.\" SPDX-License-Identifier: CDDL-1.0
.\"
.\" Copyright (c) 2013 by Turbo Fredriksson <turbo@bayour.com>. All rights reserved.
.\" Copyright (c) 2019, 2021 by Delphix. All rights reserved.
.\" Copyright (c) 2019 Datto Inc.
.\" Copyright (c) 2023, 2024, 2025, Klara, Inc.
.\" The contents of this file are subject to the terms of the Common Development
.\" and Distribution License (the "License").  You may not use this file except
.\" in compliance with the License. You can obtain a copy of the license at
.\" usr/src/OPENSOLARIS.LICENSE or https://opensource.org/licenses/CDDL-1.0.
.\"
.\" See the License for the specific language governing permissions and
.\" limitations under the License. When distributing Covered Code, include this
.\" CDDL HEADER in each file and include the License file at
.\" usr/src/OPENSOLARIS.LICENSE.  If applicable, add the following below this
.\" CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your
.\" own identifying information:
.\" Portions Copyright [yyyy] [name of copyright owner]
.\"
.Dd May 24, 2025
.Dt ZFS 4
.Os
.
.Sh NAME
.Nm zfs
.Nd tuning of the ZFS kernel module
.
.Sh DESCRIPTION
The ZFS module supports these parameters:
.Bl -tag -width Ds
.It Sy dbuf_cache_max_bytes Ns = Ns Sy UINT64_MAX Ns B Pq u64
Maximum size in bytes of the dbuf cache.
The target size is determined by the MIN versus
.No 1/2^ Ns Sy dbuf_cache_shift Pq 1/32nd
of the target ARC size.
The behavior of the dbuf cache and its associated settings
can be observed via the
.Pa /proc/spl/kstat/zfs/dbufstats
kstat.
.
.It Sy dbuf_metadata_cache_max_bytes Ns = Ns Sy UINT64_MAX Ns B Pq u64
Maximum size in bytes of the metadata dbuf cache.
The target size is determined by the MIN versus
.No 1/2^ Ns Sy dbuf_metadata_cache_shift Pq 1/64th
of the target ARC size.
The behavior of the metadata dbuf cache and its associated settings
can be observed via the
.Pa /proc/spl/kstat/zfs/dbufstats
kstat.
.
.It Sy dbuf_cache_hiwater_pct Ns = Ns Sy 10 Ns % Pq uint
The percentage over
.Sy dbuf_cache_max_bytes
when dbufs must be evicted directly.
.
.It Sy dbuf_cache_lowater_pct Ns = Ns Sy 10 Ns % Pq uint
The percentage below
.Sy dbuf_cache_max_bytes
when the evict thread stops evicting dbufs.
.
.It Sy dbuf_cache_shift Ns = Ns Sy 5 Pq uint
Set the size of the dbuf cache
.Pq Sy dbuf_cache_max_bytes
to a log2 fraction of the target ARC size.
.
.It Sy dbuf_metadata_cache_shift Ns = Ns Sy 6 Pq uint
Set the size of the dbuf metadata cache
.Pq Sy dbuf_metadata_cache_max_bytes
to a log2 fraction of the target ARC size.
.
.It Sy dbuf_mutex_cache_shift Ns = Ns Sy 0 Pq uint
Set the size of the mutex array for the dbuf cache.
When set to
.Sy 0
the array is dynamically sized based on total system memory.
.
.It Sy dmu_object_alloc_chunk_shift Ns = Ns Sy 7 Po 128 Pc Pq uint
dnode slots allocated in a single operation as a power of 2.
The default value minimizes lock contention for the bulk operation performed.
.
.It Sy dmu_ddt_copies Ns = Ns Sy 3 Pq uint
Controls the number of copies stored for DeDup Table
.Pq DDT
objects.
Reducing the number of copies to 1 from the previous default of 3
can reduce the write inflation caused by deduplication.
This assumes redundancy for this data is provided by the vdev layer.
If the DDT is damaged, space may be leaked
.Pq not freed
when the DDT can not report the correct reference count.
.
.It Sy dmu_prefetch_max Ns = Ns Sy 134217728 Ns B Po 128 MiB Pc Pq uint
Limit the amount we can prefetch with one call to this amount in bytes.
This helps to limit the amount of memory that can be used by prefetching.
.
.It Sy l2arc_feed_again Ns = Ns Sy 1 Ns | Ns 0 Pq int
Turbo L2ARC warm-up.
When the L2ARC is cold the fill interval will be set as fast as possible.
.
.It Sy l2arc_feed_min_ms Ns = Ns Sy 200 Pq u64
Min feed interval in milliseconds.
Requires
.Sy l2arc_feed_again Ns = Ns Ar 1
and only applicable in related situations.
.
.It Sy l2arc_feed_secs Ns = Ns Sy 1 Pq u64
Seconds between L2ARC writing.
.
.It Sy l2arc_headroom Ns = Ns Sy 8 Pq u64
How far through the ARC lists to search for L2ARC cacheable content,
expressed as a multiplier of
.Sy l2arc_write_max .
ARC persistence across reboots can be achieved with persistent L2ARC
by setting this parameter to
.Sy 0 ,
allowing the full length of ARC lists to be searched for cacheable content.
.
.It Sy l2arc_headroom_boost Ns = Ns Sy 200 Ns % Pq u64
Scales
.Sy l2arc_headroom
by this percentage when L2ARC contents are being successfully compressed
before writing.
A value of
.Sy 100
disables this feature.
.
.It Sy l2arc_exclude_special Ns = Ns Sy 0 Ns | Ns 1 Pq int
Controls whether buffers present on special vdevs are eligible for caching
into L2ARC.
If set to 1, exclude dbufs on special vdevs from being cached to L2ARC.
.
.It Sy l2arc_mfuonly Ns = Ns Sy 0 Ns | Ns 1 Ns | Ns 2 Pq int
Controls whether only MFU metadata and data are cached from ARC into L2ARC.
This may be desired to avoid wasting space on L2ARC when reading/writing large
amounts of data that are not expected to be accessed more than once.
.Pp
The default is 0,
meaning both MRU and MFU data and metadata are cached.
When turning off this feature (setting it to 0), some MRU buffers will
still be present in ARC and eventually cached on L2ARC.
.No If Sy l2arc_noprefetch Ns = Ns Sy 0 ,
some prefetched buffers will be cached to L2ARC, and those might later
transition to MRU, in which case the
.Sy l2arc_mru_asize No arcstat will not be Sy 0 .
.Pp
Setting it to 1 means to L2 cache only MFU data and metadata.
.Pp
Setting it to 2 means to L2 cache all metadata (MRU+MFU) but
only MFU data (i.e. MRU data are not cached). This can be the right setting
to cache as much metadata as possible even when having high data turnover.
.Pp
Regardless of
.Sy l2arc_noprefetch ,
some MFU buffers might be evicted from ARC,
accessed later on as prefetches and transition to MRU as prefetches.
If accessed again they are counted as MRU and the
.Sy l2arc_mru_asize No arcstat will not be Sy 0 .
.Pp
The ARC status of L2ARC buffers when they were first cached in
L2ARC can be seen in the
.Sy l2arc_mru_asize , Sy l2arc_mfu_asize , No and Sy l2arc_prefetch_asize
arcstats when importing the pool or onlining a cache
device if persistent L2ARC is enabled.
.Pp
The
.Sy evict_l2_eligible_mru
arcstat does not take into account if this option is enabled as the information
provided by the
.Sy evict_l2_eligible_m[rf]u
arcstats can be used to decide if toggling this option is appropriate
for the current workload.
.
.It Sy l2arc_meta_percent Ns = Ns Sy 33 Ns % Pq uint
Percent of ARC size allowed for L2ARC-only headers.
Since L2ARC buffers are not evicted on memory pressure,
too many headers on a system with an irrationally large L2ARC
can render it slow or unusable.
This parameter limits L2ARC writes and rebuilds to achieve the target.
.
.It Sy l2arc_trim_ahead Ns = Ns Sy 0 Ns % Pq u64
Trims ahead of the current write size
.Pq Sy l2arc_write_max
on L2ARC devices by this percentage of write size if we have filled the device.
If set to
.Sy 100
we TRIM twice the space required to accommodate upcoming writes.
A minimum of
.Sy 64 MiB
will be trimmed.
It also enables TRIM of the whole L2ARC device upon creation
or addition to an existing pool or if the header of the device is
invalid upon importing a pool or onlining a cache device.
A value of
.Sy 0
disables TRIM on L2ARC altogether and is the default as it can put significant
stress on the underlying storage devices.
This will vary depending of how well the specific device handles these commands.
.
.It Sy l2arc_noprefetch Ns = Ns Sy 1 Ns | Ns 0 Pq int
Do not write buffers to L2ARC if they were prefetched but not used by
applications.
In case there are prefetched buffers in L2ARC and this option
is later set, we do not read the prefetched buffers from L2ARC.
Unsetting this option is useful for caching sequential reads from the
disks to L2ARC and serve those reads from L2ARC later on.
This may be beneficial in case the L2ARC device is significantly faster
in sequential reads than the disks of the pool.
.Pp
Use
.Sy 1
to disable and
.Sy 0
to enable caching/reading prefetches to/from L2ARC.
.
.It Sy l2arc_norw Ns = Ns Sy 0 Ns | Ns 1 Pq int
No reads during writes.
.
.It Sy l2arc_write_boost Ns = Ns Sy 33554432 Ns B Po 32 MiB Pc Pq u64
Cold L2ARC devices will have
.Sy l2arc_write_max
increased by this amount while they remain cold.
.
.It Sy l2arc_write_max Ns = Ns Sy 33554432 Ns B Po 32 MiB Pc Pq u64
Max write bytes per interval.
.
.It Sy l2arc_rebuild_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Rebuild the L2ARC when importing a pool (persistent L2ARC).
This can be disabled if there are problems importing a pool
or attaching an L2ARC device (e.g. the L2ARC device is slow
in reading stored log metadata, or the metadata
has become somehow fragmented/unusable).
.
.It Sy l2arc_rebuild_blocks_min_l2size Ns = Ns Sy 1073741824 Ns B Po 1 GiB Pc Pq u64
Minimum size of an L2ARC device required in order to write log blocks in it.
The log blocks are used upon importing the pool to rebuild the persistent L2ARC.
.Pp
For L2ARC devices less than 1 GiB, the amount of data
.Fn l2arc_evict
evicts is significant compared to the amount of restored L2ARC data.
In this case, do not write log blocks in L2ARC in order not to waste space.
.
.It Sy metaslab_aliquot Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq u64
Metaslab granularity, in bytes.
This is roughly similar to what would be referred to as the "stripe size"
in traditional RAID arrays.
In normal operation, ZFS will try to write this amount of data to each disk
before moving on to the next top-level vdev.
.
.It Sy metaslab_bias_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enable metaslab group biasing based on their vdevs' over- or under-utilization
relative to the pool.
.
.It Sy metaslab_force_ganging Ns = Ns Sy 16777217 Ns B Po 16 MiB + 1 B Pc Pq u64
Make some blocks above a certain size be gang blocks.
This option is used by the test suite to facilitate testing.
.
.It Sy metaslab_force_ganging_pct Ns = Ns Sy 3 Ns % Pq uint
For blocks that could be forced to be a gang block (due to
.Sy metaslab_force_ganging ) ,
force this many of them to be gang blocks.
.
.It Sy brt_zap_prefetch Ns = Ns Sy 1 Ns | Ns 0 Pq int
Controls prefetching BRT records for blocks which are going to be cloned.
.
.It Sy brt_zap_default_bs Ns = Ns Sy 12 Po 4 KiB Pc Pq int
Default BRT ZAP data block size as a power of 2. Note that changing this after
creating a BRT on the pool will not affect existing BRTs, only newly created
ones.
.
.It Sy brt_zap_default_ibs Ns = Ns Sy 12 Po 4 KiB Pc Pq int
Default BRT ZAP indirect block size as a power of 2. Note that changing this
after creating a BRT on the pool will not affect existing BRTs, only newly
created ones.
.
.It Sy ddt_zap_default_bs Ns = Ns Sy 15 Po 32 KiB Pc Pq int
Default DDT ZAP data block size as a power of 2. Note that changing this after
creating a DDT on the pool will not affect existing DDTs, only newly created
ones.
.
.It Sy ddt_zap_default_ibs Ns = Ns Sy 15 Po 32 KiB Pc Pq int
Default DDT ZAP indirect block size as a power of 2. Note that changing this
after creating a DDT on the pool will not affect existing DDTs, only newly
created ones.
.
.It Sy zfs_default_bs Ns = Ns Sy 9 Po 512 B Pc Pq int
Default dnode block size as a power of 2.
.
.It Sy zfs_default_ibs Ns = Ns Sy 17 Po 128 KiB Pc Pq int
Default dnode indirect block size as a power of 2.
.
.It Sy zfs_dio_enabled Ns = Ns Sy 0 Ns | Ns 1 Pq int
Enable Direct I/O.
If this setting is 0, then all I/O requests will be directed through the ARC
acting as though the dataset property
.Sy direct
was set to
.Sy disabled .
.
.It Sy zfs_history_output_max Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq u64
When attempting to log an output nvlist of an ioctl in the on-disk history,
the output will not be stored if it is larger than this size (in bytes).
This must be less than
.Sy DMU_MAX_ACCESS Pq 64 MiB .
This applies primarily to
.Fn zfs_ioc_channel_program Pq cf. Xr zfs-program 8 .
.
.It Sy zfs_keep_log_spacemaps_at_export Ns = Ns Sy 0 Ns | Ns 1 Pq int
Prevent log spacemaps from being destroyed during pool exports and destroys.
.
.It Sy zfs_metaslab_segment_weight_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enable/disable segment-based metaslab selection.
.
.It Sy zfs_metaslab_switch_threshold Ns = Ns Sy 2 Pq int
When using segment-based metaslab selection, continue allocating
from the active metaslab until this option's
worth of buckets have been exhausted.
.
.It Sy metaslab_debug_load Ns = Ns Sy 0 Ns | Ns 1 Pq int
Load all metaslabs during pool import.
.
.It Sy metaslab_debug_unload Ns = Ns Sy 0 Ns | Ns 1 Pq int
Prevent metaslabs from being unloaded.
.
.It Sy metaslab_fragmentation_factor_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enable use of the fragmentation metric in computing metaslab weights.
.
.It Sy metaslab_df_max_search Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq uint
Maximum distance to search forward from the last offset.
Without this limit, fragmented pools can see
.Em >100`000
iterations and
.Fn metaslab_block_picker
becomes the performance limiting factor on high-performance storage.
.Pp
With the default setting of
.Sy 16 MiB ,
we typically see less than
.Em 500
iterations, even with very fragmented
.Sy ashift Ns = Ns Sy 9
pools.
The maximum number of iterations possible is
.Sy metaslab_df_max_search / 2^(ashift+1) .
With the default setting of
.Sy 16 MiB
this is
.Em 16*1024 Pq with Sy ashift Ns = Ns Sy 9
or
.Em 2*1024 Pq with Sy ashift Ns = Ns Sy 12 .
.
