xref: /freebsd/sys/contrib/openzfs/man/man7/zpool-features.7 (revision 924226fba12cc9a228c73b956e1b7fa24c60b055)
1.\"
2.\" Copyright (c) 2012, 2018 by Delphix. All rights reserved.
3.\" Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
4.\" Copyright (c) 2014, Joyent, Inc. All rights reserved.
5.\" The contents of this file are subject to the terms of the Common Development
6.\" and Distribution License (the "License").  You may not use this file except
7.\" in compliance with the License. You can obtain a copy of the license at
8.\" usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing.
9.\"
10.\" See the License for the specific language governing permissions and
11.\" limitations under the License. When distributing Covered Code, include this
12.\" CDDL HEADER in each file and include the License file at
13.\" usr/src/OPENSOLARIS.LICENSE.  If applicable, add the following below this
14.\" CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your
15.\" own identifying information:
16.\" Portions Copyright [yyyy] [name of copyright owner]
17.\" Copyright (c) 2019, Klara Inc.
18.\" Copyright (c) 2019, Allan Jude
19.\" Copyright (c) 2021, Colm Buckley <colm@tuatha.org>
20.\"
21.Dd May 31, 2021
22.Dt ZPOOL-FEATURES 7
23.Os
24.
25.Sh NAME
26.Nm zpool-features
27.Nd description of ZFS pool features
28.
29.Sh DESCRIPTION
30ZFS pool on-disk format versions are specified via "features" which replace
31the old on-disk format numbers (the last supported on-disk format number is 28).
32To enable a feature on a pool use the
33.Nm zpool Cm upgrade ,
34or set the
35.Sy feature Ns @ Ns Ar feature-name
36property to
37.Sy enabled .
38Please also see the
39.Sx Compatibility feature sets
40section for information on how sets of features may be enabled together.
41.Pp
42The pool format does not affect file system version compatibility or the ability
43to send file systems between pools.
44.Pp
45Since most features can be enabled independently of each other, the on-disk
46format of the pool is specified by the set of all features marked as
47.Sy active
48on the pool.
49If the pool was created by another software version
50this set may include unsupported features.
51.
52.Ss Identifying features
53Every feature has a GUID of the form
54.Ar com.example : Ns Ar feature-name .
55The reversed DNS name ensures that the feature's GUID is unique across all ZFS
56implementations.
57When unsupported features are encountered on a pool they will
58be identified by their GUIDs.
59Refer to the documentation for the ZFS
60implementation that created the pool for information about those features.
61.Pp
62Each supported feature also has a short name.
63By convention a feature's short name is the portion of its GUID which follows the
64.Sq \&:
65(i.e.
66.Ar com.example : Ns Ar feature-name
67would have the short name
68.Ar feature-name ) ,
69however a feature's short name may differ across ZFS implementations if
70following the convention would result in name conflicts.
71.
72.Ss Feature states
73Features can be in one of three states:
74.Bl -tag -width "disabled"
75.It Sy active
76This feature's on-disk format changes are in effect on the pool.
77Support for this feature is required to import the pool in read-write mode.
78If this feature is not read-only compatible,
79support is also required to import the pool in read-only mode
80.Pq see Sx Read-only compatibility .
81.It Sy enabled
82An administrator has marked this feature as enabled on the pool, but the
83feature's on-disk format changes have not been made yet.
84The pool can still be imported by software that does not support this feature,
85but changes may be made to the on-disk format at any time
86which will move the feature to the
87.Sy active
88state.
89Some features may support returning to the
90.Sy enabled
91state after becoming
92.Sy active .
93See feature-specific documentation for details.
94.It Sy disabled
95This feature's on-disk format changes have not been made and will not be made
96unless an administrator moves the feature to the
97.Sy enabled
98state.
99Features cannot be disabled once they have been enabled.
100.El
101.Pp
102The state of supported features is exposed through pool properties of the form
103.Sy feature Ns @ Ns Ar short-name .
104.
105.Ss Read-only compatibility
106Some features may make on-disk format changes that do not interfere with other
107software's ability to read from the pool.
108These features are referred to as
109.Dq read-only compatible .
110If all unsupported features on a pool are read-only compatible,
111the pool can be imported in read-only mode by setting the
112.Sy readonly
113property during import (see
114.Xr zpool-import 8
115for details on importing pools).
116.
