1 .\" Copyright (c) 2007 The DragonFly Project. All rights reserved.
3 .\" This code is derived from software contributed to The DragonFly Project
4 .\" by Matthew Dillon <dillon@backplane.com>
6 .\" Redistribution and use in source and binary forms, with or without
7 .\" modification, are permitted provided that the following conditions
10 .\" 1. Redistributions of source code must retain the above copyright
11 .\" notice, this list of conditions and the following disclaimer.
12 .\" 2. Redistributions in binary form must reproduce the above copyright
13 .\" notice, this list of conditions and the following disclaimer in
14 .\" the documentation and/or other materials provided with the
16 .\" 3. Neither the name of The DragonFly Project nor the names of its
17 .\" contributors may be used to endorse or promote products derived
18 .\" from this software without specific, prior written permission.
20 .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
21 .\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
22 .\" LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
23 .\" FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
24 .\" COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
25 .\" INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
26 .\" BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
27 .\" LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
28 .\" AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
29 .\" OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
30 .\" OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 .\" $DragonFly: src/sbin/hammer/hammer.8,v 1.58 2008/11/13 02:04:27 dillon Exp $
40 .Nd HAMMER file system utility
46 .Op Fl f Ar blkdev[:blkdev]*
47 .\" .Op Fl s Ar linkpath
50 .Op Fl C Ar cachesize[:readahead]
54 This manual page documents the
56 utility which provides miscellaneous functions related to managing a
59 For a general introduction to the
61 file system, its features, and
62 examples on how to set up and maintain one, see
65 The options are as follows:
66 .Bl -tag -width indent
70 Tell the mirror commands to use a 2-way protocol, which allows
71 automatic negotiation of transaction id ranges.
72 This option is automatically enabled by the
76 Specify a bandwidth limit in bytes per second for mirroring streams.
77 This option is typically used to prevent batch mirroring operations from
78 loading down the machine.
79 The bandwidth may be suffixed with
84 to specify values in kilobytes, megabytes, and gigabytes per second.
85 If no suffix is specified, bytes per second is assumed.
87 When pruning and reblocking you can instruction
89 to start at the object id stored in the specified file.
90 If the file does not exist
92 will start at the beginning.
96 specific period of time and is unable to complete the operation it will
97 write out the current object id so the next run can pick up where it left off.
100 runs to completion it will delete
102 .It Fl f Ar blkdev[:blkdev]*
103 Specify the volumes making up a
107 When maintaining a streaming mirroring this option specifies the
108 minimum delay after a batch ends before the next batch is allowed
110 The default is five seconds.
112 Decrease verboseness.
113 May be specified multiple times.
115 Specify recursion for those commands which support it.
117 When pruning and reblocking you can tell the utility to stop after a
118 certain period of time.
119 This option is used along with the
121 option to prune or reblock a portion of the file system incrementally.
123 Increase verboseness.
124 May be specified multiple times.
126 Force "yes" for any interactive question.
127 .It Fl C Ar cachesize[:readahead]
128 Set the memory cache size for any raw I/O. The default is 16m.
129 A suffix of 'k' for kilobytes and 'm' for megabytes is allowed,
130 else the cache size is specified in bytes.
132 The read-behind/read-ahead defaults to 4 hammer blocks.
134 This option is typically only used with diagnostic commands
135 as kernel-supported commands will use the kernel's buffer cache.
138 The commands are as follows:
139 .Bl -tag -width indent
140 .\" ==== synctid ====
141 .It Ar synctid Ar filesystem Op quick
142 Generates a guaranteed, formal 64 bit transaction id representing the
143 current state of the specified
146 The file system will be synced to the media.
150 keyword is specified the file system will be soft-synced, meaning that a
151 crash might still undo the state of the file system as of the transaction
152 id returned but any new modifications will occur after the returned
153 transaction id as expected.
155 .It Ar bstats Op interval
158 B-tree statistics until interrupted.
161 seconds between each display.
