xref: /freebsd/sys/contrib/openzfs/man/man7/zpool-features.7 (revision 66fd12cf4896eb08ad8e7a2627537f84ead84dd3)
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 https://opensource.org/licenses/CDDL-1.0.
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 June 23, 2022
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
31.Dq features
32which replace the old on-disk format numbers
33.Pq the last supported on-disk format number is 28 .
34To enable a feature on a pool use the
35.Nm zpool Cm upgrade ,
36or set the
37.Sy feature Ns @ Ns Ar feature-name
38property to
39.Sy enabled .
40Please also see the
41.Sx Compatibility feature sets
42section for information on how sets of features may be enabled together.
43.Pp
44The pool format does not affect file system version compatibility or the ability
45to send file systems between pools.
46.Pp
47Since most features can be enabled independently of each other, the on-disk
48format of the pool is specified by the set of all features marked as
49.Sy active
50on the pool.
51If the pool was created by another software version
52this set may include unsupported features.
53.
54.Ss Identifying features
55Every feature has a GUID of the form
56.Ar com.example : Ns Ar feature-name .
57The reversed DNS name ensures that the feature's GUID is unique across all ZFS
58implementations.
59When unsupported features are encountered on a pool they will
60be identified by their GUIDs.
61Refer to the documentation for the ZFS
62implementation that created the pool for information about those features.
63.Pp
64Each supported feature also has a short name.
65By convention a feature's short name is the portion of its GUID which follows
66the
67.Sq \&:
68.Po
69i.e.
70.Ar com.example : Ns Ar feature-name
71would have the short name
72.Ar feature-name
73.Pc ,
74however a feature's short name may differ across ZFS implementations if
75following the convention would result in name conflicts.
76.
77.Ss Feature states
78Features can be in one of three states:
79.Bl -tag -width "disabled"
80.It Sy active
81This feature's on-disk format changes are in effect on the pool.
82Support for this feature is required to import the pool in read-write mode.
83If this feature is not read-only compatible,
84support is also required to import the pool in read-only mode
85.Pq see Sx Read-only compatibility .
86.It Sy enabled
87An administrator has marked this feature as enabled on the pool, but the
88feature's on-disk format changes have not been made yet.
89The pool can still be imported by software that does not support this feature,
90but changes may be made to the on-disk format at any time
91which will move the feature to the
92.Sy active
93state.
94Some features may support returning to the
95.Sy enabled
96state after becoming
97.Sy active .
98See feature-specific documentation for details.
99.It Sy disabled
100This feature's on-disk format changes have not been made and will not be made
101unless an administrator moves the feature to the
102.Sy enabled
103state.
104Features cannot be disabled once they have been enabled.
105.El
106.Pp
107The state of supported features is exposed through pool properties of the form
108.Sy feature Ns @ Ns Ar short-name .
109.
110.Ss Read-only compatibility
111Some features may make on-disk format changes that do not interfere with other
112software's ability to read from the pool.
113These features are referred to as
114.Dq read-only compatible .
115If all unsupported features on a pool are read-only compatible,
116the pool can be imported in read-only mode by setting the
117.Sy readonly
118property during import
119.Po see
120.Xr zpool-import 8
121for details on importing pools
122.Pc .
123.
124.Ss Unsupported features
125For each unsupported feature enabled on an imported pool, a pool property
126named
127.Sy unsupported Ns @ Ns Ar feature-name
128will indicate why the import was allowed despite the unsupported feature.
129Possible values for this property are:
130.Bl -tag -width "readonly"
131.It Sy inactive
132The feature is in the
133.Sy enabled
134state and therefore the pool's on-disk
135format is still compatible with software that does not support this feature.
136.It Sy readonly
137The feature is read-only compatible and the pool has been imported in
138read-only mode.
139.El
140.
141.Ss Feature dependencies
142Some features depend on other features being enabled in order to function.
143Enabling a feature will automatically enable any features it depends on.