.It Sy metaslab_df_use_largest_segment Ns = Ns Sy 0 Ns | Ns 1 Pq int
If not searching forward (due to
.Sy metaslab_df_max_search , metaslab_df_free_pct ,
.No or Sy metaslab_df_alloc_threshold ) ,
this tunable controls which segment is used.
If set, we will use the largest free segment.
If unset, we will use a segment of at least the requested size.
.
.It Sy zfs_metaslab_max_size_cache_sec Ns = Ns Sy 3600 Ns s Po 1 hour Pc Pq u64
When we unload a metaslab, we cache the size of the largest free chunk.
We use that cached size to determine whether or not to load a metaslab
for a given allocation.
As more frees accumulate in that metaslab while it's unloaded,
the cached max size becomes less and less accurate.
After a number of seconds controlled by this tunable,
we stop considering the cached max size and start
considering only the histogram instead.
.
.It Sy zfs_metaslab_mem_limit Ns = Ns Sy 25 Ns % Pq uint
When we are loading a new metaslab, we check the amount of memory being used
to store metaslab range trees.
If it is over a threshold, we attempt to unload the least recently used metaslab
to prevent the system from clogging all of its memory with range trees.
This tunable sets the percentage of total system memory that is the threshold.
.
.It Sy zfs_metaslab_try_hard_before_gang Ns = Ns Sy 0 Ns | Ns 1 Pq int
.Bl -item -compact
.It
If unset, we will first try normal allocation.
.It
If that fails then we will do a gang allocation.
.It
If that fails then we will do a "try hard" gang allocation.
.It
If that fails then we will have a multi-layer gang block.
.El
.Pp
.Bl -item -compact
.It
If set, we will first try normal allocation.
.It
If that fails then we will do a "try hard" allocation.
.It
If that fails we will do a gang allocation.
.It
If that fails we will do a "try hard" gang allocation.
.It
If that fails then we will have a multi-layer gang block.
.El
.
.It Sy zfs_metaslab_find_max_tries Ns = Ns Sy 100 Pq uint
When not trying hard, we only consider this number of the best metaslabs.
This improves performance, especially when there are many metaslabs per vdev
and the allocation can't actually be satisfied
(so we would otherwise iterate all metaslabs).
.
.It Sy zfs_vdev_default_ms_count Ns = Ns Sy 200 Pq uint
When a vdev is added, target this number of metaslabs per top-level vdev.
.
.It Sy zfs_vdev_default_ms_shift Ns = Ns Sy 29 Po 512 MiB Pc Pq uint
Default lower limit for metaslab size.
.
.It Sy zfs_vdev_max_ms_shift Ns = Ns Sy 34 Po 16 GiB Pc Pq uint
Default upper limit for metaslab size.
.
.It Sy zfs_vdev_max_auto_ashift Ns = Ns Sy 14 Pq uint
Maximum ashift used when optimizing for logical \[->] physical sector size on
new
top-level vdevs.
May be increased up to
.Sy ASHIFT_MAX Po 16 Pc ,
but this may negatively impact pool space efficiency.
.
.It Sy zfs_vdev_direct_write_verify Ns = Ns Sy Linux 1 | FreeBSD 0 Pq uint
If non-zero, then a Direct I/O write's checksum will be verified every
time the write is issued and before it is committed to the block pointer.
In the event the checksum is not valid then the I/O operation will return EIO.
This module parameter can be used to detect if the
contents of the users buffer have changed in the process of doing a Direct I/O
write.
It can also help to identify if reported checksum errors are tied to Direct I/O
writes.
Each verify error causes a
.Sy dio_verify_wr
zevent.
Direct Write I/O checksum verify errors can be seen with
.Nm zpool Cm status Fl d .
The default value for this is 1 on Linux, but is 0 for
.Fx
because user pages can be placed under write protection in
.Fx
before the Direct I/O write is issued.
.
.It Sy zfs_vdev_min_auto_ashift Ns = Ns Sy ASHIFT_MIN Po 9 Pc Pq uint
Minimum ashift used when creating new top-level vdevs.
.
.It Sy zfs_vdev_min_ms_count Ns = Ns Sy 16 Pq uint
Minimum number of metaslabs to create in a top-level vdev.
.
.It Sy vdev_validate_skip Ns = Ns Sy 0 Ns | Ns 1 Pq int
Skip label validation steps during pool import.
Changing is not recommended unless you know what you're doing
and are recovering a damaged label.
.
.It Sy zfs_vdev_ms_count_limit Ns = Ns Sy 131072 Po 128k Pc Pq uint
Practical upper limit of total metaslabs per top-level vdev.
.
.It Sy metaslab_preload_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enable metaslab group preloading.
.
.It Sy metaslab_preload_limit Ns = Ns Sy 10 Pq uint
Maximum number of metaslabs per group to preload
.
.It Sy metaslab_preload_pct Ns = Ns Sy 50 Pq uint
Percentage of CPUs to run a metaslab preload taskq
.
.It Sy metaslab_lba_weighting_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Give more weight to metaslabs with lower LBAs,
assuming they have greater bandwidth,
as is typically the case on a modern constant angular velocity disk drive.
.
.It Sy metaslab_unload_delay Ns = Ns Sy 32 Pq uint
After a metaslab is used, we keep it loaded for this many TXGs, to attempt to
reduce unnecessary reloading.
Note that both this many TXGs and
.Sy metaslab_unload_delay_ms
milliseconds must pass before unloading will occur.
.
.It Sy metaslab_unload_delay_ms Ns = Ns Sy 600000 Ns ms Po 10 min Pc Pq uint
After a metaslab is used, we keep it loaded for this many milliseconds,
to attempt to reduce unnecessary reloading.
Note, that both this many milliseconds and
.Sy metaslab_unload_delay
TXGs must pass before unloading will occur.
.
.It Sy reference_history Ns = Ns Sy 3 Pq uint
Maximum reference holders being tracked when reference_tracking_enable is
active.
.It Sy raidz_expand_max_copy_bytes Ns = Ns Sy 160MB Pq ulong
Max amount of memory to use for RAID-Z expansion I/O.
This limits how much I/O can be outstanding at once.
.
.It Sy raidz_expand_max_reflow_bytes Ns = Ns Sy 0 Pq ulong
For testing, pause RAID-Z expansion when reflow amount reaches this value.
.
.It Sy raidz_io_aggregate_rows Ns = Ns Sy 4 Pq ulong
For expanded RAID-Z, aggregate reads that have more rows than this.
.
.It Sy reference_history Ns = Ns Sy 3 Pq int
Maximum reference holders being tracked when reference_tracking_enable is
active.
.
.It Sy reference_tracking_enable Ns = Ns Sy 0 Ns | Ns 1 Pq int
Track reference holders to
.Sy refcount_t
objects (debug builds only).
.
.It Sy send_holes_without_birth_time Ns = Ns Sy 1 Ns | Ns 0 Pq int
When set, the
.Sy hole_birth
optimization will not be used, and all holes will always be sent during a
.Nm zfs Cm send .
This is useful if you suspect your datasets are affected by a bug in
.Sy hole_birth .
.
.It Sy spa_config_path Ns = Ns Pa /etc/zfs/zpool.cache Pq charp
SPA config file.
.
.It Sy spa_asize_inflation Ns = Ns Sy 24 Pq uint
Multiplication factor used to estimate actual disk consumption from the
size of data being written.
The default value is a worst case estimate,
but lower values may be valid for a given pool depending on its configuration.
Pool administrators who understand the factors involved
may wish to specify a more realistic inflation factor,
particularly if they operate close to quota or capacity limits.
.
.It Sy spa_load_print_vdev_tree Ns = Ns Sy 0 Ns | Ns 1 Pq int
Whether to print the vdev tree in the debugging message buffer during pool
import.
.
.It Sy spa_load_verify_data Ns = Ns Sy 1 Ns | Ns 0 Pq int
Whether to traverse data blocks during an "extreme rewind"
.Pq Fl X
import.
.Pp
An extreme rewind import normally performs a full traversal of all
blocks in the pool for verification.
If this parameter is unset, the traversal skips non-metadata blocks.
It can be toggled once the
import has started to stop or start the traversal of non-metadata blocks.
.
.It Sy spa_load_verify_metadata  Ns = Ns Sy 1 Ns | Ns 0 Pq int
Whether to traverse blocks during an "extreme rewind"
.Pq Fl X
pool import.
.Pp
An extreme rewind import normally performs a full traversal of all
blocks in the pool for verification.
If this parameter is unset, the traversal is not performed.
It can be toggled once the import has started to stop or start the traversal.
.
.It Sy spa_load_verify_shift Ns = Ns Sy 4 Po 1/16th Pc Pq uint
Sets the maximum number of bytes to consume during pool import to the log2
fraction of the target ARC size.
.
.It Sy spa_slop_shift Ns = Ns Sy 5 Po 1/32nd Pc Pq int
Normally, we don't allow the last
.Sy 3.2% Pq Sy 1/2^spa_slop_shift
of space in the pool to be consumed.
This ensures that we don't run the pool completely out of space,
due to unaccounted changes (e.g. to the MOS).
It also limits the worst-case time to allocate space.
If we have less than this amount of free space,
most ZPL operations (e.g. write, create) will return
.Sy ENOSPC .
.
.It Sy spa_num_allocators Ns = Ns Sy 4 Pq int
Determines the number of block allocators to use per spa instance.
Capped by the number of actual CPUs in the system via
.Sy spa_cpus_per_allocator .
.Pp
Note that setting this value too high could result in performance
degradation and/or excess fragmentation.
Set value only applies to pools imported/created after that.
.
.It Sy spa_cpus_per_allocator Ns = Ns Sy 4 Pq int
Determines the minimum number of CPUs in a system for block allocator
per spa instance.
Set value only applies to pools imported/created after that.
.
.It Sy spa_upgrade_errlog_limit Ns = Ns Sy 0 Pq uint
Limits the number of on-disk error log entries that will be converted to the
new format when enabling the
.Sy head_errlog
feature.
The default is to convert all log entries.
.
.It Sy vdev_removal_max_span Ns = Ns Sy 32768 Ns B Po 32 KiB Pc Pq uint
During top-level vdev removal, chunks of data are copied from the vdev
which may include free space in order to trade bandwidth for IOPS.
This parameter determines the maximum span of free space, in bytes,
which will be included as "unnecessary" data in a chunk of copied data.
.Pp
The default value here was chosen to align with
.Sy zfs_vdev_read_gap_limit ,
which is a similar concept when doing
regular reads (but there's no reason it has to be the same).
.
.It Sy vdev_file_logical_ashift Ns = Ns Sy 9 Po 512 B Pc Pq u64
Logical ashift for file-based devices.
.
.It Sy vdev_file_physical_ashift Ns = Ns Sy 9 Po 512 B Pc Pq u64
Physical ashift for file-based devices.
.
.It Sy zap_iterate_prefetch Ns = Ns Sy 1 Ns | Ns 0 Pq int
If set, when we start iterating over a ZAP object,
prefetch the entire object (all leaf blocks).
However, this is limited by
.Sy dmu_prefetch_max .
.
.It Sy zap_micro_max_size Ns = Ns Sy 131072 Ns B Po 128 KiB Pc Pq int
Maximum micro ZAP size.
A "micro" ZAP is upgraded to a "fat" ZAP once it grows beyond the specified
size.
Sizes higher than 128KiB will be clamped to 128KiB unless the
.Sy large_microzap
feature is enabled.
.
.It Sy zap_shrink_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
If set, adjacent empty ZAP blocks will be collapsed, reducing disk space.
.
.It Sy zfetch_min_distance Ns = Ns Sy 4194304 Ns B Po 4 MiB Pc Pq uint
Min bytes to prefetch per stream.
Prefetch distance starts from the demand access size and quickly grows to
this value, doubling on each hit.
After that it may grow further by 1/8 per hit, but only if some prefetch
since last time haven't completed in time to satisfy demand request, i.e.
prefetch depth didn't cover the read latency or the pool got saturated.
.
.It Sy zfetch_max_distance Ns = Ns Sy 67108864 Ns B Po 64 MiB Pc Pq uint
Max bytes to prefetch per stream.
.
.It Sy zfetch_max_idistance Ns = Ns Sy 67108864 Ns B Po 64 MiB Pc Pq uint
Max bytes to prefetch indirects for per stream.
.
.It Sy zfetch_max_reorder Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq uint
Requests within this byte distance from the current prefetch stream position
are considered parts of the stream, reordered due to parallel processing.
Such requests do not advance the stream position immediately unless
.Sy zfetch_hole_shift
fill threshold is reached, but saved to fill holes in the stream later.
.
.It Sy zfetch_max_streams Ns = Ns Sy 8 Pq uint
Max number of streams per zfetch (prefetch streams per file).
.
.It Sy zfetch_min_sec_reap Ns = Ns Sy 1 Pq uint
Min time before inactive prefetch stream can be reclaimed
.
.It Sy zfetch_max_sec_reap Ns = Ns Sy 2 Pq uint
Max time before inactive prefetch stream can be deleted
.
.It Sy zfs_abd_scatter_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enables ARC from using scatter/gather lists and forces all allocations to be
linear in kernel memory.
Disabling can improve performance in some code paths
at the expense of fragmented kernel memory.
.
.It Sy zfs_abd_scatter_max_order Ns = Ns Sy MAX_ORDER\-1 Pq uint
Maximum number of consecutive memory pages allocated in a single block for
scatter/gather lists.
.Pp
The value of
.Sy MAX_ORDER
depends on kernel configuration.
.
.It Sy zfs_abd_scatter_min_size Ns = Ns Sy 1536 Ns B Po 1.5 KiB Pc Pq uint
This is the minimum allocation size that will use scatter (page-based) ABDs.
Smaller allocations will use linear ABDs.
.
.It Sy zfs_arc_dnode_limit Ns = Ns Sy 0 Ns B Pq u64
When the number of bytes consumed by dnodes in the ARC exceeds this number of
bytes, try to unpin some of it in response to demand for non-metadata.
This value acts as a ceiling to the amount of dnode metadata, and defaults to
.Sy 0 ,
which indicates that a percent which is based on
.Sy zfs_arc_dnode_limit_percent
of the ARC meta buffers that may be used for dnodes.
.It Sy zfs_arc_dnode_limit_percent Ns = Ns Sy 10 Ns % Pq u64
Percentage that can be consumed by dnodes of ARC meta buffers.
.Pp
See also
.Sy zfs_arc_dnode_limit ,
which serves a similar purpose but has a higher priority if nonzero.
.
.It Sy zfs_arc_dnode_reduce_percent Ns = Ns Sy 10 Ns % Pq u64
Percentage of ARC dnodes to try to scan in response to demand for non-metadata
when the number of bytes consumed by dnodes exceeds
.Sy zfs_arc_dnode_limit .
.
.It Sy zfs_arc_average_blocksize Ns = Ns Sy 8192 Ns B Po 8 KiB Pc Pq uint
The ARC's buffer hash table is sized based on the assumption of an average
block size of this value.
This works out to roughly 1 MiB of hash table per 1 GiB of physical memory
with 8-byte pointers.
For configurations with a known larger average block size,
this value can be increased to reduce the memory footprint.
.