117.Ss Unsupported features
118For each unsupported feature enabled on an imported pool, a pool property
119named
120.Sy unsupported Ns @ Ns Ar feature-name
121will indicate why the import was allowed despite the unsupported feature.
122Possible values for this property are:
123.Bl -tag -width "readonly"
124.It Sy inactive
125The feature is in the
126.Sy enabled
127state and therefore the pool's on-disk
128format is still compatible with software that does not support this feature.
129.It Sy readonly
130The feature is read-only compatible and the pool has been imported in
131read-only mode.
132.El
133.
134.Ss Feature dependencies
135Some features depend on other features being enabled in order to function.
136Enabling a feature will automatically enable any features it depends on.
137.
138.Ss Compatibility feature sets
139It is sometimes necessary for a pool to maintain compatibility with a
140specific on-disk format, by enabling and disabling particular features.
141The
142.Sy compatibility
143feature facilitates this by allowing feature sets to be read from text files.
144When set to
145.Sy off
146(the default), compatibility feature sets are disabled
147(i.e. all features are enabled); when set to
148.Sy legacy ,
149no features are enabled.
150When set to a comma-separated list of filenames
151(each filename may either be an absolute path, or relative to
152.Pa /etc/zfs/compatibility.d
153or
154.Pa /usr/share/zfs/compatibility.d ) ,
155the lists of requested features are read from those files,
156separated by whitespace and/or commas.
157Only features present in all files are enabled.
158.Pp
159Simple sanity checks are applied to the files:
160they must be between 1 B and 16 KiB in size, and must end with a newline character.
161.Pp
162The requested features are applied when a pool is created using
163.Nm zpool Cm create Fl o Sy compatibility Ns = Ns Ar …
164and controls which features are enabled when using
165.Nm zpool Cm upgrade .
166.Nm zpool Cm status
167will not show a warning about disabled features which are not part
168of the requested feature set.
169.Pp
170The special value
171.Sy legacy
172prevents any features from being enabled, either via
173.Nm zpool Cm upgrade
174or
175.Nm zpool Cm set Sy feature Ns @ Ns Ar feature-name Ns = Ns Sy enabled .
176This setting also prevents pools from being upgraded to newer on-disk versions.
177This is a safety measure to prevent new features from being
178accidentally enabled, breaking compatibility.
179.Pp
180By convention, compatibility files in
181.Pa /usr/share/zfs/compatibility.d
182are provided by the distribution, and include feature sets
183supported by important versions of popular distributions, and feature
184sets commonly supported at the start of each year.
185Compatibility files in
186.Pa /etc/zfs/compatibility.d ,
187if present, will take precedence over files with the same name in
188.Pa /usr/share/zfs/compatibility.d .
189.Pp
190If an unrecognized feature is found in these files, an error message will
191be shown.
192If the unrecognized feature is in a file in
193.Pa /etc/zfs/compatibility.d ,
194this is treated as an error and processing will stop.
195If the unrecognized feature is under
196.Pa /usr/share/zfs/compatibility.d ,
197this is treated as a warning and processing will continue.
198This difference is to allow distributions to include features
199which might not be recognized by the currently-installed binaries.
200.Pp
201Compatibility files may include comments:
202any text from
203.Sq #
204to the end of the line is ignored.
205.Pp
206.Sy Example :
207.Bd -literal -compact -offset 4n
208.No example# Nm cat Pa /usr/share/zfs/compatibility.d/grub2
209# Features which are supported by GRUB2
210async_destroy
211bookmarks
212embedded_data
213empty_bpobj
214enabled_txg
215extensible_dataset
216filesystem_limits
217hole_birth
218large_blocks
219lz4_compress
220spacemap_histogram
221
222.No example# Nm zpool Cm create Fl o Sy compatibility Ns = Ns Ar grub2 Ar bootpool Ar vdev
223.Ed
224.Pp
225See
226.Xr zpool-create 8
227and
228.Xr zpool-upgrade 8
229for more information on how these commands are affected by feature sets.
230.
231.de feature
232.It Sy \\$2
233.Bl -tag -compact -width "READ-ONLY COMPATIBLE"
234.It GUID
235.Sy \\$1:\\$2
236.if !"\\$4"" \{\
237.It DEPENDENCIES
238\fB\\$4\fP\c
239.if !"\\$5"" , \fB\\$5\fP\c
240.if !"\\$6"" , \fB\\$6\fP\c
241.if !"\\$7"" , \fB\\$7\fP\c
242.if !"\\$8"" , \fB\\$8\fP\c
243.if !"\\$9"" , \fB\\$9\fP\c
244.\}
245.It READ-ONLY COMPATIBLE
246\\$3
247.El
248.Pp
249..