144.
145.Ss Compatibility feature sets
146It is sometimes necessary for a pool to maintain compatibility with a
147specific on-disk format, by enabling and disabling particular features.
148The
149.Sy compatibility
150feature facilitates this by allowing feature sets to be read from text files.
151When set to
152.Sy off
153.Pq the default ,
154compatibility feature sets are disabled
155.Pq i.e. all features are enabled ;
156when set to
157.Sy legacy ,
158no features are enabled.
159When set to a comma-separated list of filenames
160.Po
161each filename may either be an absolute path, or relative to
162.Pa /etc/zfs/compatibility.d
163or
164.Pa /usr/share/zfs/compatibility.d
165.Pc ,
166the lists of requested features are read from those files,
167separated by whitespace and/or commas.
168Only features present in all files are enabled.
169.Pp
170Simple sanity checks are applied to the files:
171they must be between 1 B and 16 KiB in size, and must end with a newline
172character.
173.Pp
174The requested features are applied when a pool is created using
175.Nm zpool Cm create Fl o Sy compatibility Ns = Ns Ar …
176and controls which features are enabled when using
177.Nm zpool Cm upgrade .
178.Nm zpool Cm status
179will not show a warning about disabled features which are not part
180of the requested feature set.
181.Pp
182The special value
183.Sy legacy
184prevents any features from being enabled, either via
185.Nm zpool Cm upgrade
186or
187.Nm zpool Cm set Sy feature Ns @ Ns Ar feature-name Ns = Ns Sy enabled .
188This setting also prevents pools from being upgraded to newer on-disk versions.
189This is a safety measure to prevent new features from being
190accidentally enabled, breaking compatibility.
191.Pp
192By convention, compatibility files in
193.Pa /usr/share/zfs/compatibility.d
194are provided by the distribution, and include feature sets
195supported by important versions of popular distributions, and feature
196sets commonly supported at the start of each year.
197Compatibility files in
198.Pa /etc/zfs/compatibility.d ,
199if present, will take precedence over files with the same name in
200.Pa /usr/share/zfs/compatibility.d .
201.Pp
202If an unrecognized feature is found in these files, an error message will
203be shown.
204If the unrecognized feature is in a file in
205.Pa /etc/zfs/compatibility.d ,
206this is treated as an error and processing will stop.
207If the unrecognized feature is under
208.Pa /usr/share/zfs/compatibility.d ,
209this is treated as a warning and processing will continue.
210This difference is to allow distributions to include features
211which might not be recognized by the currently-installed binaries.
212.Pp
213Compatibility files may include comments:
214any text from
215.Sq #
216to the end of the line is ignored.
217.Pp
218.Sy Example :
219.Bd -literal -compact -offset 4n
220.No example# Nm cat Pa /usr/share/zfs/compatibility.d/grub2
221# Features which are supported by GRUB2
222async_destroy
223bookmarks
224embedded_data
225empty_bpobj
226enabled_txg
227extensible_dataset
228filesystem_limits
229hole_birth
230large_blocks
231livelist
232lz4_compress
233spacemap_histogram
234zpool_checkpoint
235
236.No example# Nm zpool Cm create Fl o Sy compatibility Ns = Ns Ar grub2 Ar bootpool Ar vdev
237.Ed
238.Pp
239See
240.Xr zpool-create 8
241and
242.Xr zpool-upgrade 8
243for more information on how these commands are affected by feature sets.
244.
245.de feature
246.It Sy \\$2
247.Bl -tag -compact -width "READ-ONLY COMPATIBLE"
248.It GUID
249.Sy \\$1:\\$2
250.if !"\\$4"" \{\
251.It DEPENDENCIES
252\fB\\$4\fP\c
253.if !"\\$5"" , \fB\\$5\fP\c
254.if !"\\$6"" , \fB\\$6\fP\c
255.if !"\\$7"" , \fB\\$7\fP\c
256.if !"\\$8"" , \fB\\$8\fP\c
257.if !"\\$9"" , \fB\\$9\fP\c
258.\}
259.It READ-ONLY COMPATIBLE
260\\$3
261.El
262.Pp
263..