.It Sy zfs_arc_eviction_pct Ns = Ns Sy 200 Ns % Pq uint
When
.Fn arc_is_overflowing ,
.Fn arc_get_data_impl
waits for this percent of the requested amount of data to be evicted.
For example, by default, for every
.Em 2 KiB
that's evicted,
.Em 1 KiB
of it may be "reused" by a new allocation.
Since this is above
.Sy 100 Ns % ,
it ensures that progress is made towards getting
.Sy arc_size No under Sy arc_c .
Since this is finite, it ensures that allocations can still happen,
even during the potentially long time that
.Sy arc_size No is more than Sy arc_c .
.
.It Sy zfs_arc_evict_batch_limit Ns = Ns Sy 10 Pq uint
Number ARC headers to evict per sub-list before proceeding to another sub-list.
This batch-style operation prevents entire sub-lists from being evicted at once
but comes at a cost of additional unlocking and locking.
.
.It Sy zfs_arc_evict_threads Ns = Ns Sy 0 Pq int
Sets the number of ARC eviction threads to be used.
.Pp
If set greater than 0, ZFS will dedicate up to that many threads to ARC
eviction.
Each thread will process one sub-list at a time,
until the eviction target is reached or all sub-lists have been processed.
When set to 0, ZFS will compute a reasonable number of eviction threads based
on the number of CPUs.
.TS
box;
lb l l .
	CPUs	Threads
_
	1-4	1
	5-8	2
	9-15	3
	16-31	4
	32-63	6
	64-95	8
	96-127	9
	128-160	11
	160-191	12
	192-223	13
	224-255	14
	256+	16
.TE
.Pp
More threads may improve the responsiveness of ZFS to memory pressure.
This can be important for performance when eviction from the ARC becomes
a bottleneck for reads and writes.
.Pp
This parameter can only be set at module load time.
.
.It Sy zfs_arc_grow_retry Ns = Ns Sy 0 Ns s Pq uint
If set to a non zero value, it will replace the
.Sy arc_grow_retry
value with this value.
The
.Sy arc_grow_retry
.No value Pq default Sy 5 Ns s
is the number of seconds the ARC will wait before
trying to resume growth after a memory pressure event.
.
.It Sy zfs_arc_lotsfree_percent Ns = Ns Sy 10 Ns % Pq int
Throttle I/O when free system memory drops below this percentage of total
system memory.
Setting this value to
.Sy 0
will disable the throttle.
.
.It Sy zfs_arc_max Ns = Ns Sy 0 Ns B Pq u64
Max size of ARC in bytes.
If
.Sy 0 ,
then the max size of ARC is determined by the amount of system memory installed.
The larger of
.Sy all_system_memory No \- Sy 1 GiB
and
.Sy 5/8 No \(mu Sy all_system_memory
will be used as the limit.
This value must be at least
.Sy 67108864 Ns B Pq 64 MiB .
.Pp
This value can be changed dynamically, with some caveats.
It cannot be set back to
.Sy 0
while running, and reducing it below the current ARC size will not cause
the ARC to shrink without memory pressure to induce shrinking.
.
.It Sy zfs_arc_meta_balance Ns = Ns Sy 500 Pq uint
Balance between metadata and data on ghost hits.
Values above 100 increase metadata caching by proportionally reducing effect
of ghost data hits on target data/metadata rate.
.
.It Sy zfs_arc_min Ns = Ns Sy 0 Ns B Pq u64
Min size of ARC in bytes.
.No If set to Sy 0 , arc_c_min
will default to consuming the larger of
.Sy 32 MiB
and
.Sy all_system_memory No / Sy 32 .
.
.It Sy zfs_arc_min_prefetch_ms Ns = Ns Sy 0 Ns ms Ns Po Ns ≡ Ns 1s Pc Pq uint
Minimum time prefetched blocks are locked in the ARC.
.
.It Sy zfs_arc_min_prescient_prefetch_ms Ns = Ns Sy 0 Ns ms Ns Po Ns ≡ Ns 6s Pc Pq uint
Minimum time "prescient prefetched" blocks are locked in the ARC.
These blocks are meant to be prefetched fairly aggressively ahead of
the code that may use them.
.
.It Sy zfs_arc_prune_task_threads Ns = Ns Sy 1 Pq int
Number of arc_prune threads.
.Fx
does not need more than one.
Linux may theoretically use one per mount point up to number of CPUs,
but that was not proven to be useful.
.
.It Sy zfs_max_missing_tvds Ns = Ns Sy 0 Pq int
Number of missing top-level vdevs which will be allowed during
pool import (only in read-only mode).
.
.It Sy zfs_max_nvlist_src_size Ns = Sy 0 Pq u64
Maximum size in bytes allowed to be passed as
.Sy zc_nvlist_src_size
for ioctls on
.Pa /dev/zfs .
This prevents a user from causing the kernel to allocate
an excessive amount of memory.
When the limit is exceeded, the ioctl fails with
.Sy EINVAL
and a description of the error is sent to the
.Pa zfs-dbgmsg
log.
This parameter should not need to be touched under normal circumstances.
If
.Sy 0 ,
equivalent to a quarter of the user-wired memory limit under
.Fx
and to
.Sy 134217728 Ns B Pq 128 MiB
under Linux.
.
.It Sy zfs_multilist_num_sublists Ns = Ns Sy 0 Pq uint
To allow more fine-grained locking, each ARC state contains a series
of lists for both data and metadata objects.
Locking is performed at the level of these "sub-lists".
This parameters controls the number of sub-lists per ARC state,
and also applies to other uses of the multilist data structure.
.Pp
If
.Sy 0 ,
equivalent to the greater of the number of online CPUs and
.Sy 4 .
.
.It Sy zfs_arc_overflow_shift Ns = Ns Sy 8 Pq int
The ARC size is considered to be overflowing if it exceeds the current
ARC target size
.Pq Sy arc_c
by thresholds determined by this parameter.
Exceeding by
.Sy ( arc_c No >> Sy zfs_arc_overflow_shift ) No / Sy 2
starts ARC reclamation process.
If that appears insufficient, exceeding by
.Sy ( arc_c No >> Sy zfs_arc_overflow_shift ) No \(mu Sy 1.5
blocks new buffer allocation until the reclaim thread catches up.
Started reclamation process continues till ARC size returns below the
target size.
.Pp
The default value of
.Sy 8
causes the ARC to start reclamation if it exceeds the target size by
.Em 0.2%
of the target size, and block allocations by
.Em 0.6% .
.
.It Sy zfs_arc_shrink_shift Ns = Ns Sy 0 Pq uint
If nonzero, this will update
.Sy arc_shrink_shift Pq default Sy 7
with the new value.
.
.It Sy zfs_arc_pc_percent Ns = Ns Sy 0 Ns % Po off Pc Pq uint
Percent of pagecache to reclaim ARC to.
.Pp
This tunable allows the ZFS ARC to play more nicely
with the kernel's LRU pagecache.
It can guarantee that the ARC size won't collapse under scanning
pressure on the pagecache, yet still allows the ARC to be reclaimed down to
.Sy zfs_arc_min
if necessary.
This value is specified as percent of pagecache size (as measured by
.Sy NR_ACTIVE_FILE
+
.Sy NR_INACTIVE_FILE ) ,
where that percent may exceed
.Sy 100 .
This
only operates during memory pressure/reclaim.
.
.It Sy zfs_arc_shrinker_limit Ns = Ns Sy 0 Pq int
This is a limit on how many pages the ARC shrinker makes available for
eviction in response to one page allocation attempt.
Note that in practice, the kernel's shrinker can ask us to evict
up to about four times this for one allocation attempt.
To reduce OOM risk, this limit is applied for kswapd reclaims only.
.Pp
For example a value of
.Sy 10000 Pq in practice, Em 160 MiB No per allocation attempt with 4 KiB pages
limits the amount of time spent attempting to reclaim ARC memory to
less than 100 ms per allocation attempt,
even with a small average compressed block size of ~8 KiB.
.Pp
The parameter can be set to 0 (zero) to disable the limit,
and only applies on Linux.
.
.It Sy zfs_arc_shrinker_seeks Ns = Ns Sy 2 Pq int
Relative cost of ARC eviction on Linux, AKA number of seeks needed to
restore evicted page.
Bigger values make ARC more precious and evictions smaller, comparing to
other kernel subsystems.
Value of 4 means parity with page cache.
.
.It Sy zfs_arc_sys_free Ns = Ns Sy 0 Ns B Pq u64
The target number of bytes the ARC should leave as free memory on the system.
If zero, equivalent to the bigger of
.Sy 512 KiB No and Sy all_system_memory/64 .
.
.It Sy zfs_autoimport_disable Ns = Ns Sy 1 Ns | Ns 0 Pq int
Disable pool import at module load by ignoring the cache file
.Pq Sy spa_config_path .
.
.It Sy zfs_checksum_events_per_second Ns = Ns Sy 20 Ns /s Pq uint
Rate limit checksum events to this many per second.
Note that this should not be set below the ZED thresholds
(currently 10 checksums over 10 seconds)
or else the daemon may not trigger any action.
.
.It Sy zfs_commit_timeout_pct Ns = Ns Sy 10 Ns % Pq uint
This controls the amount of time that a ZIL block (lwb) will remain "open"
when it isn't "full", and it has a thread waiting for it to be committed to
stable storage.
The timeout is scaled based on a percentage of the last lwb
latency to avoid significantly impacting the latency of each individual
transaction record (itx).
.
.It Sy zfs_condense_indirect_commit_entry_delay_ms Ns = Ns Sy 0 Ns ms Pq int
Vdev indirection layer (used for device removal) sleeps for this many
milliseconds during mapping generation.
Intended for use with the test suite to throttle vdev removal speed.
.
.It Sy zfs_condense_indirect_obsolete_pct Ns = Ns Sy 25 Ns % Pq uint
Minimum percent of obsolete bytes in vdev mapping required to attempt to
condense
.Pq see Sy zfs_condense_indirect_vdevs_enable .
Intended for use with the test suite
to facilitate triggering condensing as needed.
.
.It Sy zfs_condense_indirect_vdevs_enable Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enable condensing indirect vdev mappings.
When set, attempt to condense indirect vdev mappings
if the mapping uses more than
.Sy zfs_condense_min_mapping_bytes
bytes of memory and if the obsolete space map object uses more than
.Sy zfs_condense_max_obsolete_bytes
bytes on-disk.
The condensing process is an attempt to save memory by removing obsolete
mappings.
.
.It Sy zfs_condense_max_obsolete_bytes Ns = Ns Sy 1073741824 Ns B Po 1 GiB Pc Pq u64
Only attempt to condense indirect vdev mappings if the on-disk size
of the obsolete space map object is greater than this number of bytes
.Pq see Sy zfs_condense_indirect_vdevs_enable .
.
.It Sy zfs_condense_min_mapping_bytes Ns = Ns Sy 131072 Ns B Po 128 KiB Pc Pq u64
Minimum size vdev mapping to attempt to condense
.Pq see Sy zfs_condense_indirect_vdevs_enable .
.
.It Sy zfs_dbgmsg_enable Ns = Ns Sy 1 Ns | Ns 0 Pq int
Internally ZFS keeps a small log to facilitate debugging.
The log is enabled by default, and can be disabled by unsetting this option.
The contents of the log can be accessed by reading
.Pa /proc/spl/kstat/zfs/dbgmsg .
Writing
.Sy 0
to the file clears the log.
.Pp
This setting does not influence debug prints due to
.Sy zfs_flags .
.
.It Sy zfs_dbgmsg_maxsize Ns = Ns Sy 4194304 Ns B Po 4 MiB Pc Pq uint
Maximum size of the internal ZFS debug log.
.
.It Sy zfs_dbuf_state_index Ns = Ns Sy 0 Pq int
Historically used for controlling what reporting was available under
.Pa /proc/spl/kstat/zfs .
No effect.
.
.It Sy zfs_deadman_checktime_ms Ns = Ns Sy 60000 Ns ms Po 1 min Pc Pq u64
Check time in milliseconds.
This defines the frequency at which we check for hung I/O requests
and potentially invoke the
.Sy zfs_deadman_failmode
behavior.
.
.It Sy zfs_deadman_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
When a pool sync operation takes longer than
.Sy zfs_deadman_synctime_ms ,
or when an individual I/O operation takes longer than
.Sy zfs_deadman_ziotime_ms ,
then the operation is considered to be "hung".
If
.Sy zfs_deadman_enabled
is set, then the deadman behavior is invoked as described by
.Sy zfs_deadman_failmode .
By default, the deadman is enabled and set to
.Sy wait
which results in "hung" I/O operations only being logged.
The deadman is automatically disabled when a pool gets suspended.
.
.It Sy zfs_deadman_events_per_second Ns = Ns Sy 1 Ns /s Pq int
Rate limit deadman zevents (which report hung I/O operations) to this many per
second.
.
.It Sy zfs_deadman_failmode Ns = Ns Sy wait Pq charp
Controls the failure behavior when the deadman detects a "hung" I/O operation.
Valid values are:
.Bl -tag -compact -offset 4n -width "continue"
.It Sy wait
Wait for a "hung" operation to complete.
For each "hung" operation a "deadman" event will be posted
describing that operation.
.It Sy continue
Attempt to recover from a "hung" operation by re-dispatching it
to the I/O pipeline if possible.
.It Sy panic
Panic the system.
This can be used to facilitate automatic fail-over
to a properly configured fail-over partner.
.El
.
.It Sy zfs_deadman_synctime_ms Ns = Ns Sy 600000 Ns ms Po 10 min Pc Pq u64
Interval in milliseconds after which the deadman is triggered and also
the interval after which a pool sync operation is considered to be "hung".
Once this limit is exceeded the deadman will be invoked every
.Sy zfs_deadman_checktime_ms
milliseconds until the pool sync completes.
.
.It Sy zfs_deadman_ziotime_ms Ns = Ns Sy 300000 Ns ms Po 5 min Pc Pq u64
Interval in milliseconds after which the deadman is triggered and an
individual I/O operation is considered to be "hung".
As long as the operation remains "hung",
the deadman will be invoked every
.Sy zfs_deadman_checktime_ms
milliseconds until the operation completes.
.
.It Sy zfs_dedup_prefetch Ns = Ns Sy 0 Ns | Ns 1 Pq int
Enable prefetching dedup-ed blocks which are going to be freed.
.
.It Sy zfs_dedup_log_flush_min_time_ms Ns = Ns Sy 1000 Ns Pq uint
Minimum time to spend on dedup log flush each transaction.
.Pp
At least this long will be spent flushing dedup log entries each transaction,
up to
.Sy zfs_txg_timeout .
This occurs even if doing so would delay the transaction, that is, other IO
completes under this time.
.
.It Sy zfs_dedup_log_flush_entries_min Ns = Ns Sy 100 Ns Pq uint
Flush at least this many entries each transaction.
.Pp
OpenZFS will flush a fraction of the log every TXG, to keep the size
proportional to the ingest rate (see
.Sy zfs_dedup_log_flush_txgs ) .
This sets the minimum for that estimate, which prevents the backlog from
completely draining if the ingest rate falls.
Raising it can force OpenZFS to flush more aggressively, reducing the backlog
to zero more quickly, but can make it less able to back off if log
flushing would compete with other IO too much.
.