250.
251.ds instant-never \
252.No This feature becomes Sy active No as soon as it is enabled \
253and will never return to being Sy enabled .
254.
255.ds remount-upgrade \
256.No Each filesystem will be upgraded automatically when remounted, \
257or when a new file is created under that filesystem. \
258The upgrade can also be triggered on filesystems via \
259Nm zfs Cm set Sy version Ns = Ns Sy current Ar fs . \
260No The upgrade process runs in the background and may take a while to complete \
261for filesystems containing large amounts of files.
262.
263.de checksum-spiel
264When the
265.Sy \\$1
266feature is set to
267.Sy enabled ,
268the administrator can turn on the
269.Sy \\$1
270checksum on any dataset using
271.Nm zfs Cm set Sy checksum Ns = Ns Sy \\$1 Ar dset
272.Po see Xr zfs-set 8 Pc .
273This feature becomes
274.Sy active
275once a
276.Sy checksum
277property has been set to
278.Sy \\$1 ,
279and will return to being
280.Sy enabled
281once all filesystems that have ever had their checksum set to
282.Sy \\$1
283are destroyed.
284..
285.
286.Sh FEATURES
287The following features are supported on this system:
288.Bl -tag -width Ds
289.feature org.zfsonlinux allocation_classes yes
290This feature enables support for separate allocation classes.
291.Pp
292This feature becomes
293.Sy active
294when a dedicated allocation class vdev (dedup or special) is created with the
295.Nm zpool Cm create No or Nm zpool Cm add No commands .
296With device removal, it can be returned to the
297.Sy enabled
298state if all the dedicated allocation class vdevs are removed.
299.
300.feature com.delphix async_destroy yes
301Destroying a file system requires traversing all of its data in order to
302return its used space to the pool.
303Without
304.Sy async_destroy ,
305the file system is not fully removed until all space has been reclaimed.
306If the destroy operation is interrupted by a reboot or power outage,
307the next attempt to open the pool will need to complete the destroy
308operation synchronously.
309.Pp
310When
311.Sy async_destroy
312is enabled, the file system's data will be reclaimed by a background process,
313allowing the destroy operation to complete
314without traversing the entire file system.
315The background process is able to resume
316interrupted destroys after the pool has been opened, eliminating the need
317to finish interrupted destroys as part of the open operation.
318The amount of space remaining to be reclaimed by the background process
319is available through the
320.Sy freeing
321property.
322.Pp
323This feature is only
324.Sy active
325while
326.Sy freeing
327is non-zero.
328.
329.feature org.openzfs blake3 no extensible_dataset
330This feature enables the use of the BLAKE3 hash algorithm for checksum and dedup.
331BLAKE3 is a secure hash algorithm focused on high performance.
332.Pp
333.checksum-spiel blake3
334.
335.feature com.delphix bookmarks yes extensible_dataset
336This feature enables use of the
337.Nm zfs Cm bookmark
338command.
339.Pp
340This feature is
341.Sy active
342while any bookmarks exist in the pool.
343All bookmarks in the pool can be listed by running
344.Nm zfs Cm list Fl t Sy bookmark Fl r Ar poolname .
345.
346.feature com.datto bookmark_v2 no bookmark extensible_dataset
347This feature enables the creation and management of larger bookmarks which are
348needed for other features in ZFS.
349.Pp
350This feature becomes
351.Sy active
352when a v2 bookmark is created and will be returned to the
353.Sy enabled
354state when all v2 bookmarks are destroyed.
355.
356.feature com.delphix bookmark_written no bookmark extensible_dataset bookmark_v2
357This feature enables additional bookmark accounting fields, enabling the
358.Sy written Ns # Ns Ar bookmark
359property (space written since a bookmark) and estimates of
360send stream sizes for incrementals from bookmarks.
361.Pp
362This feature becomes
363.Sy active
364when a bookmark is created and will be
365returned to the
366.Sy enabled
367state when all bookmarks with these fields are destroyed.
368.
369.feature org.openzfs device_rebuild yes
370This feature enables the ability for the
371.Nm zpool Cm attach
372and
373.Nm zpool Cm replace
374commands to perform sequential reconstruction
375(instead of healing reconstruction) when resilvering.