264.
265.ds instant-never \
266.No This feature becomes Sy active No as soon as it is enabled \
267and will never return to being Sy enabled .
268.
269.ds remount-upgrade \
270.No Each filesystem will be upgraded automatically when remounted, \
271or when a new file is created under that filesystem. \
272The upgrade can also be triggered on filesystems via \
273Nm zfs Cm set Sy version Ns = Ns Sy current Ar fs . \
274No The upgrade process runs in the background and may take a while to complete \
275for filesystems containing large amounts of files .
276.
277.de checksum-spiel
278When the
279.Sy \\$1
280feature is set to
281.Sy enabled ,
282the administrator can turn on the
283.Sy \\$1
284checksum on any dataset using
285.Nm zfs Cm set Sy checksum Ns = Ns Sy \\$1 Ar dset
286.Po see Xr zfs-set 8 Pc .
287This feature becomes
288.Sy active
289once a
290.Sy checksum
291property has been set to
292.Sy \\$1 ,
293and will return to being
294.Sy enabled
295once all filesystems that have ever had their checksum set to
296.Sy \\$1
297are destroyed.
298..
299.
300.Sh FEATURES
301The following features are supported on this system:
302.Bl -tag -width Ds
303.feature org.zfsonlinux allocation_classes yes
304This feature enables support for separate allocation classes.
305.Pp
306This feature becomes
307.Sy active
308when a dedicated allocation class vdev
309.Pq dedup or special
310is created with the
311.Nm zpool Cm create No or Nm zpool Cm add No commands .
312With device removal, it can be returned to the
313.Sy enabled
314state if all the dedicated allocation class vdevs are removed.
315.
316.feature com.delphix async_destroy yes
317Destroying a file system requires traversing all of its data in order to
318return its used space to the pool.
319Without
320.Sy async_destroy ,
321the file system is not fully removed until all space has been reclaimed.
322If the destroy operation is interrupted by a reboot or power outage,
323the next attempt to open the pool will need to complete the destroy
324operation synchronously.
325.Pp
326When
327.Sy async_destroy
328is enabled, the file system's data will be reclaimed by a background process,
329allowing the destroy operation to complete
330without traversing the entire file system.
331The background process is able to resume
332interrupted destroys after the pool has been opened, eliminating the need
333to finish interrupted destroys as part of the open operation.
334The amount of space remaining to be reclaimed by the background process
335is available through the
336.Sy freeing
337property.
338.Pp
339This feature is only
340.Sy active
341while
342.Sy freeing
343is non-zero.
344.
345.feature org.openzfs blake3 no extensible_dataset
346This feature enables the use of the BLAKE3 hash algorithm for checksum and
347dedup.
348BLAKE3 is a secure hash algorithm focused on high performance.
349.Pp
350.checksum-spiel blake3
351.
352.feature com.fudosecurity block_cloning yes
353When this feature is enabled ZFS will use block cloning for operations like
354.Fn copy_file_range 2 .
355Block cloning allows to create multiple references to a single block.
356It is much faster than copying the data (as the actual data is neither read nor
357written) and takes no additional space.
358Blocks can be cloned across datasets under some conditions (like disabled
359encryption and equal
360.Nm recordsize ) .
361.Pp
362This feature becomes
363.Sy active
364when first block is cloned.
365When the last cloned block is freed, it goes back to the enabled state.
366.feature com.delphix bookmarks yes extensible_dataset
367This feature enables use of the
368.Nm zfs Cm bookmark
369command.
370.Pp
371This feature is
372.Sy active
373while any bookmarks exist in the pool.
374All bookmarks in the pool can be listed by running
375.Nm zfs Cm list Fl t Sy bookmark Fl r Ar poolname .