162 The default interval is one second.
163 .\" ==== iostats ====
164 .It Ar iostats Op interval
167 I/O statistics until interrupted.
170 seconds between each display.
171 The default interval is one second.
172 .\" ==== history ====
173 .It Ar history Ar path ...
174 Show the modification history for
176 file's inode and data.
177 .\" ==== blockmap ====
179 Dump the blockmap for the filesystem. The HAMMER blockmap is two-layer
180 blockmap representing the maximum possible filesystem size of 1 Exabyte.
181 Needless to say the second layer is only present for blocks which exist.
182 HAMMER's blockmap represents 8-Megabyte blocks. Each block has an append
183 point, a free byte count, and a typed zone id which allows content to be
184 reverse engineered to some degree.
186 In HAMMER allocations essentially appended to a selected big-block using
187 the append offset and deducted from the free byte count. When space is
188 freed the free byte count is adjusted but HAMMER does not track holes in
189 big-blocks for reallocation. A big-block must be completely freed, either
190 through normal filesystem operations or through reblocking, before
193 Data blocks can be shared by deducting the space used from the free byte
194 count for each shared references, though HAMMER does not yet make use of
195 this feature. This means the free byte count can legally go negative.
197 This command needs the
202 Dump the B-tree. By default this command will validate all B-Tree
203 linkages and CRCs, including data CRCs, and will report the most verbose
204 information it can dig up.
205 Any errors will show up with a 'B' in column 1 along with various
210 the command will report less information about the inode contents.
214 the command will not report the content of the inode or other typed
219 the command will not report volume header information, big-block fill
220 ratios, mirror TIDs, or report or check data CRCs. B-Tree CRCs and
221 linkages are still checked.
223 This command needs the
227 .\" Dump the B-tree, record, large-data, and small-data blockmaps, showing
228 .\" physical block assignments and free space percentages.
229 .\" ==== namekey1 ====
230 .It Ar namekey1 Ar filename
233 64 bit directory hash for the specified file name, using
234 the original directory hash algorithm in version 1 of the filesystem.
235 The low 32 bits are used as an iterator for hash collisions and will be
237 .\" ==== namekey2 ====
238 .It Ar namekey2 Ar filename
241 64 bit directory hash for the specified file name, using
242 the new directory hash algorithm in version 2 of the filesystem.
243 The low 32 bits are still used as an iterator but will start out containing
244 part of the hash key.
245 .\" ==== namekey32 ====
246 .It Ar namekey32 Ar filename
247 Generate the top 32 bits of a
249 64 bit directory hash for the specified file name.
252 Shows extended information about all the mounted HAMMER filesystems.
253 At the moment volume identification, big blocks information and space details are shown.
254 .\" ==== cleanup ====
255 .It Ar cleanup Op Ar filesystem ...
256 This is a meta-command which executes snapshot, pruning, and reblocking
257 commands on the specified
262 is specified this command will clean-up all
264 file systems in use, including PFS's.
265 To do this it will scan all
269 mounts, extract PFS id's, and clean-up each PFS found.
271 This command will by default access a
277 creating them if necessary.
278 The format of the configuration file is:
279 .Bd -literal -offset indent
280 snapshots <period> <retention-time> [any]
281 prune <period> <max-runtime>
282 .\"rebalance <period> <max-runtime>
283 reblock <period> <1/3 max-runtime>
284 recopy <period> <1/3 max-runtime>
287 snapshots 1d 60d # 0d 60d for PFS /tmp, /var/tmp, /usr/obj
294 Time is given with a suffix of
300 meaning day, hour, minute and second.
302 If the snapshots directive has a period of 0 and a retention time of 0
303 then snapshot generation is disabled, removal of old snapshots are
304 disabled, and prunes will use
305 .Ar prune-everything .
306 If the snapshots directive has a period of 0 but a non-zero retention time
307 then this command will not create any new snapshots but will remove old
308 snapshots it finds based on the retention time.