.It Sy zfs_dedup_log_flush_entries_max Ns = Ns Sy UINT_MAX Ns Pq uint
Flush at most this many entries each transaction.
.Pp
Mostly used for debugging purposes.
.It Sy zfs_dedup_log_flush_txgs Ns = Ns Sy 100 Ns Pq uint
Target number of TXGs to process the whole dedup log.
.Pp
Every TXG, OpenZFS will process the inverse of this number times the size
of the DDT backlog.
This will keep the backlog at a size roughly equal to the ingest rate
times this value.
This offers a balance between a more efficient DDT log, with better
aggregation, and shorter import times, which increase as the size of the
DDT log increases.
Increasing this value will result in a more efficient DDT log, but longer
import times.
.It Sy zfs_dedup_log_cap Ns = Ns Sy UINT_MAX Ns Pq uint
Soft cap for the size of the current dedup log.
.Pp
If the log is larger than this size, we increase the aggressiveness of
the flushing to try to bring it back down to the soft cap.
Setting it will reduce import times, but will reduce the efficiency of
the DDT log, increasing the expected number of IOs required to flush the same
amount of data.
.It Sy zfs_dedup_log_hard_cap Ns = Ns Sy 0 Ns | Ns 1 Pq uint
Whether to treat the log cap as a firm cap or not.
.Pp
When set to 0 (the default), the
.Sy zfs_dedup_log_cap
will increase the maximum number of log entries we flush in a given txg.
This will bring the backlog size down towards the cap, but not at the expense
of making TXG syncs take longer.
If this is set to 1, the cap acts more like a hard cap than a soft cap; it will
also increase the minimum number of log entries we flush per TXG.
Enabling it will reduce worst-case import times, at the cost of increased TXG
sync times.
.It Sy zfs_dedup_log_flush_flow_rate_txgs Ns = Ns Sy 10 Ns Pq uint
Number of transactions to use to compute the flow rate.
.Pp
OpenZFS will estimate number of entries changed (ingest rate), number of entries
flushed (flush rate) and time spent flushing (flush time rate) and combining
these into an overall "flow rate".
It will use an exponential weighted moving average over some number of recent
transactions to compute these rates.
This sets the number of transactions to compute these averages over.
Setting it higher can help to smooth out the flow rate in the face of spiky
workloads, but will take longer for the flow rate to adjust to a sustained
change in the ingress rate.
.
.It Sy zfs_dedup_log_txg_max Ns = Ns Sy 8 Ns Pq uint
Max transactions to before starting to flush dedup logs.
.Pp
OpenZFS maintains two dedup logs, one receiving new changes, one flushing.
If there is nothing to flush, it will accumulate changes for no more than this
many transactions before switching the logs and starting to flush entries out.
.
.It Sy zfs_dedup_log_mem_max Ns = Ns Sy 0 Ns Pq u64
Max memory to use for dedup logs.
.Pp
OpenZFS will spend no more than this much memory on maintaining the in-memory
dedup log.
Flushing will begin when around half this amount is being spent on logs.
The default value of
.Sy 0
will cause it to be set by
.Sy zfs_dedup_log_mem_max_percent
instead.
.
.It Sy zfs_dedup_log_mem_max_percent Ns = Ns Sy 1 Ns % Pq uint
Max memory to use for dedup logs, as a percentage of total memory.
.Pp
If
.Sy zfs_dedup_log_mem_max
is not set, it will be initialized as a percentage of the total memory in the
system.
.
.It Sy zfs_delay_min_dirty_percent Ns = Ns Sy 60 Ns % Pq uint
Start to delay each transaction once there is this amount of dirty data,
expressed as a percentage of
.Sy zfs_dirty_data_max .
This value should be at least
.Sy zfs_vdev_async_write_active_max_dirty_percent .
.No See Sx ZFS TRANSACTION DELAY .
.
.It Sy zfs_delay_scale Ns = Ns Sy 500000 Pq int
This controls how quickly the transaction delay approaches infinity.
Larger values cause longer delays for a given amount of dirty data.
.Pp
For the smoothest delay, this value should be about 1 billion divided
by the maximum number of operations per second.
This will smoothly handle between ten times and a tenth of this number.
.No See Sx ZFS TRANSACTION DELAY .
.Pp
.Sy zfs_delay_scale No \(mu Sy zfs_dirty_data_max Em must No be smaller than Sy 2^64 .
.
.It Sy zfs_dio_write_verify_events_per_second Ns = Ns Sy 20 Ns /s Pq uint
Rate limit Direct I/O write verify events to this many per second.
.
.It Sy zfs_disable_ivset_guid_check Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disables requirement for IVset GUIDs to be present and match when doing a raw
receive of encrypted datasets.
Intended for users whose pools were created with
OpenZFS pre-release versions and now have compatibility issues.
.
.It Sy zfs_key_max_salt_uses Ns = Ns Sy 400000000 Po 4*10^8 Pc Pq ulong
Maximum number of uses of a single salt value before generating a new one for
encrypted datasets.
The default value is also the maximum.
.
.It Sy zfs_object_mutex_size Ns = Ns Sy 64 Pq uint
Size of the znode hashtable used for holds.
.Pp
Due to the need to hold locks on objects that may not exist yet, kernel mutexes
are not created per-object and instead a hashtable is used where collisions
will result in objects waiting when there is not actually contention on the
same object.
.
.It Sy zfs_slow_io_events_per_second Ns = Ns Sy 20 Ns /s Pq int
Rate limit delay zevents (which report slow I/O operations) to this many per
second.
.
.It Sy zfs_unflushed_max_mem_amt Ns = Ns Sy 1073741824 Ns B Po 1 GiB Pc Pq u64
Upper-bound limit for unflushed metadata changes to be held by the
log spacemap in memory, in bytes.
.
.It Sy zfs_unflushed_max_mem_ppm Ns = Ns Sy 1000 Ns ppm Po 0.1% Pc Pq u64
Part of overall system memory that ZFS allows to be used
for unflushed metadata changes by the log spacemap, in millionths.
.
.It Sy zfs_unflushed_log_block_max Ns = Ns Sy 131072 Po 128k Pc Pq u64
Describes the maximum number of log spacemap blocks allowed for each pool.
The default value means that the space in all the log spacemaps
can add up to no more than
.Sy 131072
blocks (which means
.Em 16 GiB
of logical space before compression and ditto blocks,
assuming that blocksize is
.Em 128 KiB ) .
.Pp
This tunable is important because it involves a trade-off between import
time after an unclean export and the frequency of flushing metaslabs.
The higher this number is, the more log blocks we allow when the pool is
active which means that we flush metaslabs less often and thus decrease
the number of I/O operations for spacemap updates per TXG.
At the same time though, that means that in the event of an unclean export,
there will be more log spacemap blocks for us to read, inducing overhead
in the import time of the pool.
The lower the number, the amount of flushing increases, destroying log
blocks quicker as they become obsolete faster, which leaves less blocks
to be read during import time after a crash.
.Pp
Each log spacemap block existing during pool import leads to approximately
one extra logical I/O issued.
This is the reason why this tunable is exposed in terms of blocks rather
than space used.
.
.It Sy zfs_unflushed_log_block_min Ns = Ns Sy 1000 Pq u64
If the number of metaslabs is small and our incoming rate is high,
we could get into a situation that we are flushing all our metaslabs every TXG.
Thus we always allow at least this many log blocks.
.
.It Sy zfs_unflushed_log_block_pct Ns = Ns Sy 400 Ns % Pq u64
Tunable used to determine the number of blocks that can be used for
the spacemap log, expressed as a percentage of the total number of
unflushed metaslabs in the pool.
.
.It Sy zfs_unflushed_log_txg_max Ns = Ns Sy 1000 Pq u64
Tunable limiting maximum time in TXGs any metaslab may remain unflushed.
It effectively limits maximum number of unflushed per-TXG spacemap logs
that need to be read after unclean pool export.
.
.It Sy zfs_unlink_suspend_progress Ns = Ns Sy 0 Ns | Ns 1 Pq uint
When enabled, files will not be asynchronously removed from the list of pending
unlinks and the space they consume will be leaked.
Once this option has been disabled and the dataset is remounted,
the pending unlinks will be processed and the freed space returned to the pool.
This option is used by the test suite.
.
.It Sy zfs_delete_blocks Ns = Ns Sy 20480 Pq ulong
This is the used to define a large file for the purposes of deletion.
Files containing more than
.Sy zfs_delete_blocks
will be deleted asynchronously, while smaller files are deleted synchronously.
Decreasing this value will reduce the time spent in an
.Xr unlink 2
system call, at the expense of a longer delay before the freed space is
available.
This only applies on Linux.
.
.It Sy zfs_dirty_data_max Ns = Pq int
Determines the dirty space limit in bytes.
Once this limit is exceeded, new writes are halted until space frees up.
This parameter takes precedence over
.Sy zfs_dirty_data_max_percent .
.No See Sx ZFS TRANSACTION DELAY .
.Pp
Defaults to
.Sy physical_ram/10 ,
capped at
.Sy zfs_dirty_data_max_max .
.
.It Sy zfs_dirty_data_max_max Ns = Pq int
Maximum allowable value of
.Sy zfs_dirty_data_max ,
expressed in bytes.
This limit is only enforced at module load time, and will be ignored if
.Sy zfs_dirty_data_max
is later changed.
This parameter takes precedence over
.Sy zfs_dirty_data_max_max_percent .
.No See Sx ZFS TRANSACTION DELAY .
.Pp
Defaults to
.Sy min(physical_ram/4, 4GiB) ,
or
.Sy min(physical_ram/4, 1GiB)
for 32-bit systems.
.
.It Sy zfs_dirty_data_max_max_percent Ns = Ns Sy 25 Ns % Pq uint
Maximum allowable value of
.Sy zfs_dirty_data_max ,
expressed as a percentage of physical RAM.
This limit is only enforced at module load time, and will be ignored if
.Sy zfs_dirty_data_max
is later changed.
The parameter
.Sy zfs_dirty_data_max_max
takes precedence over this one.
.No See Sx ZFS TRANSACTION DELAY .
.
.It Sy zfs_dirty_data_max_percent Ns = Ns Sy 10 Ns % Pq uint
Determines the dirty space limit, expressed as a percentage of all memory.
Once this limit is exceeded, new writes are halted until space frees up.
The parameter
.Sy zfs_dirty_data_max
takes precedence over this one.
.No See Sx ZFS TRANSACTION DELAY .
.Pp
Subject to
.Sy zfs_dirty_data_max_max .
.
.It Sy zfs_dirty_data_sync_percent Ns = Ns Sy 20 Ns % Pq uint
Start syncing out a transaction group if there's at least this much dirty data
.Pq as a percentage of Sy zfs_dirty_data_max .
This should be less than
.Sy zfs_vdev_async_write_active_min_dirty_percent .
.
.It Sy zfs_wrlog_data_max Ns = Pq int
The upper limit of write-transaction ZIL log data size in bytes.
Write operations are throttled when approaching the limit until log data is
cleared out after transaction group sync.
Because of some overhead, it should be set at least 2 times the size of
.Sy zfs_dirty_data_max
.No to prevent harming normal write throughput .
It also should be smaller than the size of the slog device if slog is present.
.Pp
Defaults to
.Sy zfs_dirty_data_max*2
.
.It Sy zfs_fallocate_reserve_percent Ns = Ns Sy 110 Ns % Pq uint
Since ZFS is a copy-on-write filesystem with snapshots, blocks cannot be
preallocated for a file in order to guarantee that later writes will not
run out of space.
Instead,
.Xr fallocate 2
space preallocation only checks that sufficient space is currently available
in the pool or the user's project quota allocation,
and then creates a sparse file of the requested size.
The requested space is multiplied by
.Sy zfs_fallocate_reserve_percent
to allow additional space for indirect blocks and other internal metadata.
Setting this to
.Sy 0
disables support for
.Xr fallocate 2
and causes it to return
.Sy EOPNOTSUPP .
.
.It Sy zfs_fletcher_4_impl Ns = Ns Sy fastest Pq string
Select a fletcher 4 implementation.
.Pp
Supported selectors are:
.Sy fastest , scalar , sse2 , ssse3 , avx2 , avx512f , avx512bw ,
.No and Sy aarch64_neon .
All except
.Sy fastest No and Sy scalar
require instruction set extensions to be available,
and will only appear if ZFS detects that they are present at runtime.
If multiple implementations of fletcher 4 are available, the
.Sy fastest
will be chosen using a micro benchmark.
Selecting
.Sy scalar
results in the original CPU-based calculation being used.
Selecting any option other than
.Sy fastest No or Sy scalar
results in vector instructions
from the respective CPU instruction set being used.
.
.It Sy zfs_bclone_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enables access to the block cloning feature.
If this setting is 0, then even if feature@block_cloning is enabled,
using functions and system calls that attempt to clone blocks will act as
though the feature is disabled.
.
.It Sy zfs_bclone_wait_dirty Ns = Ns Sy 1 Ns | Ns 0 Pq int
When set to 1 the FICLONE and FICLONERANGE ioctls will wait for any dirty
data to be written to disk before proceeding.
This ensures that the clone operation reliably succeeds, even if a file is
modified and then immediately cloned.
Note that for small files this may be slower than simply copying the file.
When set to 0 the clone operation will immediately fail if it encounters
any dirty blocks.
By default waiting is enabled.
.
.It Sy zfs_blake3_impl Ns = Ns Sy fastest Pq string
Select a BLAKE3 implementation.
.Pp
Supported selectors are:
.Sy cycle , fastest , generic , sse2 , sse41 , avx2 , avx512 .
All except
.Sy cycle , fastest No and Sy generic
require instruction set extensions to be available,
and will only appear if ZFS detects that they are present at runtime.
If multiple implementations of BLAKE3 are available, the
.Sy fastest will be chosen using a micro benchmark. You can see the
benchmark results by reading this kstat file:
.Pa /proc/spl/kstat/zfs/chksum_bench .
.
.It Sy zfs_free_bpobj_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enable/disable the processing of the free_bpobj object.
.
.It Sy zfs_async_block_max_blocks Ns = Ns Sy UINT64_MAX Po unlimited Pc Pq u64
Maximum number of blocks freed in a single TXG.
.
.It Sy zfs_max_async_dedup_frees Ns = Ns Sy 100000 Po 10^5 Pc Pq u64
Maximum number of dedup blocks freed in a single TXG.
.
.It Sy zfs_vdev_async_read_max_active Ns = Ns Sy 3 Pq uint
Maximum asynchronous read I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_async_read_min_active Ns = Ns Sy 1 Pq uint
Minimum asynchronous read I/O operation active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_async_write_active_max_dirty_percent Ns = Ns Sy 60 Ns % Pq uint
When the pool has more than this much dirty data, use
.Sy zfs_vdev_async_write_max_active
to limit active async writes.
If the dirty data is between the minimum and maximum,
the active I/O limit is linearly interpolated.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_async_write_active_min_dirty_percent Ns = Ns Sy 30 Ns % Pq uint
When the pool has less than this much dirty data, use
.Sy zfs_vdev_async_write_min_active
to limit active async writes.