376.
377.feature com.datto bookmark_v2 no bookmark extensible_dataset
378This feature enables the creation and management of larger bookmarks which are
379needed for other features in ZFS.
380.Pp
381This feature becomes
382.Sy active
383when a v2 bookmark is created and will be returned to the
384.Sy enabled
385state when all v2 bookmarks are destroyed.
386.
387.feature com.delphix bookmark_written no bookmark extensible_dataset bookmark_v2
388This feature enables additional bookmark accounting fields, enabling the
389.Sy written Ns # Ns Ar bookmark
390property
391.Pq space written since a bookmark
392and estimates of send stream sizes for incrementals from bookmarks.
393.Pp
394This feature becomes
395.Sy active
396when a bookmark is created and will be
397returned to the
398.Sy enabled
399state when all bookmarks with these fields are destroyed.
400.
401.feature org.openzfs device_rebuild yes
402This feature enables the ability for the
403.Nm zpool Cm attach
404and
405.Nm zpool Cm replace
406commands to perform sequential reconstruction
407.Pq instead of healing reconstruction
408when resilvering.
409.Pp
410Sequential reconstruction resilvers a device in LBA order without immediately
411verifying the checksums.
412Once complete, a scrub is started, which then verifies the checksums.
413This approach allows full redundancy to be restored to the pool
414in the minimum amount of time.
415This two-phase approach will take longer than a healing resilver
416when the time to verify the checksums is included.
417However, unless there is additional pool damage,
418no checksum errors should be reported by the scrub.
419This feature is incompatible with raidz configurations.
420.
421This feature becomes
422.Sy active
423while a sequential resilver is in progress, and returns to
424.Sy enabled
425when the resilver completes.
426.
427.feature com.delphix device_removal no
428This feature enables the
429.Nm zpool Cm remove
430command to remove top-level vdevs,
431evacuating them to reduce the total size of the pool.
432.Pp
433This feature becomes
434.Sy active
435when the
436.Nm zpool Cm remove
437command is used
438on a top-level vdev, and will never return to being
439.Sy enabled .
440.
441.feature org.openzfs draid no
442This feature enables use of the
443.Sy draid
444vdev type.
445dRAID is a variant of RAID-Z which provides integrated distributed
446hot spares that allow faster resilvering while retaining the benefits of RAID-Z.
447Data, parity, and spare space are organized in redundancy groups
448and distributed evenly over all of the devices.
449.Pp
450This feature becomes
451.Sy active
452when creating a pool which uses the
453.Sy draid
454vdev type, or when adding a new
455.Sy draid
456vdev to an existing pool.
457.
458.feature org.illumos edonr no extensible_dataset
459This feature enables the use of the Edon-R hash algorithm for checksum,
460including for nopwrite
461.Po if compression is also enabled, an overwrite of
462a block whose checksum matches the data being written will be ignored
463.Pc .
464In an abundance of caution, Edon-R requires verification when used with
465dedup:
466.Nm zfs Cm set Sy dedup Ns = Ns Sy edonr , Ns Sy verify
467.Po see Xr zfs-set 8 Pc .
468.Pp
469Edon-R is a very high-performance hash algorithm that was part
470of the NIST SHA-3 competition.
471It provides extremely high hash performance
472.Pq over 350% faster than SHA-256 ,
473but was not selected because of its unsuitability
474as a general purpose secure hash algorithm.
475This implementation utilizes the new salted checksumming functionality
476in ZFS, which means that the checksum is pre-seeded with a secret
477256-bit random key
478.Pq stored on the pool
479before being fed the data block to be checksummed.
480Thus the produced checksums are unique to a given pool,
481preventing hash collision attacks on systems with dedup.
482.Pp
483.checksum-spiel edonr
484.
485.feature com.delphix embedded_data no
486This feature improves the performance and compression ratio of
487highly-compressible blocks.
488Blocks whose contents can compress to 112 bytes
489or smaller can take advantage of this feature.
