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)
35 Use \fB1\fR for yes (default) and \fB0\fR to disable.
41 \fBl2arc_feed_min_ms\fR (ulong)
44 Min feed interval in milliseconds
46 Default value: \fB200\fR.
52 \fBl2arc_feed_secs\fR (ulong)
55 Seconds between L2ARC writing
57 Default value: \fB1\fR.
63 \fBl2arc_headroom\fR (ulong)
66 Number of max device writes to precache
68 Default value: \fB2\fR.
74 \fBl2arc_headroom_boost\fR (ulong)
77 Compressed l2arc_headroom multiplier
79 Default value: \fB200\fR.
85 \fBl2arc_nocompress\fR (int)
88 Skip compressing L2ARC buffers
90 Use \fB1\fR for yes and \fB0\fR for no (default).
96 \fBl2arc_noprefetch\fR (int)
99 Skip caching prefetched buffers
101 Use \fB1\fR for yes (default) and \fB0\fR to disable.
107 \fBl2arc_norw\fR (int)
110 No reads during writes
112 Use \fB1\fR for yes and \fB0\fR for no (default).
118 \fBl2arc_write_boost\fR (ulong)
121 Extra write bytes during device warmup
123 Default value: \fB8,388,608\fR.
129 \fBl2arc_write_max\fR (ulong)
132 Max write bytes per interval
134 Default value: \fB8,388,608\fR.
140 \fBmetaslab_bias_enabled\fR (int)
143 Enable metaslab group biasing based on its vdev's over- or under-utilization
144 relative to the pool.
146 Use \fB1\fR for yes (default) and \fB0\fR for no.
152 \fBmetaslab_debug_load\fR (int)
155 Load all metaslabs during pool import.
157 Use \fB1\fR for yes and \fB0\fR for no (default).
163 \fBmetaslab_debug_unload\fR (int)
166 Prevent metaslabs from being unloaded.
168 Use \fB1\fR for yes and \fB0\fR for no (default).
174 \fBmetaslab_fragmentation_factor_enabled\fR (int)
177 Enable use of the fragmentation metric in computing metaslab weights.
179 Use \fB1\fR for yes (default) and \fB0\fR for no.
185 \fBmetaslabs_per_vdev\fR (int)
188 When a vdev is added, it will be divided into approximately (but no more than) this number of metaslabs.
190 Default value: \fB200\fR.
196 \fBmetaslab_preload_enabled\fR (int)
199 Enable metaslab group preloading.
201 Use \fB1\fR for yes (default) and \fB0\fR for no.
207 \fBmetaslab_lba_weighting_enabled\fR (int)
210 Give more weight to metaslabs with lower LBAs, assuming they have
211 greater bandwidth as is typically the case on a modern constant
212 angular velocity disk drive.
214 Use \fB1\fR for yes (default) and \fB0\fR for no.
220 \fBspa_config_path\fR (charp)
225 Default value: \fB/etc/zfs/zpool.cache\fR.
231 \fBspa_asize_inflation\fR (int)
234 Multiplication factor used to estimate actual disk consumption from the
235 size of data being written. The default value is a worst case estimate,
236 but lower values may be valid for a given pool depending on its
237 configuration. Pool administrators who understand the factors involved
238 may wish to specify a more realistic inflation factor, particularly if
239 they operate close to quota or capacity limits.
247 \fBspa_load_verify_data\fR (int)
250 Whether to traverse data blocks during an "extreme rewind" (\fB-X\fR)
251 import. Use 0 to disable and 1 to enable.
253 An extreme rewind import normally performs a full traversal of all
254 blocks in the pool for verification. If this parameter is set to 0,
255 the traversal skips non-metadata blocks. It can be toggled once the
256 import has started to stop or start the traversal of non-metadata blocks.
264 \fBspa_load_verify_metadata\fR (int)
267 Whether to traverse blocks during an "extreme rewind" (\fB-X\fR)
268 pool import. Use 0 to disable and 1 to enable.
270 An extreme rewind import normally performs a full traversal of all
271 blocks in the pool for verification. If this parameter is set to 1,
272 the traversal is not performed. It can be toggled once the import has
273 started to stop or start the traversal.
281 \fBspa_load_verify_maxinflight\fR (int)
284 Maximum concurrent I/Os during the traversal performed during an "extreme
285 rewind" (\fB-X\fR) pool import.