If the dirty data is between the minimum and maximum,
the active I/O limit is linearly
interpolated.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_async_write_max_active Ns = Ns Sy 10 Pq uint
Maximum asynchronous write I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_async_write_min_active Ns = Ns Sy 2 Pq uint
Minimum asynchronous write I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.Pp
Lower values are associated with better latency on rotational media but poorer
resilver performance.
The default value of
.Sy 2
was chosen as a compromise.
A value of
.Sy 3
has been shown to improve resilver performance further at a cost of
further increasing latency.
.
.It Sy zfs_vdev_initializing_max_active Ns = Ns Sy 1 Pq uint
Maximum initializing I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_initializing_min_active Ns = Ns Sy 1 Pq uint
Minimum initializing I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_max_active Ns = Ns Sy 1000 Pq uint
The maximum number of I/O operations active to each device.
Ideally, this will be at least the sum of each queue's
.Sy max_active .
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_open_timeout_ms Ns = Ns Sy 1000 Pq uint
Timeout value to wait before determining a device is missing
during import.
This is helpful for transient missing paths due
to links being briefly removed and recreated in response to
udev events.
.
.It Sy zfs_vdev_rebuild_max_active Ns = Ns Sy 3 Pq uint
Maximum sequential resilver I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_rebuild_min_active Ns = Ns Sy 1 Pq uint
Minimum sequential resilver I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_removal_max_active Ns = Ns Sy 2 Pq uint
Maximum removal I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_removal_min_active Ns = Ns Sy 1 Pq uint
Minimum removal I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_scrub_max_active Ns = Ns Sy 2 Pq uint
Maximum scrub I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_scrub_min_active Ns = Ns Sy 1 Pq uint
Minimum scrub I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_sync_read_max_active Ns = Ns Sy 10 Pq uint
Maximum synchronous read I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_sync_read_min_active Ns = Ns Sy 10 Pq uint
Minimum synchronous read I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_sync_write_max_active Ns = Ns Sy 10 Pq uint
Maximum synchronous write I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_sync_write_min_active Ns = Ns Sy 10 Pq uint
Minimum synchronous write I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_trim_max_active Ns = Ns Sy 2 Pq uint
Maximum trim/discard I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_trim_min_active Ns = Ns Sy 1 Pq uint
Minimum trim/discard I/O operations active to each device.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_nia_delay Ns = Ns Sy 5 Pq uint
For non-interactive I/O (scrub, resilver, removal, initialize and rebuild),
the number of concurrently-active I/O operations is limited to
.Sy zfs_*_min_active ,
unless the vdev is "idle".
When there are no interactive I/O operations active (synchronous or otherwise),
and
.Sy zfs_vdev_nia_delay
operations have completed since the last interactive operation,
then the vdev is considered to be "idle",
and the number of concurrently-active non-interactive operations is increased to
.Sy zfs_*_max_active .
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_nia_credit Ns = Ns Sy 5 Pq uint
Some HDDs tend to prioritize sequential I/O so strongly, that concurrent
random I/O latency reaches several seconds.
On some HDDs this happens even if sequential I/O operations
are submitted one at a time, and so setting
.Sy zfs_*_max_active Ns = Sy 1
does not help.
To prevent non-interactive I/O, like scrub,
from monopolizing the device, no more than
.Sy zfs_vdev_nia_credit operations can be sent
while there are outstanding incomplete interactive operations.
This enforced wait ensures the HDD services the interactive I/O
within a reasonable amount of time.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_queue_depth_pct Ns = Ns Sy 1000 Ns % Pq uint
Maximum number of queued allocations per top-level vdev expressed as
a percentage of
.Sy zfs_vdev_async_write_max_active ,
which allows the system to detect devices that are more capable
of handling allocations and to allocate more blocks to those devices.
This allows for dynamic allocation distribution when devices are imbalanced,
as fuller devices will tend to be slower than empty devices.
.Pp
Also see
.Sy zio_dva_throttle_enabled .
.
.It Sy zfs_vdev_def_queue_depth Ns = Ns Sy 32 Pq uint
Default queue depth for each vdev IO allocator.
Higher values allow for better coalescing of sequential writes before sending
them to the disk, but can increase transaction commit times.
.
.It Sy zfs_vdev_failfast_mask Ns = Ns Sy 1 Pq uint
Defines if the driver should retire on a given error type.
The following options may be bitwise-ored together:
.TS
box;
lbz r l l .
	Value	Name	Description
_
	1	Device	No driver retries on device errors
	2	Transport	No driver retries on transport errors.
	4	Driver	No driver retries on driver errors.
.TE
.
.It Sy zfs_vdev_disk_max_segs Ns = Ns Sy 0 Pq uint
Maximum number of segments to add to a BIO (min 4).
If this is higher than the maximum allowed by the device queue or the kernel
itself, it will be clamped.
Setting it to zero will cause the kernel's ideal size to be used.
This parameter only applies on Linux.
This parameter is ignored if
.Sy zfs_vdev_disk_classic Ns = Ns Sy 1 .
.
.It Sy zfs_vdev_disk_classic Ns = Ns Sy 0 Ns | Ns 1 Pq uint
If set to 1, OpenZFS will submit IO to Linux using the method it used in 2.2
and earlier.
This "classic" method has known issues with highly fragmented IO requests and
is slower on many workloads, but it has been in use for many years and is known
to be very stable.
If you set this parameter, please also open a bug report why you did so,
including the workload involved and any error messages.
.Pp
This parameter and the classic submission method will be removed once we have
total confidence in the new method.
.Pp
This parameter only applies on Linux, and can only be set at module load time.
.
.It Sy zfs_expire_snapshot Ns = Ns Sy 300 Ns s Pq int
Time before expiring
.Pa .zfs/snapshot .
.
.It Sy zfs_admin_snapshot Ns = Ns Sy 0 Ns | Ns 1 Pq int
Allow the creation, removal, or renaming of entries in the
.Sy .zfs/snapshot
directory to cause the creation, destruction, or renaming of snapshots.
When enabled, this functionality works both locally and over NFS exports
which have the
.Em no_root_squash
option set.
.
.It Sy zfs_snapshot_no_setuid Ns = Ns Sy 0 Ns | Ns 1 Pq int
Whether to disable
.Em setuid/setgid
support for snapshot mounts triggered by access to the
.Sy .zfs/snapshot
directory by setting the
.Em nosuid
mount option.
.
.It Sy zfs_flags Ns = Ns Sy 0 Pq int
Set additional debugging flags.
The following flags may be bitwise-ored together:
.TS
box;
lbz r l l .
	Value	Name	Description
_
	1	ZFS_DEBUG_DPRINTF	Enable dprintf entries in the debug log.
*	2	ZFS_DEBUG_DBUF_VERIFY	Enable extra dbuf verifications.
*	4	ZFS_DEBUG_DNODE_VERIFY	Enable extra dnode verifications.
	8	ZFS_DEBUG_SNAPNAMES	Enable snapshot name verification.
*	16	ZFS_DEBUG_MODIFY	Check for illegally modified ARC buffers.
	64	ZFS_DEBUG_ZIO_FREE	Enable verification of block frees.
	128	ZFS_DEBUG_HISTOGRAM_VERIFY	Enable extra spacemap histogram verifications.
	256	ZFS_DEBUG_METASLAB_VERIFY	Verify space accounting on disk matches in-memory \fBrange_trees\fP.
	512	ZFS_DEBUG_SET_ERROR	Enable \fBSET_ERROR\fP and dprintf entries in the debug log.
	1024	ZFS_DEBUG_INDIRECT_REMAP	Verify split blocks created by device removal.
	2048	ZFS_DEBUG_TRIM	Verify TRIM ranges are always within the allocatable range tree.
	4096	ZFS_DEBUG_LOG_SPACEMAP	Verify that the log summary is consistent with the spacemap log
			       and enable \fBzfs_dbgmsgs\fP for metaslab loading and flushing.
	8192	ZFS_DEBUG_METASLAB_ALLOC	Enable debugging messages when allocations fail.
	16384	ZFS_DEBUG_BRT	Enable BRT-related debugging messages.
	32768	ZFS_DEBUG_RAIDZ_RECONSTRUCT	Enabled debugging messages for raidz reconstruction.
	65536	ZFS_DEBUG_DDT	Enable DDT-related debugging messages.
.TE
.Sy \& * No Requires debug build .
.
.It Sy zfs_btree_verify_intensity Ns = Ns Sy 0 Pq uint
Enables btree verification.
The following settings are cumulative:
.TS
box;
lbz r l l .
	Value	Description

	1	Verify height.
	2	Verify pointers from children to parent.
	3	Verify element counts.
	4	Verify element order. (expensive)
*	5	Verify unused memory is poisoned. (expensive)
.TE
.Sy \& * No Requires debug build .
.
.It Sy zfs_free_leak_on_eio Ns = Ns Sy 0 Ns | Ns 1 Pq int
If destroy encounters an
.Sy EIO
while reading metadata (e.g. indirect blocks),
space referenced by the missing metadata can not be freed.
Normally this causes the background destroy to become "stalled",
as it is unable to make forward progress.
While in this stalled state, all remaining space to free
from the error-encountering filesystem is "temporarily leaked".
Set this flag to cause it to ignore the
.Sy EIO ,
permanently leak the space from indirect blocks that can not be read,
and continue to free everything else that it can.
.Pp
The default "stalling" behavior is useful if the storage partially
fails (i.e. some but not all I/O operations fail), and then later recovers.
In this case, we will be able to continue pool operations while it is
partially failed, and when it recovers, we can continue to free the
space, with no leaks.
Note, however, that this case is actually fairly rare.
.Pp
Typically pools either
.Bl -enum -compact -offset 4n -width "1."
.It
fail completely (but perhaps temporarily,
e.g. due to a top-level vdev going offline), or
.It
have localized, permanent errors (e.g. disk returns the wrong data
due to bit flip or firmware bug).
.El
In the former case, this setting does not matter because the
pool will be suspended and the sync thread will not be able to make
forward progress regardless.
In the latter, because the error is permanent, the best we can do
is leak the minimum amount of space,
which is what setting this flag will do.
It is therefore reasonable for this flag to normally be set,
but we chose the more conservative approach of not setting it,
so that there is no possibility of
leaking space in the "partial temporary" failure case.
.
.It Sy zfs_free_min_time_ms Ns = Ns Sy 1000 Ns ms Po 1s Pc Pq uint
During a
.Nm zfs Cm destroy
operation using the
.Sy async_destroy
feature,
a minimum of this much time will be spent working on freeing blocks per TXG.
.
.It Sy zfs_obsolete_min_time_ms Ns = Ns Sy 500 Ns ms Pq uint
Similar to
.Sy zfs_free_min_time_ms ,
but for cleanup of old indirection records for removed vdevs.
.
.It Sy zfs_immediate_write_sz Ns = Ns Sy 32768 Ns B Po 32 KiB Pc Pq s64
Largest data block to write to the ZIL.
Larger blocks will be treated as if the dataset being written to had the
.Sy logbias Ns = Ns Sy throughput
property set.
.
.It Sy zfs_initialize_value Ns = Ns Sy 16045690984833335022 Po 0xDEADBEEFDEADBEEE Pc Pq u64
Pattern written to vdev free space by
.Xr zpool-initialize 8 .
.
.It Sy zfs_initialize_chunk_size Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq u64
Size of writes used by
.Xr zpool-initialize 8 .
This option is used by the test suite.
.
.It Sy zfs_livelist_max_entries Ns = Ns Sy 500000 Po 5*10^5 Pc Pq u64
The threshold size (in block pointers) at which we create a new sub-livelist.
Larger sublists are more costly from a memory perspective but the fewer
sublists there are, the lower the cost of insertion.
.
.It Sy zfs_livelist_min_percent_shared Ns = Ns Sy 75 Ns % Pq int
If the amount of shared space between a snapshot and its clone drops below
this threshold, the clone turns off the livelist and reverts to the old
deletion method.
This is in place because livelists no long give us a benefit
once a clone has been overwritten enough.
.
.It Sy zfs_livelist_condense_new_alloc Ns = Ns Sy 0 Pq int
Incremented each time an extra ALLOC blkptr is added to a livelist entry while
it is being condensed.
This option is used by the test suite to track race conditions.
.
.It Sy zfs_livelist_condense_sync_cancel Ns = Ns Sy 0 Pq int
Incremented each time livelist condensing is canceled while in
.Fn spa_livelist_condense_sync .
This option is used by the test suite to track race conditions.
.
.It Sy zfs_livelist_condense_sync_pause Ns = Ns Sy 0 Ns | Ns 1 Pq int
When set, the livelist condense process pauses indefinitely before
executing the synctask \(em
.Fn spa_livelist_condense_sync .
This option is used by the test suite to trigger race conditions.
.
.It Sy zfs_livelist_condense_zthr_cancel Ns = Ns Sy 0 Pq int
Incremented each time livelist condensing is canceled while in
.Fn spa_livelist_condense_cb .
This option is used by the test suite to track race conditions.
.
.It Sy zfs_livelist_condense_zthr_pause Ns = Ns Sy 0 Ns | Ns 1 Pq int
When set, the livelist condense process pauses indefinitely before
executing the open context condensing work in
.Fn spa_livelist_condense_cb .
This option is used by the test suite to trigger race conditions.
.
.It Sy zfs_lua_max_instrlimit Ns = Ns Sy 100000000 Po 10^8 Pc Pq u64
The maximum execution time limit that can be set for a ZFS channel program,
specified as a number of Lua instructions.
.
.It Sy zfs_lua_max_memlimit Ns = Ns Sy 104857600 Po 100 MiB Pc Pq u64
The maximum memory limit that can be set for a ZFS channel program, specified
in bytes.
.
.It Sy zfs_max_dataset_nesting Ns = Ns Sy 50 Pq int
The maximum depth of nested datasets.
This value can be tuned temporarily to
fix existing datasets that exceed the predefined limit.
.
.It Sy zfs_max_log_walking Ns = Ns Sy 5 Pq u64
The number of past TXGs that the flushing algorithm of the log spacemap
feature uses to estimate incoming log blocks.
.
.It Sy zfs_max_logsm_summary_length Ns = Ns Sy 10 Pq u64
Maximum number of rows allowed in the summary of the spacemap log.
.
.It Sy zfs_max_recordsize Ns = Ns Sy 16777216 Po 16 MiB Pc Pq uint
We currently support block sizes from
.Em 512 Po 512 B Pc No to Em 16777216 Po 16 MiB Pc .
The benefits of larger blocks, and thus larger I/O,
need to be weighed against the cost of COWing a giant block to modify one byte.
Additionally, very large blocks can have an impact on I/O latency,
and also potentially on the memory allocator.
Therefore, we formerly forbade creating blocks larger than 1M.
Larger blocks could be created by changing it,
and pools with larger blocks can always be imported and used,
regardless of this setting.
.Pp
Note that it is still limited by default to
.Ar 1 MiB
on x86_32, because Linux's
3/1 memory split doesn't leave much room for 16M chunks.
.
.It Sy zfs_allow_redacted_dataset_mount Ns = Ns Sy 0 Ns | Ns 1 Pq int
Allow datasets received with redacted send/receive to be mounted.
Normally disabled because these datasets may be missing key data.
.