490.Pp
491When this feature is enabled, the contents of highly-compressible blocks are
492stored in the block
493.Dq pointer
494itself
495.Po a misnomer in this case, as it contains
496the compressed data, rather than a pointer to its location on disk
497.Pc .
498Thus the space of the block
499.Pq one sector, typically 512 B or 4 KiB
500is saved, and no additional I/O is needed to read and write the data block.
501.
502\*[instant-never]
503.
504.feature com.delphix empty_bpobj yes
505This feature increases the performance of creating and using a large
506number of snapshots of a single filesystem or volume, and also reduces
507the disk space required.
508.Pp
509When there are many snapshots, each snapshot uses many Block Pointer
510Objects
511.Pq bpobjs
512to track blocks associated with that snapshot.
513However, in common use cases, most of these bpobjs are empty.
514This feature allows us to create each bpobj on-demand,
515thus eliminating the empty bpobjs.
516.Pp
517This feature is
518.Sy active
519while there are any filesystems, volumes,
520or snapshots which were created after enabling this feature.
521.
522.feature com.delphix enabled_txg yes
523Once this feature is enabled, ZFS records the transaction group number
524in which new features are enabled.
525This has no user-visible impact, but other features may depend on this feature.
526.Pp
527This feature becomes
528.Sy active
529as soon as it is enabled and will never return to being
530.Sy enabled .
531.
532.feature com.datto encryption no bookmark_v2 extensible_dataset
533This feature enables the creation and management of natively encrypted datasets.
534.Pp
535This feature becomes
536.Sy active
537when an encrypted dataset is created and will be returned to the
538.Sy enabled
539state when all datasets that use this feature are destroyed.
540.
541.feature com.delphix extensible_dataset no
542This feature allows more flexible use of internal ZFS data structures,
543and exists for other features to depend on.
544.Pp
545This feature will be
546.Sy active
547when the first dependent feature uses it, and will be returned to the
548.Sy enabled
549state when all datasets that use this feature are destroyed.
550.
551.feature com.joyent filesystem_limits yes extensible_dataset
552This feature enables filesystem and snapshot limits.
553These limits can be used to control how many filesystems and/or snapshots
554can be created at the point in the tree on which the limits are set.
555.Pp
556This feature is
557.Sy active
558once either of the limit properties has been set on a dataset
559and will never return to being
560.Sy enabled .
561.
562.feature com.delphix head_errlog no
563This feature enables the upgraded version of errlog, which required an on-disk
564error log format change.
565Now the error log of each head dataset is stored separately in the zap object
566and keyed by the head id.
567With this feature enabled, every dataset affected by an error block is listed
568in the output of
569.Nm zpool Cm status .
570In case of encrypted filesystems with unloaded keys we are unable to check
571their snapshots or clones for errors and these will not be reported.
572An "access denied" error will be reported.
573.Pp
574\*[instant-never]
575.
576.feature com.delphix hole_birth no enabled_txg
577This feature has/had bugs, the result of which is that, if you do a
578.Nm zfs Cm send Fl i
579.Pq or Fl R , No since it uses Fl i
580from an affected dataset, the receiving party will not see any checksum
581or other errors, but the resulting destination snapshot
582will not match the source.
583Its use by
584.Nm zfs Cm send Fl i
585has been disabled by default
586.Po
587see
588.Sy send_holes_without_birth_time
589in
590.Xr zfs 4
591.Pc .
592.Pp
593This feature improves performance of incremental sends
594.Pq Nm zfs Cm send Fl i
595and receives for objects with many holes.
596The most common case of hole-filled objects is zvols.
597.Pp
598An incremental send stream from snapshot
599.Sy A No to snapshot Sy B
600contains information about every block that changed between
601.Sy A No and Sy B .
602Blocks which did not change between those snapshots can be
603identified and omitted from the stream using a piece of metadata called
604the
605.Dq block birth time ,
606but birth times are not recorded for holes
607.Pq blocks filled only with zeroes .