293 \fBzfetch_array_rd_sz\fR (ulong)
296 If prefetching is enabled, disable prefetching for reads larger than this size.
298 Default value: \fB1,048,576\fR.
304 \fBzfetch_block_cap\fR (uint)
307 Max number of blocks to prefetch at a time
309 Default value: \fB256\fR.
315 \fBzfetch_max_streams\fR (uint)
318 Max number of streams per zfetch (prefetch streams per file).
320 Default value: \fB8\fR.
326 \fBzfetch_min_sec_reap\fR (uint)
329 Min time before an active prefetch stream can be reclaimed
331 Default value: \fB2\fR.
337 \fBzfs_arc_average_blocksize\fR (int)
340 The ARC's buffer hash table is sized based on the assumption of an average
341 block size of \fBzfs_arc_average_blocksize\fR (default 8K). This works out
342 to roughly 1MB of hash table per 1GB of physical memory with 8-byte pointers.
343 For configurations with a known larger average block size this value can be
344 increased to reduce the memory footprint.
347 Default value: \fB8192\fR.
353 \fBzfs_arc_evict_batch_limit\fR (int)
356 Number ARC headers to evict per sub-list before proceding to another sub-list.
357 This batch-style operation prevents entire sub-lists from being evicted at once
358 but comes at a cost of additional unlocking and locking.
360 Default value: \fB10\fR.
366 \fBzfs_arc_grow_retry\fR (int)
369 Seconds before growing arc size
371 Default value: \fB5\fR.
377 \fBzfs_arc_max\fR (ulong)
382 Default value: \fB0\fR.
388 \fBzfs_arc_memory_throttle_disable\fR (int)
391 Disable memory throttle
393 Use \fB1\fR for yes (default) and \fB0\fR to disable.
399 \fBzfs_arc_meta_limit\fR (ulong)
402 The maximum allowed size in bytes that meta data buffers are allowed to
403 consume in the ARC. When this limit is reached meta data buffers will
404 be reclaimed even if the overall arc_c_max has not been reached. This
405 value defaults to 0 which indicates that 3/4 of the ARC may be used
408 Default value: \fB0\fR.
414 \fBzfs_arc_meta_min\fR (ulong)
417 The minimum allowed size in bytes that meta data buffers may consume in
418 the ARC. This value defaults to 0 which disables a floor on the amount
419 of the ARC devoted meta data.
421 Default value: \fB0\fR.
427 \fBzfs_arc_meta_prune\fR (int)
430 The number of dentries and inodes to be scanned looking for entries
431 which can be dropped. This may be required when the ARC reaches the
432 \fBzfs_arc_meta_limit\fR because dentries and inodes can pin buffers
433 in the ARC. Increasing this value will cause to dentry and inode caches
434 to be pruned more aggressively. Setting this value to 0 will disable
435 pruning the inode and dentry caches.
437 Default value: \fB10,000\fR.
443 \fBzfs_arc_meta_adjust_restarts\fR (ulong)
446 The number of restart passes to make while scanning the ARC attempting
447 the free buffers in order to stay below the \fBzfs_arc_meta_limit\fR.
448 This value should not need to be tuned but is available to facilitate
449 performance analysis.
451 Default value: \fB4096\fR.
457 \fBzfs_arc_min\fR (ulong)
462 Default value: \fB100\fR.
468 \fBzfs_arc_min_prefetch_lifespan\fR (int)
471 Min life of prefetch block
473 Default value: \fB100\fR.
479 \fBzfs_arc_num_sublists_per_state\fR (int)
482 To allow more fine-grained locking, each ARC state contains a series
483 of lists for both data and meta data objects. Locking is performed at
484 the level of these "sub-lists". This parameters controls the number of
485 sub-lists per ARC state.
487 Default value: 1 or the number of on-online CPUs, whichever is greater
493 \fBzfs_arc_overflow_shift\fR (int)
496 The ARC size is considered to be overflowing if it exceeds the current
497 ARC target size (arc_c) by a threshold determined by this parameter.
498 The threshold is calculated as a fraction of arc_c using the formula
499 "arc_c >> \fBzfs_arc_overflow_shift\fR".
501 The default value of 8 causes the ARC to be considered to be overflowing
502 if it exceeds the target size by 1/256th (0.3%) of the target size.
504 When the ARC is overflowing, new buffer allocations are stalled until
505 the reclaim thread catches up and the overflow condition no longer exists.