310 By default only snapshots in the form: snap-yyyymmdd[-hhmm] are processed.
313 directive is specified as a third argument on the snapshots config line
314 then any softlink of the form *[- or .]yyyymmdd[-hhmm] will be processed.
316 A prune max-runtime of 0 means unlimited.
318 If period hasn't passed since the previous
321 For example a day has passed when midnight is passed (localtime).
329 The default configuration file will create a daily snapshot, do a daily
330 pruning and reblocking run and a monthly recopy run.
331 Reblocking is defragmentation with a level of 95%,
332 and recopy is full defragmentation.
334 By default prune and reblock operations are limited to 5 minutes per function,
335 and recopy operations are limited to 10 minutes per function.
336 Reblocking and recopy runs are each broken down into three separate functions
337 (btree, inodes and data)
338 and are thus by default limited to 15 and 30 minutes respectively.
339 Also note that this directive will by default disable snapshots on
346 The defaults may be adjusted by modifying the
349 The pruning and reblocking commands automatically maintain a cyclefile
350 for incremental operation.
351 If you interrupt (^C) the program the cyclefile will be updated, but a sub-command
352 may continue to run in the background for a few seconds until the
354 ioctl detects the interrupt.
357 PFS option can be set to use another location for the snapshots directory.
359 Work on this command is still in progress.
360 Expected additions: An ability to remove snapshots dynamically as the
361 file system becomes full.
362 .\" ==== snapshot ====
363 .It Ar snapshot Oo Ar filesystem Oc Ar snapshot-dir
364 Takes a snapshot of the file system either explicitly given by
366 or implicitly derived from the
368 argument and creates a symlink in the directory provided by
370 pointing to the snapshot.
373 is not a directory, it is assumed to be a format string passed to
375 with the current time as parameter.
378 refers to an existing directory, a default format string of "snap-%Y%d%m-%H%M"
379 is assumed and used as name for the newly created symlink.
381 Snapshot is a per PFS operation, so a
383 file system and each PFS in it have to be snapshot separately.
385 Example, assuming that
391 is a file system on its own, the following invocations:
392 .Bd -literal -offset indent
393 hammer snapshot /mysnapshots
395 hammer snapshot /mysnapshots/%Y-%m-%d
397 hammer snapshot /obj /mysnapshots/obj-%Y-%m-%d
400 would create symlinks similar to:
401 .Bd -literal -offset indent
402 /mysnapshots/snap-20080627-1210 -> /@@0x10d2cd05b7270d16
404 /mysnapshots/2008-06-27 -> /@@0x10d2cd05b7270d16
406 /mysnapshots/obj-2008-06-27 -> /obj@@0x10d2cd05b7270d16
409 .It Ar prune Ar softlink-dir
410 Prune the file system based on previously created snapshot softlinks.
411 Pruning is the act of deleting file system history.
415 will delete file system history such that
416 the file system state is retained for the given snapshots,
417 and all history after the latest snapshot,
418 but all other history is deleted.
420 The target directory is expected to contain softlinks pointing to
421 snapshots of the file systems you wish to retain.
422 The directory is scanned non-recursively and the mount points and
423 transaction ids stored in the softlinks are extracted and sorted.
424 The file system is then explicitly pruned according to what is found.
425 Cleaning out portions of the file system is as simple as removing a softlink
430 As a safety measure pruning only occurs if one or more softlinks are found
431 containing the @@ snapshot id extension.
432 Currently the scanned softlink directory must contain softlinks pointing
436 The softlinks may specify absolute or relative paths.
437 Softlinks must use 20-character (@@0x%016llx) transaction ids,
438 as might be returned from
439 .Dq Nm Ar synctid filesystem .
441 Pruning is a per PFS operation, so a
443 file system and each PFS in it have to be pruned separately.