608Since holes created after
609.Sy A No cannot be distinguished from holes created before Sy A ,
610information about every hole in the entire filesystem or zvol
611is included in the send stream.
612.Pp
613For workloads where holes are rare this is not a problem.
614However, when incrementally replicating filesystems or zvols with many holes
615.Pq for example a zvol formatted with another filesystem
616a lot of time will be spent sending and receiving unnecessary information
617about holes that already exist on the receiving side.
618.Pp
619Once the
620.Sy hole_birth
621feature has been enabled the block birth times
622of all new holes will be recorded.
623Incremental sends between snapshots created after this feature is enabled
624will use this new metadata to avoid sending information about holes that
625already exist on the receiving side.
626.Pp
627\*[instant-never]
628.
629.feature org.open-zfs large_blocks no extensible_dataset
630This feature allows the record size on a dataset to be set larger than 128 KiB.
631.Pp
632This feature becomes
633.Sy active
634once a dataset contains a file with a block size larger than 128 KiB,
635and will return to being
636.Sy enabled
637once all filesystems that have ever had their recordsize larger than 128 KiB
638are destroyed.
639.
640.feature org.zfsonlinux large_dnode no extensible_dataset
641This feature allows the size of dnodes in a dataset to be set larger than 512 B.
642.
643This feature becomes
644.Sy active
645once a dataset contains an object with a dnode larger than 512 B,
646which occurs as a result of setting the
647.Sy dnodesize
648dataset property to a value other than
649.Sy legacy .
650The feature will return to being
651.Sy enabled
652once all filesystems that have ever contained a dnode larger than 512 B
653are destroyed.
654Large dnodes allow more data to be stored in the bonus buffer,
655thus potentially improving performance by avoiding the use of spill blocks.
656.
657.feature com.delphix livelist yes
658This feature allows clones to be deleted faster than the traditional method
659when a large number of random/sparse writes have been made to the clone.
660All blocks allocated and freed after a clone is created are tracked by the
661the clone's livelist which is referenced during the deletion of the clone.
662The feature is activated when a clone is created and remains
663.Sy active
664until all clones have been destroyed.
665.
666.feature com.delphix log_spacemap yes com.delphix:spacemap_v2
667This feature improves performance for heavily-fragmented pools,
668especially when workloads are heavy in random-writes.
669It does so by logging all the metaslab changes on a single spacemap every TXG
670instead of scattering multiple writes to all the metaslab spacemaps.
671.Pp
672\*[instant-never]
673.
674.feature org.illumos lz4_compress no
675.Sy lz4
676is a high-performance real-time compression algorithm that
677features significantly faster compression and decompression as well as a
678higher compression ratio than the older
679.Sy lzjb
680compression.
681Typically,
682.Sy lz4
683compression is approximately 50% faster on compressible data and 200% faster
684on incompressible data than
685.Sy lzjb .
686It is also approximately 80% faster on decompression,
687while giving approximately a 10% better compression ratio.
688.Pp
689When the
690.Sy lz4_compress
691feature is set to
692.Sy enabled ,
693the administrator can turn on
694.Sy lz4
695compression on any dataset on the pool using the
696.Xr zfs-set 8
697command.
698All newly written metadata will be compressed with the
699.Sy lz4
700algorithm.
701.Pp
702\*[instant-never]
703.
704.feature com.joyent multi_vdev_crash_dump no
705This feature allows a dump device to be configured with a pool comprised
706of multiple vdevs.
707Those vdevs may be arranged in any mirrored or raidz configuration.
708.Pp
709When the
710.Sy multi_vdev_crash_dump
711feature is set to
712.Sy enabled ,
713the administrator can use
714.Xr dumpadm 8
715to configure a dump device on a pool comprised of multiple vdevs.
716.Pp
717Under
718.Fx
719and Linux this feature is unused, but registered for compatibility.