507 Default value: \fB8\fR.
513 \fBzfs_arc_p_aggressive_disable\fR (int)
516 Disable aggressive arc_p growth
518 Use \fB1\fR for yes (default) and \fB0\fR to disable.
524 \fBzfs_arc_p_dampener_disable\fR (int)
527 Disable arc_p adapt dampener
529 Use \fB1\fR for yes (default) and \fB0\fR to disable.
535 \fBzfs_arc_shrink_shift\fR (int)
538 log2(fraction of arc to reclaim)
540 Default value: \fB5\fR.
546 \fBzfs_autoimport_disable\fR (int)
549 Disable pool import at module load by ignoring the cache file (typically \fB/etc/zfs/zpool.cache\fR).
551 Use \fB1\fR for yes (default) and \fB0\fR for no.
557 \fBzfs_dbuf_state_index\fR (int)
560 Calculate arc header index
562 Default value: \fB0\fR.
568 \fBzfs_deadman_enabled\fR (int)
573 Use \fB1\fR for yes (default) and \fB0\fR to disable.
579 \fBzfs_deadman_synctime_ms\fR (ulong)
582 Expiration time in milliseconds. This value has two meanings. First it is
583 used to determine when the spa_deadman() logic should fire. By default the
584 spa_deadman() will fire if spa_sync() has not completed in 1000 seconds.
585 Secondly, the value determines if an I/O is considered "hung". Any I/O that
586 has not completed in zfs_deadman_synctime_ms is considered "hung" resulting
587 in a zevent being logged.
589 Default value: \fB1,000,000\fR.
595 \fBzfs_dedup_prefetch\fR (int)
598 Enable prefetching dedup-ed blks
600 Use \fB1\fR for yes and \fB0\fR to disable (default).
606 \fBzfs_delay_min_dirty_percent\fR (int)
609 Start to delay each transaction once there is this amount of dirty data,
610 expressed as a percentage of \fBzfs_dirty_data_max\fR.
611 This value should be >= zfs_vdev_async_write_active_max_dirty_percent.
612 See the section "ZFS TRANSACTION DELAY".
614 Default value: \fB60\fR.
620 \fBzfs_delay_scale\fR (int)
623 This controls how quickly the transaction delay approaches infinity.
624 Larger values cause longer delays for a given amount of dirty data.
626 For the smoothest delay, this value should be about 1 billion divided
627 by the maximum number of operations per second. This will smoothly
628 handle between 10x and 1/10th this number.
630 See the section "ZFS TRANSACTION DELAY".
632 Note: \fBzfs_delay_scale\fR * \fBzfs_dirty_data_max\fR must be < 2^64.
634 Default value: \fB500,000\fR.
640 \fBzfs_dirty_data_max\fR (int)
643 Determines the dirty space limit in bytes. Once this limit is exceeded, new
644 writes are halted until space frees up. This parameter takes precedence
645 over \fBzfs_dirty_data_max_percent\fR.
646 See the section "ZFS TRANSACTION DELAY".
648 Default value: 10 percent of all memory, capped at \fBzfs_dirty_data_max_max\fR.
654 \fBzfs_dirty_data_max_max\fR (int)
657 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed in bytes.
658 This limit is only enforced at module load time, and will be ignored if
659 \fBzfs_dirty_data_max\fR is later changed. This parameter takes
660 precedence over \fBzfs_dirty_data_max_max_percent\fR. See the section
661 "ZFS TRANSACTION DELAY".
663 Default value: 25% of physical RAM.
669 \fBzfs_dirty_data_max_max_percent\fR (int)
672 Maximum allowable value of \fBzfs_dirty_data_max\fR, expressed as a
673 percentage of physical RAM. This limit is only enforced at module load
674 time, and will be ignored if \fBzfs_dirty_data_max\fR is later changed.
675 The parameter \fBzfs_dirty_data_max_max\fR takes precedence over this
676 one. See the section "ZFS TRANSACTION DELAY".
684 \fBzfs_dirty_data_max_percent\fR (int)
687 Determines the dirty space limit, expressed as a percentage of all
688 memory. Once this limit is exceeded, new writes are halted until space frees
689 up. The parameter \fBzfs_dirty_data_max\fR takes precedence over this
690 one. See the section "ZFS TRANSACTION DELAY".