445 Note that pruning a file system may not immediately free-up space,
446 though typically some space will be freed if a large number of records are
448 The file system must be reblocked to completely recover all available space.
450 Example, lets say your snapshot directory contains the following links:
451 .Bd -literal -offset indent
452 lrwxr-xr-x 1 root wheel 29 May 31 17:57 snap1 ->
453 /usr/obj/@@0x10d2cd05b7270d16
455 lrwxr-xr-x 1 root wheel 29 May 31 17:58 snap2 ->
456 /usr/obj/@@0x10d2cd13f3fde98f
458 lrwxr-xr-x 1 root wheel 29 May 31 17:59 snap3 ->
459 /usr/obj/@@0x10d2cd222adee364
462 If you were to run the
464 command on this directory, then the
467 mount will be pruned to retain the above three snapshots.
468 In addition, history for modifications made to the file system older than
469 the oldest snapshot will be destroyed and history for potentially fine-grained
470 modifications made to the file system more recently than the most recent
471 snapshot will be retained.
473 If you then delete the
475 softlink and rerun the
478 history for modifications pertaining to that snapshot would be destroyed.
479 .\" ==== prune-everything ====
480 .It Ar prune-everything Ar filesystem
481 This command will remove all historical records from the file system.
482 This directive is not normally used on a production system.
483 .\" ==== rebalance ====
484 .It Ar rebalance Ar filesystem Op Ar saturation_level
485 This command will rebalance the B-Tree, nodes with small numbers of
486 elements will be combined and element counts will be smoothed out
489 The saturation level is a percentage between 50 and 100. The default
491 .\" ==== reblock ====
492 .It Ar reblock Ar filesystem Op Ar fill_percentage
493 .It Ar reblock-btree Ar filesystem Op Ar fill_percentage
494 .It Ar reblock-inodes Ar filesystem Op Ar fill_percentage
495 .It Ar reblock-dirs Ar filesystem Op Ar fill_percentage
496 .It Ar reblock-data Ar filesystem Op Ar fill_percentage
497 Attempt to defragment and free space for reuse by reblocking a live
500 Big blocks cannot be reused by
502 until they are completely free.
503 This command also has the effect of reordering all elements, effectively
504 defragmenting the file system.
506 The default fill percentage is 100% and will cause the file system to be
507 completely defragmented.
508 All specified element types will be reallocated and rewritten.
509 If you wish to quickly free up space instead try specifying
510 a smaller fill percentage, such as 90% or 80% (the
512 suffix is not needed).
514 Since this command may rewrite the entire contents of the disk it is
515 best to do it incrementally from a
521 options to limit the run time.
522 The file system would thus be defragmented over long period of time.
524 It is recommended that separate invocations be used for each data type.
525 B-tree nodes, inodes, and directories are typically the most important
526 elements needing defragmentation.
527 Data can be defragmented over a longer period of time.
529 Reblocking is a per PFS operation, so a
531 file system and each PFS in it have to be reblocked separately.
532 .\" ==== pfs-status ====
533 .It Ar pfs-status Ar dirpath ...
534 Retrieve the mirroring configuration parameters for the specified
536 file systems or pseudo-filesystems (PFS's).
537 .\" ==== pfs-master ====
538 .It Ar pfs-master Ar dirpath Op options
539 Create a pseudo-filesystem (PFS) inside a
542 Up to 65535 such file systems can be created.
543 Each PFS uses an independent inode numbering space making it suitable
544 for use as a replication source or target.
548 directive creates a PFS that you can read, write, and use as a mirroring
551 It is recommended to use a
553 mount to access a PFS, for more information see
555 .\" ==== pfs-slave ====
556 .It Ar pfs-slave Ar dirpath Op options
557 Create a pseudo-filesystem (PFS) inside a
560 Up to 65535 such file systems can be created.
561 Each PFS uses an independent inode numbering space making it suitable
562 for use as a replication source or target.
566 directive creates a PFS that you can use as a mirroring target.