.It Sy zfs_min_metaslabs_to_flush Ns = Ns Sy 1 Pq u64
Minimum number of metaslabs to flush per dirty TXG.
.
.It Sy zfs_metaslab_fragmentation_threshold Ns = Ns Sy 77 Ns % Pq uint
Allow metaslabs to keep their active state as long as their fragmentation
percentage is no more than this value.
An active metaslab that exceeds this threshold
will no longer keep its active status allowing better metaslabs to be selected.
.
.It Sy zfs_mg_fragmentation_threshold Ns = Ns Sy 95 Ns % Pq uint
Metaslab groups are considered eligible for allocations if their
fragmentation metric (measured as a percentage) is less than or equal to
this value.
If a metaslab group exceeds this threshold then it will be
skipped unless all metaslab groups within the metaslab class have also
crossed this threshold.
.
.It Sy zfs_mg_noalloc_threshold Ns = Ns Sy 0 Ns % Pq uint
Defines a threshold at which metaslab groups should be eligible for allocations.
The value is expressed as a percentage of free space
beyond which a metaslab group is always eligible for allocations.
If a metaslab group's free space is less than or equal to the
threshold, the allocator will avoid allocating to that group
unless all groups in the pool have reached the threshold.
Once all groups have reached the threshold, all groups are allowed to accept
allocations.
The default value of
.Sy 0
disables the feature and causes all metaslab groups to be eligible for
allocations.
.Pp
This parameter allows one to deal with pools having heavily imbalanced
vdevs such as would be the case when a new vdev has been added.
Setting the threshold to a non-zero percentage will stop allocations
from being made to vdevs that aren't filled to the specified percentage
and allow lesser filled vdevs to acquire more allocations than they
otherwise would under the old
.Sy zfs_mg_alloc_failures
facility.
.
.It Sy zfs_ddt_data_is_special Ns = Ns Sy 1 Ns | Ns 0 Pq int
If enabled, ZFS will place DDT data into the special allocation class.
.
.It Sy zfs_user_indirect_is_special Ns = Ns Sy 1 Ns | Ns 0 Pq int
If enabled, ZFS will place user data indirect blocks
into the special allocation class.
.
.It Sy zfs_multihost_history Ns = Ns Sy 0 Pq uint
Historical statistics for this many latest multihost updates will be available
in
.Pa /proc/spl/kstat/zfs/ Ns Ao Ar pool Ac Ns Pa /multihost .
.
.It Sy zfs_multihost_interval Ns = Ns Sy 1000 Ns ms Po 1 s Pc Pq u64
Used to control the frequency of multihost writes which are performed when the
.Sy multihost
pool property is on.
This is one of the factors used to determine the
length of the activity check during import.
.Pp
The multihost write period is
.Sy zfs_multihost_interval No / Sy leaf-vdevs .
On average a multihost write will be issued for each leaf vdev
every
.Sy zfs_multihost_interval
milliseconds.
In practice, the observed period can vary with the I/O load
and this observed value is the delay which is stored in the uberblock.
.
.It Sy zfs_multihost_import_intervals Ns = Ns Sy 20 Pq uint
Used to control the duration of the activity test on import.
Smaller values of
.Sy zfs_multihost_import_intervals
will reduce the import time but increase
the risk of failing to detect an active pool.
The total activity check time is never allowed to drop below one second.
.Pp
On import the activity check waits a minimum amount of time determined by
.Sy zfs_multihost_interval No \(mu Sy zfs_multihost_import_intervals ,
or the same product computed on the host which last had the pool imported,
whichever is greater.
The activity check time may be further extended if the value of MMP
delay found in the best uberblock indicates actual multihost updates happened
at longer intervals than
.Sy zfs_multihost_interval .
A minimum of
.Em 100 ms
is enforced.
.Pp
.Sy 0 No is equivalent to Sy 1 .
.
.It Sy zfs_multihost_fail_intervals Ns = Ns Sy 10 Pq uint
Controls the behavior of the pool when multihost write failures or delays are
detected.
.Pp
When
.Sy 0 ,
multihost write failures or delays are ignored.
The failures will still be reported to the ZED which depending on
its configuration may take action such as suspending the pool or offlining a
device.
.Pp
Otherwise, the pool will be suspended if
.Sy zfs_multihost_fail_intervals No \(mu Sy zfs_multihost_interval
milliseconds pass without a successful MMP write.
This guarantees the activity test will see MMP writes if the pool is imported.
.Sy 1 No is equivalent to Sy 2 ;
this is necessary to prevent the pool from being suspended
due to normal, small I/O latency variations.
.
.It Sy zfs_no_scrub_io Ns = Ns Sy 0 Ns | Ns 1 Pq int
Set to disable scrub I/O.
This results in scrubs not actually scrubbing data and
simply doing a metadata crawl of the pool instead.
.
.It Sy zfs_no_scrub_prefetch Ns = Ns Sy 0 Ns | Ns 1 Pq int
Set to disable block prefetching for scrubs.
.
.It Sy zfs_nocacheflush Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disable cache flush operations on disks when writing.
Setting this will cause pool corruption on power loss
if a volatile out-of-order write cache is enabled.
.
.It Sy zfs_nopwrite_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Allow no-operation writes.
The occurrence of nopwrites will further depend on other pool properties
.Pq i.a. the checksumming and compression algorithms .
.
.It Sy zfs_dmu_offset_next_sync Ns = Ns Sy 1 Ns | Ns 0 Pq int
Enable forcing TXG sync to find holes.
When enabled forces ZFS to sync data when
.Sy SEEK_HOLE No or Sy SEEK_DATA
flags are used allowing holes in a file to be accurately reported.
When disabled holes will not be reported in recently dirtied files.
.
.It Sy zfs_pd_bytes_max Ns = Ns Sy 52428800 Ns B Po 50 MiB Pc Pq int
The number of bytes which should be prefetched during a pool traversal, like
.Nm zfs Cm send
or other data crawling operations.
.
.It Sy zfs_traverse_indirect_prefetch_limit Ns = Ns Sy 32 Pq uint
The number of blocks pointed by indirect (non-L0) block which should be
prefetched during a pool traversal, like
.Nm zfs Cm send
or other data crawling operations.
.
.It Sy zfs_per_txg_dirty_frees_percent Ns = Ns Sy 30 Ns % Pq u64
Control percentage of dirtied indirect blocks from frees allowed into one TXG.
After this threshold is crossed, additional frees will wait until the next TXG.
.Sy 0 No disables this throttle .
.
.It Sy zfs_prefetch_disable Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disable predictive prefetch.
Note that it leaves "prescient" prefetch
.Pq for, e.g., Nm zfs Cm send
intact.
Unlike predictive prefetch, prescient prefetch never issues I/O
that ends up not being needed, so it can't hurt performance.
.
.It Sy zfs_qat_checksum_disable Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disable QAT hardware acceleration for SHA256 checksums.
May be unset after the ZFS modules have been loaded to initialize the QAT
hardware as long as support is compiled in and the QAT driver is present.
.
.It Sy zfs_qat_compress_disable Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disable QAT hardware acceleration for gzip compression.
May be unset after the ZFS modules have been loaded to initialize the QAT
hardware as long as support is compiled in and the QAT driver is present.
.
.It Sy zfs_qat_encrypt_disable Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disable QAT hardware acceleration for AES-GCM encryption.
May be unset after the ZFS modules have been loaded to initialize the QAT
hardware as long as support is compiled in and the QAT driver is present.
.
.It Sy zfs_vnops_read_chunk_size Ns = Ns Sy 33554432 Ns B Po 32 MiB Pc Pq u64
Bytes to read per chunk.
.
.It Sy zfs_read_history Ns = Ns Sy 0 Pq uint
Historical statistics for this many latest reads will be available in
.Pa /proc/spl/kstat/zfs/ Ns Ao Ar pool Ac Ns Pa /reads .
.
.It Sy zfs_read_history_hits Ns = Ns Sy 0 Ns | Ns 1 Pq int
Include cache hits in read history
.
.It Sy zfs_rebuild_max_segment Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq u64
Maximum read segment size to issue when sequentially resilvering a
top-level vdev.
.
.It Sy zfs_rebuild_scrub_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Automatically start a pool scrub when the last active sequential resilver
completes in order to verify the checksums of all blocks which have been
resilvered.
This is enabled by default and strongly recommended.
.
.It Sy zfs_rebuild_vdev_limit Ns = Ns Sy 67108864 Ns B Po 64 MiB Pc Pq u64
Maximum amount of I/O that can be concurrently issued for a sequential
resilver per leaf device, given in bytes.
.
.It Sy zfs_reconstruct_indirect_combinations_max Ns = Ns Sy 4096 Pq int
If an indirect split block contains more than this many possible unique
combinations when being reconstructed, consider it too computationally
expensive to check them all.
Instead, try at most this many randomly selected
combinations each time the block is accessed.
This allows all segment copies to participate fairly
in the reconstruction when all combinations
cannot be checked and prevents repeated use of one bad copy.
.
.It Sy zfs_recover Ns = Ns Sy 0 Ns | Ns 1 Pq int
Set to attempt to recover from fatal errors.
This should only be used as a last resort,
as it typically results in leaked space, or worse.
.
.It Sy zfs_removal_ignore_errors Ns = Ns Sy 0 Ns | Ns 1 Pq int
Ignore hard I/O errors during device removal.
When set, if a device encounters a hard I/O error during the removal process
the removal will not be canceled.
This can result in a normally recoverable block becoming permanently damaged
and is hence not recommended.
This should only be used as a last resort when the
pool cannot be returned to a healthy state prior to removing the device.
.
.It Sy zfs_removal_suspend_progress Ns = Ns Sy 0 Ns | Ns 1 Pq uint
This is used by the test suite so that it can ensure that certain actions
happen while in the middle of a removal.
.
.It Sy zfs_remove_max_segment Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq uint
The largest contiguous segment that we will attempt to allocate when removing
a device.
If there is a performance problem with attempting to allocate large blocks,
consider decreasing this.
The default value is also the maximum.
.
.It Sy zfs_resilver_disable_defer Ns = Ns Sy 0 Ns | Ns 1 Pq int
Ignore the
.Sy resilver_defer
feature, causing an operation that would start a resilver to
immediately restart the one in progress.
.
.It Sy zfs_resilver_defer_percent Ns = Ns Sy 10 Ns % Pq uint
If the ongoing resilver progress is below this threshold, a new resilver will
restart from scratch instead of being deferred after the current one finishes,
even if the
.Sy resilver_defer
feature is enabled.
.
.It Sy zfs_resilver_min_time_ms Ns = Ns Sy 3000 Ns ms Po 3 s Pc Pq uint
Resilvers are processed by the sync thread.
While resilvering, it will spend at least this much time
working on a resilver between TXG flushes.
.
.It Sy zfs_scan_ignore_errors Ns = Ns Sy 0 Ns | Ns 1 Pq int
If set, remove the DTL (dirty time list) upon completion of a pool scan (scrub),
even if there were unrepairable errors.
Intended to be used during pool repair or recovery to
stop resilvering when the pool is next imported.
.
.It Sy zfs_scrub_after_expand Ns = Ns Sy 1 Ns | Ns 0 Pq int
Automatically start a pool scrub after a RAIDZ expansion completes
in order to verify the checksums of all blocks which have been
copied during the expansion.
This is enabled by default and strongly recommended.
.
.It Sy zfs_scrub_min_time_ms Ns = Ns Sy 1000 Ns ms Po 1 s Pc Pq uint
Scrubs are processed by the sync thread.
While scrubbing, it will spend at least this much time
working on a scrub between TXG flushes.
.
.It Sy zfs_scrub_error_blocks_per_txg Ns = Ns Sy 4096 Pq uint
Error blocks to be scrubbed in one txg.
.
.It Sy zfs_scan_checkpoint_intval Ns = Ns Sy 7200 Ns s Po 2 hour Pc Pq uint
To preserve progress across reboots, the sequential scan algorithm periodically
needs to stop metadata scanning and issue all the verification I/O to disk.
The frequency of this flushing is determined by this tunable.
.
.It Sy zfs_scan_fill_weight Ns = Ns Sy 3 Pq uint
This tunable affects how scrub and resilver I/O segments are ordered.
A higher number indicates that we care more about how filled in a segment is,
while a lower number indicates we care more about the size of the extent without
considering the gaps within a segment.
This value is only tunable upon module insertion.
Changing the value afterwards will have no effect on scrub or resilver
performance.
.
.It Sy zfs_scan_issue_strategy Ns = Ns Sy 0 Pq uint
Determines the order that data will be verified while scrubbing or resilvering:
.Bl -tag -compact -offset 4n -width "a"
.It Sy 1
Data will be verified as sequentially as possible, given the
amount of memory reserved for scrubbing
.Pq see Sy zfs_scan_mem_lim_fact .
This may improve scrub performance if the pool's data is very fragmented.
.It Sy 2
The largest mostly-contiguous chunk of found data will be verified first.
By deferring scrubbing of small segments, we may later find adjacent data
to coalesce and increase the segment size.
.It Sy 0
.No Use strategy Sy 1 No during normal verification
.No and strategy Sy 2 No while taking a checkpoint .
.El
.
.It Sy zfs_scan_legacy Ns = Ns Sy 0 Ns | Ns 1 Pq int
If unset, indicates that scrubs and resilvers will gather metadata in
memory before issuing sequential I/O.
Otherwise indicates that the legacy algorithm will be used,
where I/O is initiated as soon as it is discovered.
Unsetting will not affect scrubs or resilvers that are already in progress.
.
.It Sy zfs_scan_max_ext_gap Ns = Ns Sy 2097152 Ns B Po 2 MiB Pc Pq int
Sets the largest gap in bytes between scrub/resilver I/O operations
that will still be considered sequential for sorting purposes.
Changing this value will not
affect scrubs or resilvers that are already in progress.
.
.It Sy zfs_scan_mem_lim_fact Ns = Ns Sy 20 Ns ^-1 Pq uint
Maximum fraction of RAM used for I/O sorting by sequential scan algorithm.
This tunable determines the hard limit for I/O sorting memory usage.
When the hard limit is reached we stop scanning metadata and start issuing
data verification I/O.
This is done until we get below the soft limit.
.
.It Sy zfs_scan_mem_lim_soft_fact Ns = Ns Sy 20 Ns ^-1 Pq uint
The fraction of the hard limit used to determined the soft limit for I/O sorting
by the sequential scan algorithm.
When we cross this limit from below no action is taken.
When we cross this limit from above it is because we are issuing verification
I/O.
In this case (unless the metadata scan is done) we stop issuing verification I/O
and start scanning metadata again until we get to the hard limit.
.
.It Sy zfs_scan_report_txgs Ns = Ns Sy 0 Ns | Ns 1 Pq uint
When reporting resilver throughput and estimated completion time use the
performance observed over roughly the last
.Sy zfs_scan_report_txgs
TXGs.
When set to zero performance is calculated over the time between checkpoints.
.
.It Sy zfs_scan_strict_mem_lim Ns = Ns Sy 0 Ns | Ns 1 Pq int
Enforce tight memory limits on pool scans when a sequential scan is in progress.
When disabled, the memory limit may be exceeded by fast disks.
.