376.Pp
377Sequential reconstruction resilvers a device in LBA order without immediately
378verifying the checksums.
379Once complete, a scrub is started, which then verifies the checksums.
380This approach allows full redundancy to be restored to the pool
381in the minimum amount of time.
382This two-phase approach will take longer than a healing resilver
383when the time to verify the checksums is included.
384However, unless there is additional pool damage,
385no checksum errors should be reported by the scrub.
386This feature is incompatible with raidz configurations.
387.
388This feature becomes
389.Sy active
390while a sequential resilver is in progress, and returns to
391.Sy enabled
392when the resilver completes.
393.
394.feature com.delphix device_removal no
395This feature enables the
396.Nm zpool Cm remove
397command to remove top-level vdevs,
398evacuating them to reduce the total size of the pool.
399.Pp
400This feature becomes
401.Sy active
402when the
403.Nm zpool Cm remove
404command is used
405on a top-level vdev, and will never return to being
406.Sy enabled .
407.
408.feature org.openzfs draid no
409This feature enables use of the
410.Sy draid
411vdev type.
412dRAID is a variant of RAID-Z which provides integrated distributed
413hot spares that allow faster resilvering while retaining the benefits of RAID-Z.
414Data, parity, and spare space are organized in redundancy groups
415and distributed evenly over all of the devices.
416.Pp
417This feature becomes
418.Sy active
419when creating a pool which uses the
420.Sy draid
421vdev type, or when adding a new
422.Sy draid
423vdev to an existing pool.
424.
425.feature org.illumos edonr no extensible_dataset
426This feature enables the use of the Edon-R hash algorithm for checksum,
427including for nopwrite (if compression is also enabled, an overwrite of
428a block whose checksum matches the data being written will be ignored).
429In an abundance of caution, Edon-R requires verification when used with
430dedup:
431.Nm zfs Cm set Sy dedup Ns = Ns Sy edonr , Ns Sy verify
432.Po see Xr zfs-set 8 Pc .
433.Pp
434Edon-R is a very high-performance hash algorithm that was part
435of the NIST SHA-3 competition.
436It provides extremely high hash performance (over 350% faster than SHA-256),
437but was not selected because of its unsuitability
438as a general purpose secure hash algorithm.
439This implementation utilizes the new salted checksumming functionality
440in ZFS, which means that the checksum is pre-seeded with a secret
441256-bit random key (stored on the pool) before being fed the data block
442to be checksummed.
443Thus the produced checksums are unique to a given pool,
444preventing hash collision attacks on systems with dedup.
445.Pp
446.checksum-spiel edonr
447.
448.feature com.delphix embedded_data no
449This feature improves the performance and compression ratio of
450highly-compressible blocks.
451Blocks whose contents can compress to 112 bytes
452or smaller can take advantage of this feature.
453.Pp
454When this feature is enabled, the contents of highly-compressible blocks are
455stored in the block "pointer" itself (a misnomer in this case, as it contains
456the compressed data, rather than a pointer to its location on disk).
457Thus the space of the block (one sector, typically 512 B or 4 KiB) is saved,
458and no additional I/O is needed to read and write the data block.
459.
460\*[instant-never]
461.
462.feature com.delphix empty_bpobj yes
463This feature increases the performance of creating and using a large
464number of snapshots of a single filesystem or volume, and also reduces
465the disk space required.
466.Pp
467When there are many snapshots, each snapshot uses many Block Pointer
468Objects (bpobjs) to track blocks associated with that snapshot.
469However, in common use cases, most of these bpobjs are empty.
470This feature allows us to create each bpobj on-demand,
471thus eliminating the empty bpobjs.
472.Pp
473This feature is
474.Sy active
475while there are any filesystems, volumes,
476or snapshots which were created after enabling this feature.
477.
478.feature com.delphix enabled_txg yes
479Once this feature is enabled, ZFS records the transaction group number
480in which new features are enabled.
481This has no user-visible impact, but other features may depend on this feature.
482.Pp
483This feature becomes
484.Sy active
485as soon as it is enabled and will never return to being
486.Sy enabled .
487.
488.feature com.datto encryption no bookmark_v2 extensible_dataset
489This feature enables the creation and management of natively encrypted datasets.
490.Pp
491This feature becomes
492.Sy active
493when an encrypted dataset is created and will be returned to the
494.Sy enabled
495state when all datasets that use this feature are destroyed.