567 You will not be able to access a slave PFS until you have completed the
568 first mirroring operation with it as the target (its root directory will
569 not exist until then).
571 Access to the pfs-slave via the special softlink,
577 dynamically modify the snapshot transaction id by returning a dynamic result
582 A PFS can only be truly destroyed with the
585 Removing the softlink will not destroy the underlying PFS.
587 It is recommended to use a
589 mount to access a PFS, for more information see
591 .\" ==== pfs-update ====
592 .It Ar pfs-update Ar dirpath Op options
593 Update the configuration parameters for an existing
596 or pseudo-filesystem.
597 Options that may be specified:
598 .Bl -tag -width indent
599 .It sync-beg-tid=0x16llx
600 This is the automatic snapshot access starting transaction id for
602 This parameter is normally updated automatically by the
606 It is important to note that accessing a mirroring slave
607 with a transaction id greater than the last fully synchronized transaction
608 id can result in an unreliable snapshot since you will be accessing
609 data that is still undergoing synchronization.
611 Manually modifying this field is dangerous and can result in a broken
613 .It sync-end-tid=0x16llx
614 This is the current synchronization point for mirroring slaves.
615 This parameter is normally updated automatically by the
619 Manually modifying this field is dangerous and can result in a broken mirror.
620 .It shared-uuid=<uuid>
621 Set the shared UUID for this file system.
622 All mirrors must have the same shared UUID.
623 For safety purposes the
625 directives will refuse to operate on a target with a different shared UUID.
627 Changing the shared UUID on an existing, non-empty mirroring target,
628 including an empty but not completely pruned target,
629 can lead to corruption of the mirroring target.
630 .It unique-uuid=<uuid>
631 Set the unique UUID for this file system.
632 This UUID should not be used anywhere else,
633 even on exact copies of the file system.
635 Set a descriptive label for this file system.
636 .It snapshots=<string>
637 Specify the snapshots directory which
640 will use to manage this PFS.
641 The snapshots directory does not need to be configured for
642 PFS masters and will default to
643 .Pa <pfs>/snapshots .
645 PFS slaves are mirroring slaves so you cannot configure a snapshots
646 directory on the slave itself to be managed by the slave's machine.
647 In fact, the slave will likely have a
649 sub-directory mirrored
650 from the master, but that directory contains the configuration the master
651 is using for its copy of the file system, not the configuration that we
652 want to use for our slave.
654 It is recommended that
655 .Pa <fs>/var/slaves/<name>
656 be configured for a PFS slave, where
662 is an appropriate label.
663 You can control snapshot retention on your slave independent of the master.
665 Zero out the snapshots directory path for this PFS.
667 .\" ==== pfs-upgrade ====
668 .It Ar pfs-upgrade Ar dirpath
669 Upgrade a PFS from slave to master operation.
670 The PFS will be rolled back to the current end synchronization tid
671 (removing any partial synchronizations), and will then become writable.
675 currently supports only single masters and using
676 this command can easily result in file system corruption
677 if you don't know what you are doing.
679 This directive will refuse to run if any programs have open descriptors
680 in the PFS, including programs chdir'd into the PFS.
681 .\" ==== pfs-downgrade ====
682 .It Ar pfs-downgrade Ar dirpath
683 Downgrade a master PFS from master to slave operation
684 The PFS becomes read-only and access will be locked to its
687 This directive will refuse to run if any programs have open descriptors
688 in the PFS, including programs chdir'd into the PFS.
689 .\" ==== pfs-destroy ====
690 .It Ar pfs-destroy Ar dirpath
691 This permanently destroys a PFS.
693 This directive will refuse to run if any programs have open descriptors
694 in the PFS, including programs chdir'd into the PFS.