.It Sy zfs_scan_suspend_progress Ns = Ns Sy 0 Ns | Ns 1 Pq int
Freezes a scrub/resilver in progress without actually pausing it.
Intended for testing/debugging.
.
.It Sy zfs_scan_vdev_limit Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq int
Maximum amount of data that can be concurrently issued at once for scrubs and
resilvers per leaf device, given in bytes.
.
.It Sy zfs_send_corrupt_data Ns = Ns Sy 0 Ns | Ns 1 Pq int
Allow sending of corrupt data (ignore read/checksum errors when sending).
.
.It Sy zfs_send_unmodified_spill_blocks Ns = Ns Sy 1 Ns | Ns 0 Pq int
Include unmodified spill blocks in the send stream.
Under certain circumstances, previous versions of ZFS could incorrectly
remove the spill block from an existing object.
Including unmodified copies of the spill blocks creates a backwards-compatible
stream which will recreate a spill block if it was incorrectly removed.
.
.It Sy zfs_send_no_prefetch_queue_ff Ns = Ns Sy 20 Ns ^\-1 Pq uint
The fill fraction of the
.Nm zfs Cm send
internal queues.
The fill fraction controls the timing with which internal threads are woken up.
.
.It Sy zfs_send_no_prefetch_queue_length Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq uint
The maximum number of bytes allowed in
.Nm zfs Cm send Ns 's
internal queues.
.
.It Sy zfs_send_queue_ff Ns = Ns Sy 20 Ns ^\-1 Pq uint
The fill fraction of the
.Nm zfs Cm send
prefetch queue.
The fill fraction controls the timing with which internal threads are woken up.
.
.It Sy zfs_send_queue_length Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq uint
The maximum number of bytes allowed that will be prefetched by
.Nm zfs Cm send .
This value must be at least twice the maximum block size in use.
.
.It Sy zfs_recv_queue_ff Ns = Ns Sy 20 Ns ^\-1 Pq uint
The fill fraction of the
.Nm zfs Cm receive
queue.
The fill fraction controls the timing with which internal threads are woken up.
.
.It Sy zfs_recv_queue_length Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq uint
The maximum number of bytes allowed in the
.Nm zfs Cm receive
queue.
This value must be at least twice the maximum block size in use.
.
.It Sy zfs_recv_write_batch_size Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq uint
The maximum amount of data, in bytes, that
.Nm zfs Cm receive
will write in one DMU transaction.
This is the uncompressed size, even when receiving a compressed send stream.
This setting will not reduce the write size below a single block.
Capped at a maximum of
.Sy 32 MiB .
.
.It Sy zfs_recv_best_effort_corrective Ns = Ns Sy 0 Pq int
When this variable is set to non-zero a corrective receive:
.Bl -enum -compact -offset 4n -width "1."
.It
Does not enforce the restriction of source & destination snapshot GUIDs
matching.
.It
If there is an error during healing, the healing receive is not
terminated instead it moves on to the next record.
.El
.
.It Sy zfs_override_estimate_recordsize Ns = Ns Sy 0 Ns | Ns 1 Pq uint
Setting this variable overrides the default logic for estimating block
sizes when doing a
.Nm zfs Cm send .
The default heuristic is that the average block size
will be the current recordsize.
Override this value if most data in your dataset is not of that size
and you require accurate zfs send size estimates.
.
.It Sy zfs_sync_pass_deferred_free Ns = Ns Sy 2 Pq uint
Flushing of data to disk is done in passes.
Defer frees starting in this pass.
.
.It Sy zfs_spa_discard_memory_limit Ns = Ns Sy 16777216 Ns B Po 16 MiB Pc Pq int
Maximum memory used for prefetching a checkpoint's space map on each
vdev while discarding the checkpoint.
.
.It Sy zfs_special_class_metadata_reserve_pct Ns = Ns Sy 25 Ns % Pq uint
Only allow small data blocks to be allocated on the special and dedup vdev
types when the available free space percentage on these vdevs exceeds this
value.
This ensures reserved space is available for pool metadata as the
special vdevs approach capacity.
.
.It Sy zfs_sync_pass_dont_compress Ns = Ns Sy 8 Pq uint
Starting in this sync pass, disable compression (including of metadata).
With the default setting, in practice, we don't have this many sync passes,
so this has no effect.
.Pp
The original intent was that disabling compression would help the sync passes
to converge.
However, in practice, disabling compression increases
the average number of sync passes; because when we turn compression off,
many blocks' size will change, and thus we have to re-allocate
(not overwrite) them.
It also increases the number of
.Em 128 KiB
allocations (e.g. for indirect blocks and spacemaps)
because these will not be compressed.
The
.Em 128 KiB
allocations are especially detrimental to performance
on highly fragmented systems, which may have very few free segments of this
size,
and may need to load new metaslabs to satisfy these allocations.
.
.It Sy zfs_sync_pass_rewrite Ns = Ns Sy 2 Pq uint
Rewrite new block pointers starting in this pass.
.
.It Sy zfs_trim_extent_bytes_max Ns = Ns Sy 134217728 Ns B Po 128 MiB Pc Pq uint
Maximum size of TRIM command.
Larger ranges will be split into chunks no larger than this value before
issuing.
.
.It Sy zfs_trim_extent_bytes_min Ns = Ns Sy 32768 Ns B Po 32 KiB Pc Pq uint
Minimum size of TRIM commands.
TRIM ranges smaller than this will be skipped,
unless they're part of a larger range which was chunked.
This is done because it's common for these small TRIMs
to negatively impact overall performance.
.
.It Sy zfs_trim_metaslab_skip Ns = Ns Sy 0 Ns | Ns 1 Pq uint
Skip uninitialized metaslabs during the TRIM process.
This option is useful for pools constructed from large thinly-provisioned
devices
where TRIM operations are slow.
As a pool ages, an increasing fraction of the pool's metaslabs
will be initialized, progressively degrading the usefulness of this option.
This setting is stored when starting a manual TRIM and will
persist for the duration of the requested TRIM.
.
.It Sy zfs_trim_queue_limit Ns = Ns Sy 10 Pq uint
Maximum number of queued TRIMs outstanding per leaf vdev.
The number of concurrent TRIM commands issued to the device is controlled by
.Sy zfs_vdev_trim_min_active No and Sy zfs_vdev_trim_max_active .
.
.It Sy zfs_trim_txg_batch Ns = Ns Sy 32 Pq uint
The number of transaction groups' worth of frees which should be aggregated
before TRIM operations are issued to the device.
This setting represents a trade-off between issuing larger,
more efficient TRIM operations and the delay
before the recently trimmed space is available for use by the device.
.Pp
Increasing this value will allow frees to be aggregated for a longer time.
This will result is larger TRIM operations and potentially increased memory
usage.
Decreasing this value will have the opposite effect.
The default of
.Sy 32
was determined to be a reasonable compromise.
.
.It Sy zfs_txg_history Ns = Ns Sy 100 Pq uint
Historical statistics for this many latest TXGs will be available in
.Pa /proc/spl/kstat/zfs/ Ns Ao Ar pool Ac Ns Pa /TXGs .
.
.It Sy zfs_txg_timeout Ns = Ns Sy 5 Ns s Pq uint
Flush dirty data to disk at least every this many seconds (maximum TXG
duration).
.
.It Sy zfs_vdev_aggregation_limit Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq uint
Max vdev I/O aggregation size.
.
.It Sy zfs_vdev_aggregation_limit_non_rotating Ns = Ns Sy 131072 Ns B Po 128 KiB Pc Pq uint
Max vdev I/O aggregation size for non-rotating media.
.
.It Sy zfs_vdev_mirror_rotating_inc Ns = Ns Sy 0 Pq int
A number by which the balancing algorithm increments the load calculation for
the purpose of selecting the least busy mirror member when an I/O operation
immediately follows its predecessor on rotational vdevs
for the purpose of making decisions based on load.
.
.It Sy zfs_vdev_mirror_rotating_seek_inc Ns = Ns Sy 5 Pq int
A number by which the balancing algorithm increments the load calculation for
the purpose of selecting the least busy mirror member when an I/O operation
lacks locality as defined by
.Sy zfs_vdev_mirror_rotating_seek_offset .
Operations within this that are not immediately following the previous operation
are incremented by half.
.
.It Sy zfs_vdev_mirror_rotating_seek_offset Ns = Ns Sy 1048576 Ns B Po 1 MiB Pc Pq int
The maximum distance for the last queued I/O operation in which
the balancing algorithm considers an operation to have locality.
.No See Sx ZFS I/O SCHEDULER .
.
.It Sy zfs_vdev_mirror_non_rotating_inc Ns = Ns Sy 0 Pq int
A number by which the balancing algorithm increments the load calculation for
the purpose of selecting the least busy mirror member on non-rotational vdevs
when I/O operations do not immediately follow one another.
.
.It Sy zfs_vdev_mirror_non_rotating_seek_inc Ns = Ns Sy 1 Pq int
A number by which the balancing algorithm increments the load calculation for
the purpose of selecting the least busy mirror member when an I/O operation
lacks
locality as defined by the
.Sy zfs_vdev_mirror_rotating_seek_offset .
Operations within this that are not immediately following the previous operation
are incremented by half.
.
.It Sy zfs_vdev_read_gap_limit Ns = Ns Sy 32768 Ns B Po 32 KiB Pc Pq uint
Aggregate read I/O operations if the on-disk gap between them is within this
threshold.
.
.It Sy zfs_vdev_write_gap_limit Ns = Ns Sy 4096 Ns B Po 4 KiB Pc Pq uint
Aggregate write I/O operations if the on-disk gap between them is within this
threshold.
.
.It Sy zfs_vdev_raidz_impl Ns = Ns Sy fastest Pq string
Select the raidz parity implementation to use.
.Pp
Variants that don't depend on CPU-specific features
may be selected on module load, as they are supported on all systems.
The remaining options may only be set after the module is loaded,
as they are available only if the implementations are compiled in
and supported on the running system.
.Pp
Once the module is loaded,
.Pa /sys/module/zfs/parameters/zfs_vdev_raidz_impl
will show the available options,
with the currently selected one enclosed in square brackets.
.Pp
.TS
lb l l .
fastest	selected by built-in benchmark
original	original implementation
scalar	scalar implementation
sse2	SSE2 instruction set	64-bit x86
ssse3	SSSE3 instruction set	64-bit x86
avx2	AVX2 instruction set	64-bit x86
avx512f	AVX512F instruction set	64-bit x86
avx512bw	AVX512F & AVX512BW instruction sets	64-bit x86
aarch64_neon	NEON	Aarch64/64-bit ARMv8
aarch64_neonx2	NEON with more unrolling	Aarch64/64-bit ARMv8
powerpc_altivec	Altivec	PowerPC
.TE
.
.It Sy zfs_vdev_scheduler Pq charp
.Sy DEPRECATED .
Prints warning to kernel log for compatibility.
.
.It Sy zfs_zevent_len_max Ns = Ns Sy 512 Pq uint
Max event queue length.
Events in the queue can be viewed with
.Xr zpool-events 8 .
.
.It Sy zfs_zevent_retain_max Ns = Ns Sy 2000 Pq int
Maximum recent zevent records to retain for duplicate checking.
Setting this to
.Sy 0
disables duplicate detection.
.
.It Sy zfs_zevent_retain_expire_secs Ns = Ns Sy 900 Ns s Po 15 min Pc Pq int
Lifespan for a recent ereport that was retained for duplicate checking.
.
.It Sy zfs_zil_clean_taskq_maxalloc Ns = Ns Sy 1048576 Pq int
The maximum number of taskq entries that are allowed to be cached.
When this limit is exceeded transaction records (itxs)
will be cleaned synchronously.
.
.It Sy zfs_zil_clean_taskq_minalloc Ns = Ns Sy 1024 Pq int
The number of taskq entries that are pre-populated when the taskq is first
created and are immediately available for use.
.
.It Sy zfs_zil_clean_taskq_nthr_pct Ns = Ns Sy 100 Ns % Pq int
This controls the number of threads used by
.Sy dp_zil_clean_taskq .
The default value of
.Sy 100%
will create a maximum of one thread per CPU.
.
.It Sy zil_maxblocksize Ns = Ns Sy 131072 Ns B Po 128 KiB Pc Pq uint
This sets the maximum block size used by the ZIL.
On very fragmented pools, lowering this
.Pq typically to Sy 36 KiB
can improve performance.
.
.It Sy zil_maxcopied Ns = Ns Sy 7680 Ns B Po 7.5 KiB Pc Pq uint
This sets the maximum number of write bytes logged via WR_COPIED.
It tunes a tradeoff between additional memory copy and possibly worse log
space efficiency vs additional range lock/unlock.
.
.It Sy zil_nocacheflush Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disable the cache flush commands that are normally sent to disk by
the ZIL after an LWB write has completed.
Setting this will cause ZIL corruption on power loss
if a volatile out-of-order write cache is enabled.
.
.It Sy zil_replay_disable Ns = Ns Sy 0 Ns | Ns 1 Pq int
Disable intent logging replay.
Can be disabled for recovery from corrupted ZIL.
.
.It Sy zil_slog_bulk Ns = Ns Sy 67108864 Ns B Po 64 MiB Pc Pq u64
Limit SLOG write size per commit executed with synchronous priority.
Any writes above that will be executed with lower (asynchronous) priority
to limit potential SLOG device abuse by single active ZIL writer.
.
.It Sy zfs_zil_saxattr Ns = Ns Sy 1 Ns | Ns 0 Pq int
Setting this tunable to zero disables ZIL logging of new
.Sy xattr Ns = Ns Sy sa
records if the
.Sy org.openzfs:zilsaxattr
feature is enabled on the pool.
This would only be necessary to work around bugs in the ZIL logging or replay
code for this record type.
The tunable has no effect if the feature is disabled.
.
.It Sy zfs_embedded_slog_min_ms Ns = Ns Sy 64 Pq uint
Usually, one metaslab from each normal-class vdev is dedicated for use by
the ZIL to log synchronous writes.
However, if there are fewer than
.Sy zfs_embedded_slog_min_ms
metaslabs in the vdev, this functionality is disabled.
This ensures that we don't set aside an unreasonable amount of space for the
ZIL.
.
.It Sy zstd_earlyabort_pass Ns = Ns Sy 1 Pq uint
Whether heuristic for detection of incompressible data with zstd levels >= 3
using LZ4 and zstd-1 passes is enabled.
.
.It Sy zstd_abort_size Ns = Ns Sy 131072 Pq uint
Minimal uncompressed size (inclusive) of a record before the early abort
heuristic will be attempted.
.
.It Sy zio_deadman_log_all Ns = Ns Sy 0 Ns | Ns 1 Pq int
If non-zero, the zio deadman will produce debugging messages
.Pq see Sy zfs_dbgmsg_enable
for all zios, rather than only for leaf zios possessing a vdev.
This is meant to be used by developers to gain
diagnostic information for hang conditions which don't involve a mutex
or other locking primitive: typically conditions in which a thread in
the zio pipeline is looping indefinitely.
.
.It Sy zio_slow_io_ms Ns = Ns Sy 30000 Ns ms Po 30 s Pc Pq int
When an I/O operation takes more than this much time to complete,
it's marked as slow.
Each slow operation causes a delay zevent.