496.
497.feature com.delphix extensible_dataset no
498This feature allows more flexible use of internal ZFS data structures,
499and exists for other features to depend on.
500.Pp
501This feature will be
502.Sy active
503when the first dependent feature uses it, and will be returned to the
504.Sy enabled
505state when all datasets that use this feature are destroyed.
506.
507.feature com.joyent filesystem_limits yes extensible_dataset
508This feature enables filesystem and snapshot limits.
509These limits can be used to control how many filesystems and/or snapshots
510can be created at the point in the tree on which the limits are set.
511.Pp
512This feature is
513.Sy active
514once either of the limit properties has been set on a dataset
515and will never return to being
516.Sy enabled .
517.
518.feature com.delphix head_errlog no
519This feature enables the upgraded version of errlog, which required an on-disk
520error log format change.
521Now the error log of each head dataset is stored separately in the zap object
522and keyed by the head id.
523With this feature enabled, every dataset affected by an error block is listed
524in the output of
525.Nm zpool Cm status .
526.Pp
527\*[instant-never]
528.
529.feature com.delphix hole_birth no enabled_txg
530This feature has/had bugs, the result of which is that, if you do a
531.Nm zfs Cm send Fl i
532.Pq or Fl R , No since it uses Fl i
533from an affected dataset, the receiving party will not see any checksum
534or other errors, but the resulting destination snapshot
535will not match the source.
536Its use by
537.Nm zfs Cm send Fl i
538has been disabled by default
539.Pq see Sy send_holes_without_birth_time No in Xr zfs 4 .
540.Pp
541This feature improves performance of incremental sends
542.Pq Nm zfs Cm send Fl i
543and receives for objects with many holes.
544The most common case of hole-filled objects is zvols.
545.Pp
546An incremental send stream from snapshot
547.Sy A No to snapshot Sy B
548contains information about every block that changed between
549.Sy A No and Sy B .
550Blocks which did not change between those snapshots can be
551identified and omitted from the stream using a piece of metadata called
552the "block birth time", but birth times are not recorded for holes
553(blocks filled only with zeroes).
554Since holes created after
555.Sy A No cannot be distinguished from holes created before Sy A ,
556information about every hole in the entire filesystem or zvol
557is included in the send stream.
558.Pp
559For workloads where holes are rare this is not a problem.
560However, when incrementally replicating filesystems or zvols with many holes
561(for example a zvol formatted with another filesystem) a lot of time will
562be spent sending and receiving unnecessary information about holes that
563already exist on the receiving side.
564.Pp
565Once the
566.Sy hole_birth
567feature has been enabled the block birth times
568of all new holes will be recorded.
569Incremental sends between snapshots created after this feature is enabled
570will use this new metadata to avoid sending information about holes that
571already exist on the receiving side.
572.Pp
573\*[instant-never]
574.
575.feature org.open-zfs large_blocks no extensible_dataset
576This feature allows the record size on a dataset to be set larger than 128 KiB.
577.Pp
578This feature becomes
579.Sy active
580once a dataset contains a file with a block size larger than 128 KiB,
581and will return to being
582.Sy enabled
583once all filesystems that have ever had their recordsize larger than 128 KiB
584are destroyed.
585.
586.feature org.zfsonlinux large_dnode no extensible_dataset
587This feature allows the size of dnodes in a dataset to be set larger than 512 B.
588.
589This feature becomes
590.Sy active
591once a dataset contains an object with a dnode larger than 512 B,
592which occurs as a result of setting the
593.Sy dnodesize
594dataset property to a value other than
595.Sy legacy .
596The feature will return to being
597.Sy enabled
598once all filesystems that have ever contained a dnode larger than 512 B
599are destroyed.
600Large dnodes allow more data to be stored in the bonus buffer,
601thus potentially improving performance by avoiding the use of spill blocks.
602.
603.feature com.delphix livelist yes
604This feature allows clones to be deleted faster than the traditional method
605when a large number of random/sparse writes have been made to the clone.
606All blocks allocated and freed after a clone is created are tracked by the
607the clone's livelist which is referenced during the deletion of the clone.
608The feature is activated when a clone is created and remains
609.Sy active
610until all clones have been destroyed.
611.
612.feature com.delphix log_spacemap yes com.delphix:spacemap_v2
613This feature improves performance for heavily-fragmented pools,
614especially when workloads are heavy in random-writes.