695 .\" ==== mirror-read ====
696 .It Ar mirror-read Ar filesystem Op Ar <begin-tid>
697 Generate a mirroring stream to stdout.
698 The stream ends when the transaction id space has been exhausted.
699 .\" ==== mirror-read-stream ====
700 .It Ar mirror-read-stream Ar filesystem Op Ar <begin-tid>
701 Generate a mirroring stream to stdout.
702 Upon completion the stream is paused until new data is synced to the
703 master, then resumed.
704 Operation continues until the pipe is broken.
705 .\" ==== mirror-write ====
706 .It Ar mirror-write Ar filesystem
707 Take a mirroring stream on stdin.
709 This command will fail if the
711 configuration field for the two file systems do not match.
713 If the target PFS does not exist this command will ask you whether
714 you want to create a compatible PFS slave for the target or not.
715 .\" ==== mirror-dump ====
721 to dump an ASCII representation of the mirroring stream.
722 .\" ==== mirror-copy ====
723 .It Ar mirror-copy Ar [[user@]host:]filesystem Ar [[user@]host:]filesystem
724 This is a shortcut which pipes a
729 If a remote host specification is made the program forks a
735 on the appropriate host.
736 The source may be a master or slave PFS, and the target must be a slave PFS.
738 This command also established full duplex communication and turns on
739 the two-way protocol feature which automatically negotiates transaction id
740 ranges without having to use a cyclefile.
741 If the operation completes successfully the target PFS's
744 Note that you must re-chdir into the target PFS to see the updated information.
745 If you do not you will still be in the previous snapshot.
747 If the target PFS does not exist this command will ask you whether
748 you want to create a compatible PFS slave for the target or not.
749 .\" ==== mirror-stream ====
750 .It Ar mirror-stream Ar [[user@]host:]filesystem Ar [[user@]host:]filesystem
751 This command works similarly to
753 but does not exit unless the pipe is broken.
754 This command will resume the mirroring operation whenever the master is synced.
755 The command is commonly used with
759 options to keep the mirroring target in sync with the source on a continuing
761 .\" ==== version ====
762 .It Ar version Ar filesystem
763 This command returns the
765 filesystem version for the specified
766 filesystem as well as the range of versions supported in the kernel.
769 option may be used to remove the summary at the end.
770 .\" ==== version-upgrade ====
771 .It Ar version-upgrade Ar filesystem Ar version Op Ar force
772 This command upgrades the
774 filesystem to the specified version.
775 Once upgraded a filesystem may not be downgraded.
776 If you wish to upgrade a filesystem to a version greater or equal to the
777 work-in-progress version number you must specify the
780 Use of WIP versions should be relegated to testing and may require wiping
781 the filesystem as development progresses, even though the WIP version might
784 NOTE! This command operates on the entire
786 filesystem and is not a per-PFS operation.
787 All PFS's will be affected.
788 .Bl -tag -width indent
791 default version, first
795 Work-in-progress version.
796 This version is developing a new directory hash key.
800 .Sh PSEUDO FILESYSTEM (PFS) NOTES
801 The root of a PFS is not hooked into the primary
803 file system as a directory.
806 creates a special softlink called "@@PFS%05d" (exactly 10 characters long)
811 then modifies the contents of the softlink as read by
813 and thus what you see with an
815 command or if you were to
818 If the PFS is a master the link reflects the current state of the PFS.
819 If the PFS is a slave the link reflects the last completed snapshot, and the
820 contents of the link will change when the next snapshot is completed, and
823 PFS support is currently very new and experimental.
826 utility employs numerous safeties to reduce user foot-shooting.
829 directive requires that the target be configured as a slave and that the
831 field of the mirroring source and target match.
833 .Bl -tag -width ".It Pa <fs>/var/slaves/<name>" -compact
835 default per PFS snapshots directory
836 .It Pa <snapshots>/config
840 .It Pa <fs>/var/slaves/<name>
841 recommended slave PFS snapshots directory
848 .Xr periodic.conf 5 ,
855 utility first appeared in
858 .An Matthew Dillon Aq dillon@backplane.com