692 Default value: 10%, subject to \fBzfs_dirty_data_max_max\fR.
698 \fBzfs_dirty_data_sync\fR (int)
701 Start syncing out a transaction group if there is at least this much dirty data.
703 Default value: \fB67,108,864\fR.
709 \fBzfs_free_max_blocks\fR (ulong)
712 Maximum number of blocks freed in a single txg.
714 Default value: \fB100,000\fR.
720 \fBzfs_vdev_async_read_max_active\fR (int)
723 Maxium asynchronous read I/Os active to each device.
724 See the section "ZFS I/O SCHEDULER".
726 Default value: \fB3\fR.
732 \fBzfs_vdev_async_read_min_active\fR (int)
735 Minimum asynchronous read I/Os active to each device.
736 See the section "ZFS I/O SCHEDULER".
738 Default value: \fB1\fR.
744 \fBzfs_vdev_async_write_active_max_dirty_percent\fR (int)
747 When the pool has more than
748 \fBzfs_vdev_async_write_active_max_dirty_percent\fR dirty data, use
749 \fBzfs_vdev_async_write_max_active\fR to limit active async writes. If
750 the dirty data is between min and max, the active I/O limit is linearly
751 interpolated. See the section "ZFS I/O SCHEDULER".
753 Default value: \fB60\fR.
759 \fBzfs_vdev_async_write_active_min_dirty_percent\fR (int)
762 When the pool has less than
763 \fBzfs_vdev_async_write_active_min_dirty_percent\fR dirty data, use
764 \fBzfs_vdev_async_write_min_active\fR to limit active async writes. If
765 the dirty data is between min and max, the active I/O limit is linearly
766 interpolated. See the section "ZFS I/O SCHEDULER".
768 Default value: \fB30\fR.
774 \fBzfs_vdev_async_write_max_active\fR (int)
777 Maxium asynchronous write I/Os active to each device.
778 See the section "ZFS I/O SCHEDULER".
780 Default value: \fB10\fR.
786 \fBzfs_vdev_async_write_min_active\fR (int)
789 Minimum asynchronous write I/Os active to each device.
790 See the section "ZFS I/O SCHEDULER".
792 Default value: \fB1\fR.
798 \fBzfs_vdev_max_active\fR (int)
801 The maximum number of I/Os active to each device. Ideally, this will be >=
802 the sum of each queue's max_active. It must be at least the sum of each
803 queue's min_active. See the section "ZFS I/O SCHEDULER".
805 Default value: \fB1,000\fR.
811 \fBzfs_vdev_scrub_max_active\fR (int)
814 Maxium scrub I/Os active to each device.
815 See the section "ZFS I/O SCHEDULER".
817 Default value: \fB2\fR.
823 \fBzfs_vdev_scrub_min_active\fR (int)
826 Minimum scrub I/Os active to each device.
827 See the section "ZFS I/O SCHEDULER".
829 Default value: \fB1\fR.
835 \fBzfs_vdev_sync_read_max_active\fR (int)
838 Maxium synchronous read I/Os active to each device.
839 See the section "ZFS I/O SCHEDULER".
841 Default value: \fB10\fR.
847 \fBzfs_vdev_sync_read_min_active\fR (int)
850 Minimum synchronous read I/Os active to each device.
851 See the section "ZFS I/O SCHEDULER".
853 Default value: \fB10\fR.
859 \fBzfs_vdev_sync_write_max_active\fR (int)
862 Maxium synchronous write I/Os active to each device.
863 See the section "ZFS I/O SCHEDULER".
865 Default value: \fB10\fR.
871 \fBzfs_vdev_sync_write_min_active\fR (int)
874 Minimum synchronous write I/Os active to each device.
875 See the section "ZFS I/O SCHEDULER".
877 Default value: \fB10\fR.
883 \fBzfs_disable_dup_eviction\fR (int)
886 Disable duplicate buffer eviction
888 Use \fB1\fR for yes and \fB0\fR for no (default).
894 \fBzfs_expire_snapshot\fR (int)
897 Seconds to expire .zfs/snapshot
899 Default value: \fB300\fR.
905 \fBzfs_flags\fR (int)
908 Set additional debugging flags. The following flags may be bitwise-or'd
920 Enable dprintf entries in the debug log.
922 2 ZFS_DEBUG_DBUF_VERIFY *
923 Enable extra dbuf verifications.
925 4 ZFS_DEBUG_DNODE_VERIFY *
926 Enable extra dnode verifications.
928 8 ZFS_DEBUG_SNAPNAMES
929 Enable snapshot name verification.
932 Check for illegally modified ARC buffers.
935 Enable spa_dbgmsg entries in the debug log.
937 64 ZFS_DEBUG_ZIO_FREE
938 Enable verification of block frees.
940 128 ZFS_DEBUG_HISTOGRAM_VERIFY
941 Enable extra spacemap histogram verifications.
944 * Requires debug build.