615It does so by logging all the metaslab changes on a single spacemap every TXG
616instead of scattering multiple writes to all the metaslab spacemaps.
617.Pp
618\*[instant-never]
619.
620.feature org.illumos lz4_compress no
621.Sy lz4
622is a high-performance real-time compression algorithm that
623features significantly faster compression and decompression as well as a
624higher compression ratio than the older
625.Sy lzjb
626compression.
627Typically,
628.Sy lz4
629compression is approximately 50% faster on compressible data and 200% faster
630on incompressible data than
631.Sy lzjb .
632It is also approximately 80% faster on decompression,
633while giving approximately a 10% better compression ratio.
634.Pp
635When the
636.Sy lz4_compress
637feature is set to
638.Sy enabled ,
639the administrator can turn on
640.Sy lz4
641compression on any dataset on the pool using the
642.Xr zfs-set 8
643command.
644All newly written metadata will be compressed with the
645.Sy lz4
646algorithm.
647.Pp
648\*[instant-never]
649.
650.feature com.joyent multi_vdev_crash_dump no
651This feature allows a dump device to be configured with a pool comprised
652of multiple vdevs.
653Those vdevs may be arranged in any mirrored or raidz configuration.
654.Pp
655When the
656.Sy multi_vdev_crash_dump
657feature is set to
658.Sy enabled ,
659the administrator can use
660.Xr dumpadm 1M
661to configure a dump device on a pool comprised of multiple vdevs.
662.Pp
663Under
664.Fx
665and Linux this feature is unused, but registered for compatibility.
666New pools created on these systems will have the feature
667.Sy enabled
668but will never transition to
669.Sy active ,
670as this functionality is not required for crash dump support.
671Existing pools where this feature is
672.Sy active
673can be imported.
674.
675.feature com.delphix obsolete_counts yes device_removal
676This feature is an enhancement of
677.Sy device_removal ,
678which will over time reduce the memory used to track removed devices.
679When indirect blocks are freed or remapped,
680we note that their part of the indirect mapping is "obsolete" – no longer needed.
681.Pp
682This feature becomes
683.Sy active
684when the
685.Nm zpool Cm remove
686command is used on a top-level vdev, and will never return to being
687.Sy enabled .
688.
689.feature org.zfsonlinux project_quota yes extensible_dataset
690This feature allows administrators to account the spaces and objects usage
691information against the project identifier (ID).
692.Pp
693The project ID is an object-based attribute.
694When upgrading an existing filesystem,
695objects without a project ID will be assigned a zero project ID.
696When this feature is enabled, newly created objects inherit
697their parent directories' project ID if the parent's inherit flag is set
698.Pq via Nm chattr Sy [+-]P No or Nm zfs Cm project Fl s Ns | Ns Fl C .
699Otherwise, the new object's project ID will be zero.
700An object's project ID can be changed at any time by the owner
701(or privileged user) via
702.Nm chattr Fl p Ar prjid
703or
704.Nm zfs Cm project Fl p Ar prjid .
705.Pp
706This feature will become
707.Sy active
708as soon as it is enabled and will never return to being
709.Sy disabled .
710\*[remount-upgrade]
711.
712.feature com.delphix redaction_bookmarks no bookmarks extensible_dataset
713This feature enables the use of redacted
714.Nm zfs Cm send Ns s ,
715which create redaction bookmarks storing the list of blocks
716redacted by the send that created them.
717For more information about redacted sends, see
718.Xr zfs-send 8 .
719.
720.feature com.delphix redacted_datasets no extensible_dataset
721This feature enables the receiving of redacted
722.Nm zfs Cm send
723streams, which create redacted datasets when received.
724These datasets are missing some of their blocks,
725and so cannot be safely mounted, and their contents cannot be safely read.
726For more information about redacted receives, see
727.Xr zfs-send 8 .
728.
729.feature com.datto resilver_defer yes
730This feature allows ZFS to postpone new resilvers if an existing one is already
731in progress.
732Without this feature, any new resilvers will cause the currently
733running one to be immediately restarted from the beginning.
734.Pp
735This feature becomes
736.Sy active
737once a resilver has been deferred, and returns to being
738.Sy enabled
739when the deferred resilver begins.
740.
741.feature org.illumos sha512 no extensible_dataset
742This feature enables the use of the SHA-512/256 truncated hash algorithm
743(FIPS 180-4) for checksum and dedup.