Slow I/O counters can be seen with
.Nm zpool Cm status Fl s .
.
.It Sy zio_dva_throttle_enabled Ns = Ns Sy 1 Ns | Ns 0 Pq int
Throttle block allocations in the I/O pipeline.
This allows for dynamic allocation distribution when devices are imbalanced.
When enabled, the maximum number of pending allocations per top-level vdev
is limited by
.Sy zfs_vdev_queue_depth_pct .
.
.It Sy zfs_xattr_compat Ns = Ns 0 Ns | Ns 1 Pq int
Control the naming scheme used when setting new xattrs in the user namespace.
If
.Sy 0
.Pq the default on Linux ,
user namespace xattr names are prefixed with the namespace, to be backwards
compatible with previous versions of ZFS on Linux.
If
.Sy 1
.Pq the default on Fx ,
user namespace xattr names are not prefixed, to be backwards compatible with
previous versions of ZFS on illumos and
.Fx .
.Pp
Either naming scheme can be read on this and future versions of ZFS, regardless
of this tunable, but legacy ZFS on illumos or
.Fx
are unable to read user namespace xattrs written in the Linux format, and
legacy versions of ZFS on Linux are unable to read user namespace xattrs written
in the legacy ZFS format.
.Pp
An existing xattr with the alternate naming scheme is removed when overwriting
the xattr so as to not accumulate duplicates.
.
.It Sy zio_requeue_io_start_cut_in_line Ns = Ns Sy 0 Ns | Ns 1 Pq int
Prioritize requeued I/O.
.
.It Sy zio_taskq_batch_pct Ns = Ns Sy 80 Ns % Pq uint
Percentage of online CPUs which will run a worker thread for I/O.
These workers are responsible for I/O work such as compression, encryption,
checksum and parity calculations.
Fractional number of CPUs will be rounded down.
.Pp
The default value of
.Sy 80%
was chosen to avoid using all CPUs which can result in
latency issues and inconsistent application performance,
especially when slower compression and/or checksumming is enabled.
Set value only applies to pools imported/created after that.
.
.It Sy zio_taskq_batch_tpq Ns = Ns Sy 0 Pq uint
Number of worker threads per taskq.
Higher values improve I/O ordering and CPU utilization,
while lower reduce lock contention.
Set value only applies to pools imported/created after that.
.Pp
If
.Sy 0 ,
generate a system-dependent value close to 6 threads per taskq.
Set value only applies to pools imported/created after that.
.
.It Sy zio_taskq_write_tpq Ns = Ns Sy 16 Pq uint
Determines the minimum number of threads per write issue taskq.
Higher values improve CPU utilization on high throughput,
while lower reduce taskq locks contention on high IOPS.
Set value only applies to pools imported/created after that.
.
.It Sy zio_taskq_read Ns = Ns Sy fixed,1,8 null scale null Pq charp
Set the queue and thread configuration for the IO read queues.
This is an advanced debugging parameter.
Don't change this unless you understand what it does.
Set values only apply to pools imported/created after that.
.
.It Sy zio_taskq_write Ns = Ns Sy sync null scale null Pq charp
Set the queue and thread configuration for the IO write queues.
This is an advanced debugging parameter.
Don't change this unless you understand what it does.
Set values only apply to pools imported/created after that.
.
.It Sy zvol_inhibit_dev Ns = Ns Sy 0 Ns | Ns 1 Pq uint
Do not create zvol device nodes.
This may slightly improve startup time on
systems with a very large number of zvols.
.
.It Sy zvol_major Ns = Ns Sy 230 Pq uint
Major number for zvol block devices.
.
.It Sy zvol_max_discard_blocks Ns = Ns Sy 16384 Pq long
Discard (TRIM) operations done on zvols will be done in batches of this
many blocks, where block size is determined by the
.Sy volblocksize
property of a zvol.
.
.It Sy zvol_prefetch_bytes Ns = Ns Sy 131072 Ns B Po 128 KiB Pc Pq uint
When adding a zvol to the system, prefetch this many bytes
from the start and end of the volume.
Prefetching these regions of the volume is desirable,
because they are likely to be accessed immediately by
.Xr blkid 8
or the kernel partitioner.
.
.It Sy zvol_request_sync Ns = Ns Sy 0 Ns | Ns 1 Pq uint
When processing I/O requests for a zvol, submit them synchronously.
This effectively limits the queue depth to
.Em 1
for each I/O submitter.
When unset, requests are handled asynchronously by a thread pool.
The number of requests which can be handled concurrently is controlled by
.Sy zvol_threads .
.Sy zvol_request_sync
is ignored when running on a kernel that supports block multiqueue
.Pq Li blk-mq .
.
.It Sy zvol_num_taskqs Ns = Ns Sy 0 Pq uint
Number of zvol taskqs.
If
.Sy 0
(the default) then scaling is done internally to prefer 6 threads per taskq.
This only applies on Linux.
.
.It Sy zvol_threads Ns = Ns Sy 0 Pq uint
The number of system wide threads to use for processing zvol block IOs.
If
.Sy 0
(the default) then internally set
.Sy zvol_threads
to the number of CPUs present or 32 (whichever is greater).
.
.It Sy zvol_blk_mq_threads Ns = Ns Sy 0 Pq uint
The number of threads per zvol to use for queuing IO requests.
This parameter will only appear if your kernel supports
.Li blk-mq
and is only read and assigned to a zvol at zvol load time.
If
.Sy 0
(the default) then internally set
.Sy zvol_blk_mq_threads
to the number of CPUs present.
.
.It Sy zvol_use_blk_mq Ns = Ns Sy 0 Ns | Ns 1 Pq uint
Set to
.Sy 1
to use the
.Li blk-mq
API for zvols.
Set to
.Sy 0
(the default) to use the legacy zvol APIs.
This setting can give better or worse zvol performance depending on
the workload.
This parameter will only appear if your kernel supports
.Li blk-mq
and is only read and assigned to a zvol at zvol load time.
.
.It Sy zvol_blk_mq_blocks_per_thread Ns = Ns Sy 8 Pq uint
If
.Sy zvol_use_blk_mq
is enabled, then process this number of
.Sy volblocksize Ns -sized blocks per zvol thread.
This tunable can be use to favor better performance for zvol reads (lower
values) or writes (higher values).
If set to
.Sy 0 ,
then the zvol layer will process the maximum number of blocks
per thread that it can.
This parameter will only appear if your kernel supports
.Li blk-mq
and is only applied at each zvol's load time.
.
.It Sy zvol_blk_mq_queue_depth Ns = Ns Sy 0 Pq uint
The queue_depth value for the zvol
.Li blk-mq
interface.
This parameter will only appear if your kernel supports
.Li blk-mq
and is only applied at each zvol's load time.
If
.Sy 0
(the default) then use the kernel's default queue depth.
Values are clamped to the kernel's
.Dv BLKDEV_MIN_RQ
and
.Dv BLKDEV_MAX_RQ Ns / Ns Dv BLKDEV_DEFAULT_RQ
limits.
.
.It Sy zvol_volmode Ns = Ns Sy 1 Pq uint
Defines zvol block devices behavior when
.Sy volmode Ns = Ns Sy default :
.Bl -tag -compact -offset 4n -width "a"
.It Sy 1
.No equivalent to Sy full
.It Sy 2
.No equivalent to Sy dev
.It Sy 3
.No equivalent to Sy none
.El
.
.It Sy zvol_enforce_quotas Ns = Ns Sy 0 Ns | Ns 1 Pq uint
Enable strict ZVOL quota enforcement.
The strict quota enforcement may have a performance impact.
.El
.
.Sh ZFS I/O SCHEDULER
ZFS issues I/O operations to leaf vdevs to satisfy and complete I/O operations.
The scheduler determines when and in what order those operations are issued.
The scheduler divides operations into five I/O classes,
prioritized in the following order: sync read, sync write, async read,
async write, and scrub/resilver.
Each queue defines the minimum and maximum number of concurrent operations
that may be issued to the device.
In addition, the device has an aggregate maximum,
.Sy zfs_vdev_max_active .
Note that the sum of the per-queue minima must not exceed the aggregate maximum.
If the sum of the per-queue maxima exceeds the aggregate maximum,
then the number of active operations may reach
.Sy zfs_vdev_max_active ,
in which case no further operations will be issued,
regardless of whether all per-queue minima have been met.
.Pp
For many physical devices, throughput increases with the number of
concurrent operations, but latency typically suffers.
Furthermore, physical devices typically have a limit
at which more concurrent operations have no
effect on throughput or can actually cause it to decrease.
.Pp
The scheduler selects the next operation to issue by first looking for an
I/O class whose minimum has not been satisfied.
Once all are satisfied and the aggregate maximum has not been hit,
the scheduler looks for classes whose maximum has not been satisfied.
Iteration through the I/O classes is done in the order specified above.
No further operations are issued
if the aggregate maximum number of concurrent operations has been hit,
or if there are no operations queued for an I/O class that has not hit its
maximum.
Every time an I/O operation is queued or an operation completes,
the scheduler looks for new operations to issue.
.Pp
In general, smaller
.Sy max_active Ns s
will lead to lower latency of synchronous operations.
Larger
.Sy max_active Ns s
may lead to higher overall throughput, depending on underlying storage.
.Pp
The ratio of the queues'
.Sy max_active Ns s
determines the balance of performance between reads, writes, and scrubs.
For example, increasing
.Sy zfs_vdev_scrub_max_active
will cause the scrub or resilver to complete more quickly,
but reads and writes to have higher latency and lower throughput.
.Pp
All I/O classes have a fixed maximum number of outstanding operations,
except for the async write class.
Asynchronous writes represent the data that is committed to stable storage
during the syncing stage for transaction groups.
Transaction groups enter the syncing state periodically,
so the number of queued async writes will quickly burst up
and then bleed down to zero.
Rather than servicing them as quickly as possible,
the I/O scheduler changes the maximum number of active async write operations
according to the amount of dirty data in the pool.
Since both throughput and latency typically increase with the number of
concurrent operations issued to physical devices, reducing the
burstiness in the number of simultaneous operations also stabilizes the
response time of operations from other queues, in particular synchronous ones.
In broad strokes, the I/O scheduler will issue more concurrent operations
from the async write queue as there is more dirty data in the pool.
.
.Ss Async Writes
The number of concurrent operations issued for the async write I/O class
follows a piece-wise linear function defined by a few adjustable points:
.Bd -literal
       |              o---------| <-- \fBzfs_vdev_async_write_max_active\fP
  ^    |             /^         |
  |    |            / |         |
active |           /  |         |
 I/O   |          /   |         |
count  |         /    |         |
       |        /     |         |
       |-------o      |         | <-- \fBzfs_vdev_async_write_min_active\fP
      0|_______^______|_________|
       0%      |      |       100% of \fBzfs_dirty_data_max\fP
               |      |
               |      `-- \fBzfs_vdev_async_write_active_max_dirty_percent\fP
               `--------- \fBzfs_vdev_async_write_active_min_dirty_percent\fP
.Ed
.Pp
Until the amount of dirty data exceeds a minimum percentage of the dirty
data allowed in the pool, the I/O scheduler will limit the number of
concurrent operations to the minimum.
As that threshold is crossed, the number of concurrent operations issued
increases linearly to the maximum at the specified maximum percentage
of the dirty data allowed in the pool.
.Pp
Ideally, the amount of dirty data on a busy pool will stay in the sloped
part of the function between
.Sy zfs_vdev_async_write_active_min_dirty_percent
and
.Sy zfs_vdev_async_write_active_max_dirty_percent .
If it exceeds the maximum percentage,
this indicates that the rate of incoming data is
greater than the rate that the backend storage can handle.
In this case, we must further throttle incoming writes,
as described in the next section.
.
.Sh ZFS TRANSACTION DELAY
We delay transactions when we've determined that the backend storage
isn't able to accommodate the rate of incoming writes.
.Pp
If there is already a transaction waiting, we delay relative to when
that transaction will finish waiting.
This way the calculated delay time
is independent of the number of threads concurrently executing transactions.
.Pp
If we are the only waiter, wait relative to when the transaction started,
rather than the current time.
This credits the transaction for "time already served",
e.g. reading indirect blocks.
.Pp
The minimum time for a transaction to take is calculated as
.D1 min_time = min( Ns Sy zfs_delay_scale No \(mu Po Sy dirty No \- Sy min Pc / Po Sy max No \- Sy dirty Pc , 100ms)
.Pp
The delay has two degrees of freedom that can be adjusted via tunables.
The percentage of dirty data at which we start to delay is defined by
.Sy zfs_delay_min_dirty_percent .
This should typically be at or above
.Sy zfs_vdev_async_write_active_max_dirty_percent ,
so that we only start to delay after writing at full speed
has failed to keep up with the incoming write rate.
The scale of the curve is defined by
.Sy zfs_delay_scale .
Roughly speaking, this variable determines the amount of delay at the midpoint
of the curve.
.Bd -literal
delay
 10ms +-------------------------------------------------------------*+
      |                                                             *|
  9ms +                                                             *+
      |                                                             *|
  8ms +                                                             *+
      |                                                            * |
  7ms +                                                            * +
      |                                                            * |
  6ms +                                                            * +
      |                                                            * |
  5ms +                                                           *  +
      |                                                           *  |
  4ms +                                                           *  +
      |                                                           *  |
  3ms +                                                          *   +
      |                                                          *   |
  2ms +                                              (midpoint) *    +
      |                                                  |    **     |
  1ms +                                                  v ***       +
      |             \fBzfs_delay_scale\fP ---------->     ********         |
    0 +-------------------------------------*********----------------+
      0%                    <- \fBzfs_dirty_data_max\fP ->               100%
.Ed
.Pp
Note, that since the delay is added to the outstanding time remaining on the
most recent transaction it's effectively the inverse of IOPS.
Here, the midpoint of
.Em 500 us
translates to
.Em 2000 IOPS .
The shape of the curve
was chosen such that small changes in the amount of accumulated dirty data
in the first three quarters of the curve yield relatively small differences
in the amount of delay.
.Pp
The effects can be easier to understand when the amount of delay is
represented on a logarithmic scale:
.Bd -literal
delay
100ms +-------------------------------------------------------------++
      +                                                              +
      |                                                              |
      +                                                             *+
 10ms +                                                             *+
      +                                                           ** +
      |                                              (midpoint)  **  |
      +                                                  |     **    +
  1ms +                                                  v ****      +
      +             \fBzfs_delay_scale\fP ---------->        *****         +
      |                                             ****             |
      +                                          ****                +
100us +                                        **                    +
      +                                       *                      +
      |                                      *                       |
      +                                     *                        +
 10us +                                     *                        +
      +                                                              +
      |                                                              |
      +                                                              +
      +--------------------------------------------------------------+
      0%                    <- \fBzfs_dirty_data_max\fP ->               100%
.Ed
.Pp
Note here that only as the amount of dirty data approaches its limit does
the delay start to increase rapidly.
The goal of a properly tuned system should be to keep the amount of dirty data
out of that range by first ensuring that the appropriate limits are set
for the I/O scheduler to reach optimal throughput on the back-end storage,
and then by changing the value of
.Sy zfs_delay_scale
to increase the steepness of the curve.