946 Default value: \fB0\fR.
952 \fBzfs_free_leak_on_eio\fR (int)
955 If destroy encounters an EIO while reading metadata (e.g. indirect
956 blocks), space referenced by the missing metadata can not be freed.
957 Normally this causes the background destroy to become "stalled", as
958 it is unable to make forward progress. While in this stalled state,
959 all remaining space to free from the error-encountering filesystem is
960 "temporarily leaked". Set this flag to cause it to ignore the EIO,
961 permanently leak the space from indirect blocks that can not be read,
962 and continue to free everything else that it can.
964 The default, "stalling" behavior is useful if the storage partially
965 fails (i.e. some but not all i/os fail), and then later recovers. In
966 this case, we will be able to continue pool operations while it is
967 partially failed, and when it recovers, we can continue to free the
968 space, with no leaks. However, note that this case is actually
971 Typically pools either (a) fail completely (but perhaps temporarily,
972 e.g. a top-level vdev going offline), or (b) have localized,
973 permanent errors (e.g. disk returns the wrong data due to bit flip or
974 firmware bug). In case (a), this setting does not matter because the
975 pool will be suspended and the sync thread will not be able to make
976 forward progress regardless. In case (b), because the error is
977 permanent, the best we can do is leak the minimum amount of space,
978 which is what setting this flag will do. Therefore, it is reasonable
979 for this flag to normally be set, but we chose the more conservative
980 approach of not setting it, so that there is no possibility of
981 leaking space in the "partial temporary" failure case.
983 Default value: \fB0\fR.
989 \fBzfs_free_min_time_ms\fR (int)
992 Min millisecs to free per txg
994 Default value: \fB1,000\fR.
1000 \fBzfs_immediate_write_sz\fR (long)
1003 Largest data block to write to zil
1005 Default value: \fB32,768\fR.
1011 \fBzfs_max_recordsize\fR (int)
1014 We currently support block sizes from 512 bytes to 16MB. The benefits of
1015 larger blocks, and thus larger IO, need to be weighed against the cost of
1016 COWing a giant block to modify one byte. Additionally, very large blocks
1017 can have an impact on i/o latency, and also potentially on the memory
1018 allocator. Therefore, we do not allow the recordsize to be set larger than
1019 zfs_max_recordsize (default 1MB). Larger blocks can be created by changing
1020 this tunable, and pools with larger blocks can always be imported and used,
1021 regardless of this setting.
1023 Default value: \fB1,048,576\fR.
1029 \fBzfs_mdcomp_disable\fR (int)
1032 Disable meta data compression
1034 Use \fB1\fR for yes and \fB0\fR for no (default).
1040 \fBzfs_metaslab_fragmentation_threshold\fR (int)
1043 Allow metaslabs to keep their active state as long as their fragmentation
1044 percentage is less than or equal to this value. An active metaslab that
1045 exceeds this threshold will no longer keep its active status allowing
1046 better metaslabs to be selected.
1048 Default value: \fB70\fR.
1054 \fBzfs_mg_fragmentation_threshold\fR (int)
1057 Metaslab groups are considered eligible for allocations if their
1058 fragmenation metric (measured as a percentage) is less than or equal to
1059 this value. If a metaslab group exceeds this threshold then it will be
1060 skipped unless all metaslab groups within the metaslab class have also
1061 crossed this threshold.
1063 Default value: \fB85\fR.
1069 \fBzfs_mg_noalloc_threshold\fR (int)
1072 Defines a threshold at which metaslab groups should be eligible for
1073 allocations. The value is expressed as a percentage of free space
1074 beyond which a metaslab group is always eligible for allocations.
1075 If a metaslab group's free space is less than or equal to the
1076 the threshold, the allocator will avoid allocating to that group
1077 unless all groups in the pool have reached the threshold. Once all
1078 groups have reached the threshold, all groups are allowed to accept
1079 allocations. The default value of 0 disables the feature and causes
1080 all metaslab groups to be eligible for allocations.
1082 This parameter allows to deal with pools having heavily imbalanced
1083 vdevs such as would be the case when a new vdev has been added.
1084 Setting the threshold to a non-zero percentage will stop allocations
1085 from being made to vdevs that aren't filled to the specified percentage
1086 and allow lesser filled vdevs to acquire more allocations than they
1087 otherwise would under the old \fBzfs_mg_alloc_failures\fR facility.