744The native 64-bit arithmetic of SHA-512 provides an approximate 50%
745performance boost over SHA-256 on 64-bit hardware
746and is thus a good minimum-change replacement candidate
747for systems where hash performance is important,
748but these systems cannot for whatever reason utilize the faster
749.Sy skein No and Sy edonr
750algorithms.
751.Pp
752.checksum-spiel sha512
753.
754.feature org.illumos skein no extensible_dataset
755This feature enables the use of the Skein hash algorithm for checksum and dedup.
756Skein is a high-performance secure hash algorithm that was a
757finalist in the NIST SHA-3 competition.
758It provides a very high security margin and high performance on 64-bit hardware
759(80% faster than SHA-256).
760This implementation also utilizes the new salted checksumming
761functionality in ZFS, which means that the checksum is pre-seeded with a
762secret 256-bit random key (stored on the pool) before being fed the data
763block to be checksummed.
764Thus the produced checksums are unique to a given pool,
765preventing hash collision attacks on systems with dedup.
766.Pp
767.checksum-spiel skein
768.
769.feature com.delphix spacemap_histogram yes
770This features allows ZFS to maintain more information about how free space
771is organized within the pool.
772If this feature is
773.Sy enabled ,
774it will be activated when a new space map object is created, or
775an existing space map is upgraded to the new format,
776and never returns back to being
777.Sy enabled .
778.
779.feature com.delphix spacemap_v2 yes
780This feature enables the use of the new space map encoding which
781consists of two words (instead of one) whenever it is advantageous.
782The new encoding allows space maps to represent large regions of
783space more efficiently on-disk while also increasing their maximum
784addressable offset.
785.Pp
786This feature becomes
787.Sy active
788once it is
789.Sy enabled ,
790and never returns back to being
791.Sy enabled .
792.
793.feature org.zfsonlinux userobj_accounting yes extensible_dataset
794This feature allows administrators to account the object usage information
795by user and group.
796.Pp
797\*[instant-never]
798\*[remount-upgrade]
799.
800.feature org.openzfs zilsaxattr yes extensible_dataset
801This feature enables
802.Sy xattr Ns = Ns Sy sa
803extended attribute logging in the ZIL.
804If enabled, extended attribute changes
805.Pq both Sy xattrdir Ns = Ns Sy dir No and Sy xattr Ns = Ns Sy sa
806are guaranteed to be durable if either the dataset had
807.Sy sync Ns = Ns Sy always
808set at the time the changes were made, or
809.Xr sync 2
810is called on the dataset after the changes were made.
811.Pp
812This feature becomes
813.Sy active
814when a ZIL is created for at least one dataset and will be returned to the
815.Sy enabled
816state when it is destroyed for all datasets that use this feature.
817.
818.feature com.delphix zpool_checkpoint yes
819This feature enables the
820.Nm zpool Cm checkpoint
821command that can checkpoint the state of the pool
822at the time it was issued and later rewind back to it or discard it.
823.Pp
824This feature becomes
825.Sy active
826when the
827.Nm zpool Cm checkpoint
828command is used to checkpoint the pool.
829The feature will only return back to being
830.Sy enabled
831when the pool is rewound or the checkpoint has been discarded.
832.
833.feature org.freebsd zstd_compress no extensible_dataset
834.Sy zstd
835is a high-performance compression algorithm that features a
836combination of high compression ratios and high speed.
837Compared to
838.Sy gzip ,
839.Sy zstd
840offers slightly better compression at much higher speeds.
841Compared to
842.Sy lz4 ,
843.Sy zstd
844offers much better compression while being only modestly slower.
845Typically,
846.Sy zstd
847compression speed ranges from 250 to 500 MB/s per thread
848and decompression speed is over 1 GB/s per thread.
849.Pp
850When the
851.Sy zstd
852feature is set to
853.Sy enabled ,
854the administrator can turn on
855.Sy zstd
856compression of any dataset using
857.Nm zfs Cm set Sy compress Ns = Ns Sy zstd Ar dset
858.Po see Xr zfs-set 8 Pc .
859This feature becomes
860.Sy active
861once a
862.Sy compress
863property has been set to
864.Sy zstd ,
865and will return to being
866.Sy enabled
867once all filesystems that have ever had their
868.Sy compress
869property set to
870.Sy zstd
871are destroyed.
872.El
873.
874.Sh SEE ALSO
875.Xr zpool 8
876