fBzvol_prefetch_bytes\fR (uint)
2752 When adding a zvol to the system prefetch \fBzvol_prefetch_bytes\fR
2753 from the start and end of the volume. Prefetching these regions
2754 of the volume is desirable because they are likely to be accessed
2755 immediately by \fBblkid(8)\fR or by the kernel scanning for a partition
2758 Default value: \fB131,072\fR.
2764 \fBzvol_request_sync\fR (uint)
2767 When processing I/O requests for a zvol submit them synchronously. This
2768 effectively limits the queue depth to 1 for each I/O submitter. When set
2769 to 0 requests are handled asynchronously by a thread pool. The number of
2770 requests which can be handled concurrently is controller by \fBzvol_threads\fR.
2772 Default value: \fB0\fR.
2778 \fBzvol_threads\fR (uint)
2781 Max number of threads which can handle zvol I/O requests concurrently.
2783 Default value: \fB32\fR.
2789 \fBzvol_volmode\fR (uint)
2792 Defines zvol block devices behaviour when \fBvolmode\fR is set to \fBdefault\fR.
2793 Valid values are \fB1\fR (full), \fB2\fR (dev) and \fB3\fR (none).
2795 Default value: \fB1\fR.
2801 \fBzfs_qat_disable\fR (int)
2804 This tunable disables qat hardware acceleration for gzip compression and.
2805 AES-GCM encryption. It is available only if qat acceleration is compiled in
2806 and the qat driver is present.
2808 Use \fB1\fR for yes and \fB0\fR for no (default).
2811 .SH ZFS I/O SCHEDULER
2812 ZFS issues I/O operations to leaf vdevs to satisfy and complete I/Os.
2813 The I/O scheduler determines when and in what order those operations are
2814 issued. The I/O scheduler divides operations into five I/O classes
2815 prioritized in the following order: sync read, sync write, async read,
2816 async write, and scrub/resilver. Each queue defines the minimum and
2817 maximum number of concurrent operations that may be issued to the
2818 device. In addition, the device has an aggregate maximum,
2819 \fBzfs_vdev_max_active\fR. Note that the sum of the per-queue minimums
2820 must not exceed the aggregate maximum. If the sum of the per-queue
2821 maximums exceeds the aggregate maximum, then the number of active I/Os
2822 may reach \fBzfs_vdev_max_active\fR, in which case no further I/Os will
2823 be issued regardless of whether all per-queue minimums have been met.
2825 For many physical devices, throughput increases with the number of
2826 concurrent operations, but latency typically suffers. Further, physical
2827 devices typically have a limit at which more concurrent operations have no
2828 effect on throughput or can actually cause it to decrease.
2830 The scheduler selects the next operation to issue by first looking for an
2831 I/O class whose minimum has not been satisfied. Once all are satisfied and
2832 the aggregate maximum has not been hit, the scheduler looks for classes
2833 whose maximum has not been satisfied. Iteration through the I/O classes is
2834 done in the order specified above. No further operations are issued if the
2835 aggregate maximum number of concurrent operations has been hit or if there
2836 are no operations queued for an I/O class that has not hit its maximum.
2837 Every time an I/O is queued or an operation completes, the I/O scheduler
2838 looks for new operations to issue.
2840 In general, smaller max_active's will lead to lower latency of synchronous
2841 operations. Larger max_active's may lead to higher overall throughput,
2842 depending on underlying storage.
2844 The ratio of the queues' max_actives determines the balance of performance
2845 between reads, writes, and scrubs. E.g., increasing
2846 \fBzfs_vdev_scrub_max_active\fR will cause the scrub or resilver to complete
2847 more quickly, but reads and writes to have higher latency and lower throughput.
2849 All I/O classes have a fixed maximum number of outstanding operations
2850 except for the async write class. Asynchronous writes represent the data
2851 that is committed to stable storage during the syncing stage for
2852 transaction groups. Transaction groups enter the syncing state
2853 periodically so the number of queued async writes will quickly burst up
2854 and then bleed down to zero. Rather than servicing them as quickly as
2855 possible, the I/O scheduler changes the maximum number of active async
2856 write I/Os according to the amount of dirty data in the pool. Since
2857 both throughput and latency typically increase with the number of
2858 concurrent operations issued to physical devices, reducing the
2859 burstiness in the number of concurrent operations also stabilizes the
2860 response time of operations from other -- and in particular synchronous
2861 -- queues. In broad strokes, the I/O scheduler will issue more
2862 concurrent operations from the async write queue as there's more dirty
2867 The number of concurrent operations issued for the async write I/O class
2868 follows a piece-wise linear function defined by a few adjustable points.