1089 Default value: \fB0\fR.
1095 \fBzfs_no_scrub_io\fR (int)
1098 Set for no scrub I/O
1100 Use \fB1\fR for yes and \fB0\fR for no (default).
1106 \fBzfs_no_scrub_prefetch\fR (int)
1109 Set for no scrub prefetching
1111 Use \fB1\fR for yes and \fB0\fR for no (default).
1117 \fBzfs_nocacheflush\fR (int)
1120 Disable cache flushes
1122 Use \fB1\fR for yes and \fB0\fR for no (default).
1128 \fBzfs_nopwrite_enabled\fR (int)
1133 Use \fB1\fR for yes (default) and \fB0\fR to disable.
1139 \fBzfs_pd_bytes_max\fR (int)
1142 The number of bytes which should be prefetched.
1144 Default value: \fB52,428,800\fR.
1150 \fBzfs_prefetch_disable\fR (int)
1153 Disable all ZFS prefetching
1155 Use \fB1\fR for yes and \fB0\fR for no (default).
1161 \fBzfs_read_chunk_size\fR (long)
1164 Bytes to read per chunk
1166 Default value: \fB1,048,576\fR.
1172 \fBzfs_read_history\fR (int)
1175 Historic statistics for the last N reads
1177 Default value: \fB0\fR.
1183 \fBzfs_read_history_hits\fR (int)
1186 Include cache hits in read history
1188 Use \fB1\fR for yes and \fB0\fR for no (default).
1194 \fBzfs_recover\fR (int)
1197 Set to attempt to recover from fatal errors. This should only be used as a
1198 last resort, as it typically results in leaked space, or worse.
1200 Use \fB1\fR for yes and \fB0\fR for no (default).
1206 \fBzfs_resilver_delay\fR (int)
1209 Number of ticks to delay prior to issuing a resilver I/O operation when
1210 a non-resilver or non-scrub I/O operation has occurred within the past
1211 \fBzfs_scan_idle\fR ticks.
1213 Default value: \fB2\fR.
1219 \fBzfs_resilver_min_time_ms\fR (int)
1222 Min millisecs to resilver per txg
1224 Default value: \fB3,000\fR.
1230 \fBzfs_scan_idle\fR (int)
1233 Idle window in clock ticks. During a scrub or a resilver, if
1234 a non-scrub or non-resilver I/O operation has occurred during this
1235 window, the next scrub or resilver operation is delayed by, respectively
1236 \fBzfs_scrub_delay\fR or \fBzfs_resilver_delay\fR ticks.
1238 Default value: \fB50\fR.
1244 \fBzfs_scan_min_time_ms\fR (int)
1247 Min millisecs to scrub per txg
1249 Default value: \fB1,000\fR.
1255 \fBzfs_scrub_delay\fR (int)
1258 Number of ticks to delay prior to issuing a scrub I/O operation when
1259 a non-scrub or non-resilver I/O operation has occurred within the past
1260 \fBzfs_scan_idle\fR ticks.
1262 Default value: \fB4\fR.
1268 \fBzfs_send_corrupt_data\fR (int)
1271 Allow to send corrupt data (ignore read/checksum errors when sending data)
1273 Use \fB1\fR for yes and \fB0\fR for no (default).
1279 \fBzfs_sync_pass_deferred_free\fR (int)
1282 Defer frees starting in this pass
1284 Default value: \fB2\fR.
1290 \fBzfs_sync_pass_dont_compress\fR (int)
1293 Don't compress starting in this pass
1295 Default value: \fB5\fR.
1301 \fBzfs_sync_pass_rewrite\fR (int)
1304 Rewrite new bps starting in this pass
1306 Default value: \fB2\fR.
1312 \fBzfs_top_maxinflight\fR (int)
1315 Max I/Os per top-level vdev during scrub or resilver operations.
1317 Default value: \fB32\fR.
1323 \fBzfs_txg_history\fR (int)
1326 Historic statistics for the last N txgs
1328 Default value: \fB0\fR.
1334 \fBzfs_txg_timeout\fR (int)
1337 Max seconds worth of delta per txg
1339 Default value: \fB5\fR.
1345 \fBzfs_vdev_aggregation_limit\fR (int)
1348 Max vdev I/O aggregation size
1350 Default value: \fB131,072\fR.
1356 \fBzfs_vdev_cache_bshift\fR (int)
1359 Shift size to inflate reads too
1361 Default value: \fB16\fR.
1367 \fBzfs_vdev_cache_max\fR (int)
1370 Inflate reads small than max
1376 \fBzfs_vdev_cache_size\fR (int)