720New pools created on these systems will have the feature
721.Sy enabled
722but will never transition to
723.Sy active ,
724as this functionality is not required for crash dump support.
725Existing pools where this feature is
726.Sy active
727can be imported.
728.
729.feature com.delphix obsolete_counts yes device_removal
730This feature is an enhancement of
731.Sy device_removal ,
732which will over time reduce the memory used to track removed devices.
733When indirect blocks are freed or remapped,
734we note that their part of the indirect mapping is
735.Dq obsolete
736– no longer needed.
737.Pp
738This feature becomes
739.Sy active
740when the
741.Nm zpool Cm remove
742command is used on a top-level vdev, and will never return to being
743.Sy enabled .
744.
745.feature org.zfsonlinux project_quota yes extensible_dataset
746This feature allows administrators to account the spaces and objects usage
747information against the project identifier
748.Pq ID .
749.Pp
750The project ID is an object-based attribute.
751When upgrading an existing filesystem,
752objects without a project ID will be assigned a zero project ID.
753When this feature is enabled, newly created objects inherit
754their parent directories' project ID if the parent's inherit flag is set
755.Pq via Nm chattr Sy [+-]P No or Nm zfs Cm project Fl s Ns | Ns Fl C .
756Otherwise, the new object's project ID will be zero.
757An object's project ID can be changed at any time by the owner
758.Pq or privileged user
759via
760.Nm chattr Fl p Ar prjid
761or
762.Nm zfs Cm project Fl p Ar prjid .
763.Pp
764This feature will become
765.Sy active
766as soon as it is enabled and will never return to being
767.Sy disabled .
768\*[remount-upgrade]
769.
770.feature com.delphix redaction_bookmarks no bookmarks extensible_dataset
771This feature enables the use of redacted
772.Nm zfs Cm send Ns s ,
773which create redaction bookmarks storing the list of blocks
774redacted by the send that created them.
775For more information about redacted sends, see
776.Xr zfs-send 8 .
777.
778.feature com.delphix redacted_datasets no extensible_dataset
779This feature enables the receiving of redacted
780.Nm zfs Cm send
781streams, which create redacted datasets when received.
782These datasets are missing some of their blocks,
783and so cannot be safely mounted, and their contents cannot be safely read.
784For more information about redacted receives, see
785.Xr zfs-send 8 .
786.
787.feature com.datto resilver_defer yes
788This feature allows ZFS to postpone new resilvers if an existing one is already
789in progress.
790Without this feature, any new resilvers will cause the currently
791running one to be immediately restarted from the beginning.
792.Pp
793This feature becomes
794.Sy active
795once a resilver has been deferred, and returns to being
796.Sy enabled
797when the deferred resilver begins.
798.
799.feature org.illumos sha512 no extensible_dataset
800This feature enables the use of the SHA-512/256 truncated hash algorithm
801.Pq FIPS 180-4
802for checksum and dedup.
803The native 64-bit arithmetic of SHA-512 provides an approximate 50%
804performance boost over SHA-256 on 64-bit hardware
805and is thus a good minimum-change replacement candidate
806for systems where hash performance is important,
807but these systems cannot for whatever reason utilize the faster
808.Sy skein No and Sy edonr
809algorithms.
810.Pp
811.checksum-spiel sha512
812.
813.feature org.illumos skein no extensible_dataset
814This feature enables the use of the Skein hash algorithm for checksum and dedup.
815Skein is a high-performance secure hash algorithm that was a
816finalist in the NIST SHA-3 competition.
817It provides a very high security margin and high performance on 64-bit hardware
818.Pq 80% faster than SHA-256 .
819This implementation also utilizes the new salted checksumming
820functionality in ZFS, which means that the checksum is pre-seeded with a
821secret 256-bit random key
822.Pq stored on the pool
823before being fed the data block to be checksummed.
824Thus the produced checksums are unique to a given pool,
825preventing hash collision attacks on systems with dedup.