2871 | o---------| <-- zfs_vdev_async_write_max_active
2878 |-------o | | <-- zfs_vdev_async_write_min_active
2879 0|_______^______|_________|
2880 0% | | 100% of zfs_dirty_data_max
2882 | `-- zfs_vdev_async_write_active_max_dirty_percent
2883 `--------- zfs_vdev_async_write_active_min_dirty_percent
2886 Until the amount of dirty data exceeds a minimum percentage of the dirty
2887 data allowed in the pool, the I/O scheduler will limit the number of
2888 concurrent operations to the minimum. As that threshold is crossed, the
2889 number of concurrent operations issued increases linearly to the maximum at
2890 the specified maximum percentage of the dirty data allowed in the pool.
2892 Ideally, the amount of dirty data on a busy pool will stay in the sloped
2893 part of the function between \fBzfs_vdev_async_write_active_min_dirty_percent\fR
2894 and \fBzfs_vdev_async_write_active_max_dirty_percent\fR. If it exceeds the
2895 maximum percentage, this indicates that the rate of incoming data is
2896 greater than the rate that the backend storage can handle. In this case, we
2897 must further throttle incoming writes, as described in the next section.
2899 .SH ZFS TRANSACTION DELAY
2900 We delay transactions when we've determined that the backend storage
2901 isn't able to accommodate the rate of incoming writes.
2903 If there is already a transaction waiting, we delay relative to when
2904 that transaction will finish waiting. This way the calculated delay time
2905 is independent of the number of threads concurrently executing
2908 If we are the only waiter, wait relative to when the transaction
2909 started, rather than the current time. This credits the transaction for
2910 "time already served", e.g. reading indirect blocks.
2912 The minimum time for a transaction to take is calculated as:
2914 min_time = zfs_delay_scale * (dirty - min) / (max - dirty)
2915 min_time is then capped at 100 milliseconds.
2918 The delay has two degrees of freedom that can be adjusted via tunables. The
2919 percentage of dirty data at which we start to delay is defined by
2920 \fBzfs_delay_min_dirty_percent\fR. This should typically be at or above
2921 \fBzfs_vdev_async_write_active_max_dirty_percent\fR so that we only start to
2922 delay after writing at full speed has failed to keep up with the incoming write
2923 rate. The scale of the curve is defined by \fBzfs_delay_scale\fR. Roughly speaking,
2924 this variable determines the amount of delay at the midpoint of the curve.
2928 10ms +-------------------------------------------------------------*+
2944 2ms + (midpoint) * +
2947 | zfs_delay_scale ----------> ******** |
2948 0 +-------------------------------------*********----------------+
2949 0% <- zfs_dirty_data_max -> 100%
2952 Note that since the delay is added to the outstanding time remaining on the
2953 most recent transaction, the delay is effectively the inverse of IOPS.
2954 Here the midpoint of 500us translates to 2000 IOPS. The shape of the curve
2955 was chosen such that small changes in the amount of accumulated dirty data
2956 in the first 3/4 of the curve yield relatively small differences in the
2959 The effects can be easier to understand when the amount of delay is
2960 represented on a log scale:
2964 100ms +-------------------------------------------------------------++
2973 + zfs_delay_scale ----------> ***** +
2984 +--------------------------------------------------------------+
2985 0% <- zfs_dirty_data_max -> 100%
2988 Note here that only as the amount of dirty data approaches its limit does
2989 the delay start to increase rapidly. The goal of a properly tuned system
2990 should be to keep the amount of dirty data out of that range by first
2991 ensuring that the appropriate limits are set for the I/O scheduler to reach
2992 optimal throughput on the backend storage, and then by changing the value
2993 of \fBzfs_delay_scale\fR to increase the steepness of the curve.