1379 Total size of the per-disk cache
1381 Default value: \fB0\fR.
1387 \fBzfs_vdev_mirror_switch_us\fR (int)
1390 Switch mirrors every N usecs
1392 Default value: \fB10,000\fR.
1398 \fBzfs_vdev_read_gap_limit\fR (int)
1401 Aggregate read I/O over gap
1403 Default value: \fB32,768\fR.
1409 \fBzfs_vdev_scheduler\fR (charp)
1414 Default value: \fBnoop\fR.
1420 \fBzfs_vdev_write_gap_limit\fR (int)
1423 Aggregate write I/O over gap
1425 Default value: \fB4,096\fR.
1431 \fBzfs_zevent_cols\fR (int)
1434 Max event column width
1436 Default value: \fB80\fR.
1442 \fBzfs_zevent_console\fR (int)
1445 Log events to the console
1447 Use \fB1\fR for yes and \fB0\fR for no (default).
1453 \fBzfs_zevent_len_max\fR (int)
1456 Max event queue length
1458 Default value: \fB0\fR.
1464 \fBzil_replay_disable\fR (int)
1467 Disable intent logging replay
1469 Use \fB1\fR for yes and \fB0\fR for no (default).
1475 \fBzil_slog_limit\fR (ulong)
1478 Max commit bytes to separate log device
1480 Default value: \fB1,048,576\fR.
1486 \fBzio_delay_max\fR (int)
1489 Max zio millisec delay before posting event
1491 Default value: \fB30,000\fR.
1497 \fBzio_requeue_io_start_cut_in_line\fR (int)
1500 Prioritize requeued I/O
1502 Default value: \fB0\fR.
1508 \fBzvol_inhibit_dev\fR (uint)
1511 Do not create zvol device nodes
1513 Use \fB1\fR for yes and \fB0\fR for no (default).
1519 \fBzvol_major\fR (uint)
1522 Major number for zvol device
1524 Default value: \fB230\fR.
1530 \fBzvol_max_discard_blocks\fR (ulong)
1533 Max number of blocks to discard at once
1535 Default value: \fB16,384\fR.
1541 \fBzvol_threads\fR (uint)
1544 Number of threads for zvol device
1546 Default value: \fB32\fR.
1549 .SH ZFS I/O SCHEDULER
1550 ZFS issues I/O operations to leaf vdevs to satisfy and complete I/Os.
1551 The I/O scheduler determines when and in what order those operations are
1552 issued. The I/O scheduler divides operations into five I/O classes
1553 prioritized in the following order: sync read, sync write, async read,
1554 async write, and scrub/resilver. Each queue defines the minimum and
1555 maximum number of concurrent operations that may be issued to the
1556 device. In addition, the device has an aggregate maximum,
1557 \fBzfs_vdev_max_active\fR. Note that the sum of the per-queue minimums
1558 must not exceed the aggregate maximum. If the sum of the per-queue
1559 maximums exceeds the aggregate maximum, then the number of active I/Os
1560 may reach \fBzfs_vdev_max_active\fR, in which case no further I/Os will
1561 be issued regardless of whether all per-queue minimums have been met.
1563 For many physical devices, throughput increases with the number of
1564 concurrent operations, but latency typically suffers. Further, physical
1565 devices typically have a limit at which more concurrent operations have no
1566 effect on throughput or can actually cause it to decrease.
1568 The scheduler selects the next operation to issue by first looking for an
1569 I/O class whose minimum has not been satisfied. Once all are satisfied and
1570 the aggregate maximum has not been hit, the scheduler looks for classes
1571 whose maximum has not been satisfied. Iteration through the I/O classes is
1572 done in the order specified above. No further operations are issued if the
1573 aggregate maximum number of concurrent operations has been hit or if there
1574 are no operations queued for an I/O class that has not hit its maximum.
1575 Every time an I/O is queued or an operation completes, the I/O scheduler
1576 looks for new operations to issue.
1578 In general, smaller max_active's will lead to lower latency of synchronous
1579 operations. Larger max_active's may lead to higher overall throughput,
1580 depending on underlying storage.
1582 The ratio of the queues' max_actives determines the balance of performance
1583 between reads, writes, and scrubs. E.g., increasing
1584 \fBzfs_vdev_scrub_max_active\fR will cause the scrub or resilver to complete
1585 more quickly, but reads and writes to have higher latency and lower throughput.