826.Pp
827.checksum-spiel skein
828.
829.feature com.delphix spacemap_histogram yes
830This features allows ZFS to maintain more information about how free space
831is organized within the pool.
832If this feature is
833.Sy enabled ,
834it will be activated when a new space map object is created, or
835an existing space map is upgraded to the new format,
836and never returns back to being
837.Sy enabled .
838.
839.feature com.delphix spacemap_v2 yes
840This feature enables the use of the new space map encoding which
841consists of two words
842.Pq instead of one
843whenever it is advantageous.
844The new encoding allows space maps to represent large regions of
845space more efficiently on-disk while also increasing their maximum
846addressable offset.
847.Pp
848This feature becomes
849.Sy active
850once it is
851.Sy enabled ,
852and never returns back to being
853.Sy enabled .
854.
855.feature org.zfsonlinux userobj_accounting yes extensible_dataset
856This feature allows administrators to account the object usage information
857by user and group.
858.Pp
859\*[instant-never]
860\*[remount-upgrade]
861.
862.feature com.klarasystems vdev_zaps_v2 no
863This feature creates a ZAP object for the root vdev.
864.Pp
865This feature becomes active after the next
866.Nm zpool Cm import
867or
868.Nm zpool reguid .
869.
870Properties can be retrieved or set on the root vdev using
871.Nm zpool Cm get
872and
873.Nm zpool Cm set
874with
875.Sy root
876as the vdev name which is an alias for
877.Sy root-0 .
878.feature org.openzfs zilsaxattr yes extensible_dataset
879This feature enables
880.Sy xattr Ns = Ns Sy sa
881extended attribute logging in the ZIL.
882If enabled, extended attribute changes
883.Pq both Sy xattrdir Ns = Ns Sy dir No and Sy xattr Ns = Ns Sy sa
884are guaranteed to be durable if either the dataset had
885.Sy sync Ns = Ns Sy always
886set at the time the changes were made, or
887.Xr sync 2
888is called on the dataset after the changes were made.
889.Pp
890This feature becomes
891.Sy active
892when a ZIL is created for at least one dataset and will be returned to the
893.Sy enabled
894state when it is destroyed for all datasets that use this feature.
895.
896.feature com.delphix zpool_checkpoint yes
897This feature enables the
898.Nm zpool Cm checkpoint
899command that can checkpoint the state of the pool
900at the time it was issued and later rewind back to it or discard it.
901.Pp
902This feature becomes
903.Sy active
904when the
905.Nm zpool Cm checkpoint
906command is used to checkpoint the pool.
907The feature will only return back to being
908.Sy enabled
909when the pool is rewound or the checkpoint has been discarded.
910.
911.feature org.freebsd zstd_compress no extensible_dataset
912.Sy zstd
913is a high-performance compression algorithm that features a
914combination of high compression ratios and high speed.
915Compared to
916.Sy gzip ,
917.Sy zstd
918offers slightly better compression at much higher speeds.
919Compared to
920.Sy lz4 ,
921.Sy zstd
922offers much better compression while being only modestly slower.
923Typically,
924.Sy zstd
925compression speed ranges from 250 to 500 MB/s per thread
926and decompression speed is over 1 GB/s per thread.
927.Pp
928When the
929.Sy zstd
930feature is set to
931.Sy enabled ,
932the administrator can turn on
933.Sy zstd
934compression of any dataset using
935.Nm zfs Cm set Sy compress Ns = Ns Sy zstd Ar dset
936.Po see Xr zfs-set 8 Pc .
937This feature becomes
938.Sy active
939once a
940.Sy compress
941property has been set to
942.Sy zstd ,
943and will return to being
944.Sy enabled
945once all filesystems that have ever had their
946.Sy compress
947property set to
948.Sy zstd
949are destroyed.
950.El
951.
952.Sh SEE ALSO
953.Xr zfs 8 ,
954.Xr zpool 8
955