1587 All I/O classes have a fixed maximum number of outstanding operations
1588 except for the async write class. Asynchronous writes represent the data
1589 that is committed to stable storage during the syncing stage for
1590 transaction groups. Transaction groups enter the syncing state
1591 periodically so the number of queued async writes will quickly burst up
1592 and then bleed down to zero. Rather than servicing them as quickly as
1593 possible, the I/O scheduler changes the maximum number of active async
1594 write I/Os according to the amount of dirty data in the pool. Since
1595 both throughput and latency typically increase with the number of
1596 concurrent operations issued to physical devices, reducing the
1597 burstiness in the number of concurrent operations also stabilizes the
1598 response time of operations from other -- and in particular synchronous
1599 -- queues. In broad strokes, the I/O scheduler will issue more
1600 concurrent operations from the async write queue as there's more dirty
1605 The number of concurrent operations issued for the async write I/O class
1606 follows a piece-wise linear function defined by a few adjustable points.
1609 | o---------| <-- zfs_vdev_async_write_max_active
1616 |-------o | | <-- zfs_vdev_async_write_min_active
1617 0|_______^______|_________|
1618 0% | | 100% of zfs_dirty_data_max
1620 | `-- zfs_vdev_async_write_active_max_dirty_percent
1621 `--------- zfs_vdev_async_write_active_min_dirty_percent
1624 Until the amount of dirty data exceeds a minimum percentage of the dirty
1625 data allowed in the pool, the I/O scheduler will limit the number of
1626 concurrent operations to the minimum. As that threshold is crossed, the
1627 number of concurrent operations issued increases linearly to the maximum at
1628 the specified maximum percentage of the dirty data allowed in the pool.
1630 Ideally, the amount of dirty data on a busy pool will stay in the sloped
1631 part of the function between \fBzfs_vdev_async_write_active_min_dirty_percent\fR
1632 and \fBzfs_vdev_async_write_active_max_dirty_percent\fR. If it exceeds the
1633 maximum percentage, this indicates that the rate of incoming data is
1634 greater than the rate that the backend storage can handle. In this case, we
1635 must further throttle incoming writes, as described in the next section.
1637 .SH ZFS TRANSACTION DELAY
1638 We delay transactions when we've determined that the backend storage
1639 isn't able to accommodate the rate of incoming writes.
1641 If there is already a transaction waiting, we delay relative to when
1642 that transaction will finish waiting. This way the calculated delay time
1643 is independent of the number of threads concurrently executing
1646 If we are the only waiter, wait relative to when the transaction
1647 started, rather than the current time. This credits the transaction for
1648 "time already served", e.g. reading indirect blocks.
1650 The minimum time for a transaction to take is calculated as:
1652 min_time = zfs_delay_scale * (dirty - min) / (max - dirty)
1653 min_time is then capped at 100 milliseconds.
1656 The delay has two degrees of freedom that can be adjusted via tunables. The
1657 percentage of dirty data at which we start to delay is defined by
1658 \fBzfs_delay_min_dirty_percent\fR. This should typically be at or above
1659 \fBzfs_vdev_async_write_active_max_dirty_percent\fR so that we only start to
1660 delay after writing at full speed has failed to keep up with the incoming write
1661 rate. The scale of the curve is defined by \fBzfs_delay_scale\fR. Roughly speaking,
1662 this variable determines the amount of delay at the midpoint of the curve.
1666 10ms +-------------------------------------------------------------*+
1682 2ms + (midpoint) * +
1685 | zfs_delay_scale ----------> ******** |
1686 0 +-------------------------------------*********----------------+
1687 0% <- zfs_dirty_data_max -> 100%
1690 Note that since the delay is added to the outstanding time remaining on the
1691 most recent transaction, the delay is effectively the inverse of IOPS.
1692 Here the midpoint of 500us translates to 2000 IOPS. The shape of the curve
1693 was chosen such that small changes in the amount of accumulated dirty data
1694 in the first 3/4 of the curve yield relatively small differences in the
1697 The effects can be easier to understand when the amount of delay is
1698 represented on a log scale:
1702 100ms +-------------------------------------------------------------++
1711 + zfs_delay_scale ----------> ***** +
1722 +--------------------------------------------------------------+
1723 0% <- zfs_dirty_data_max -> 100%
1726 Note here that only as the amount of dirty data approaches its limit does
1727 the delay start to increase rapidly. The goal of a properly tuned system
1728 should be to keep the amount of dirty data out of that range by first
1729 ensuring that the appropriate limits are set for the I/O scheduler to reach
1730 optimal throughput on the backend storage, and then by changing the value
1731 of \fBzfs_delay_scale\fR to increase the steepness of the curve.
projects / mirror_zfs.git / blob