1.\" Copyright (c) 2007 The DragonFly Project. All rights reserved. 2.\" 3.\" This code is derived from software contributed to The DragonFly Project 4.\" by Matthew Dillon <dillon@backplane.com> 5.\" 6.\" Redistribution and use in source and binary forms, with or without 7.\" modification, are permitted provided that the following conditions 8.\" are met: 9.\" 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 15.\" distribution. 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. 19.\" 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 31.\" SUCH DAMAGE. 32.\" 33.Dd January 2, 2011 34.Dt HAMMER 8 35.Os 36.Sh NAME 37.Nm hammer 38.Nd HAMMER file system utility 39.Sh SYNOPSIS 40.Nm 41.Fl h 42.Nm 43.Op Fl 2BqrvXy 44.Op Fl b Ar bandwidth 45.Op Fl c Ar cyclefile 46.Op Fl f Ar blkdevs 47.\" .Op Fl s Ar linkpath 48.Op Fl i Ar delay 49.Op Fl p Ar ssh-port 50.Op Fl t Ar seconds 51.Op Fl C Ar cachesize Ns Op Ns Cm \&: Ns Ar readahead 52.Op Fl S Ar splitsize 53.Ar command 54.Op Ar argument ... 55.Sh DESCRIPTION 56This manual page documents the 57.Nm 58utility which provides miscellaneous functions related to managing a 59.Nm HAMMER 60file system. 61For a general introduction to the 62.Nm HAMMER 63file system, its features, and 64examples on how to set up and maintain one, see 65.Xr HAMMER 5 . 66.Pp 67The options are as follows: 68.Bl -tag -width indent 69.It Fl h 70Get help. 71.It Fl 2 72Tell the mirror commands to use a 2-way protocol, which allows 73automatic negotiation of transaction id ranges. 74This option is automatically enabled by the 75.Cm mirror-copy 76command. 77.It Fl b Ar bandwidth 78Specify a bandwidth limit in bytes per second for mirroring streams. 79This option is typically used to prevent batch mirroring operations from 80loading down the machine. 81The bandwidth may be suffixed with 82.Cm k , m , 83or 84.Cm g 85to specify values in kilobytes, megabytes, and gigabytes per second. 86If no suffix is specified, bytes per second is assumed. 87.Pp 88Unfortunately this is only applicable to the pre-compression bandwidth 89when compression is used, so a better solution would probably be to 90use a 91.Xr ipfw 8 92pipe or a 93.Xr pf 4 94queue. 95.It Fl c Ar cyclefile 96When pruning, rebalancing or reblocking you can tell the utility 97to start at the object id stored in the specified file. 98If the file does not exist 99.Nm 100will start at the beginning. 101If 102.Nm 103is told to run for a 104specific period of time and is unable to complete the operation it will 105write out the current object id so the next run can pick up where it left off. 106If 107.Nm 108runs to completion it will delete 109.Ar cyclefile . 110.It Fl f Ar blkdevs 111Specify the volumes making up a 112.Nm HAMMER 113file system. 114.Ar Blkdevs 115is a colon-separated list of devices, each specifying a 116.Nm HAMMER 117volume. 118.It Fl i Ar delay 119When maintaining a streaming mirroring this option specifies the 120minimum delay after a batch ends before the next batch is allowed 121to start. 122The default is five seconds. 123.It Fl p Ar ssh-port 124This passes the 125.Fl p Ar ssh-port 126option to 127.Xr ssh 1 128when using a remote 129specification for the source and/or destination. 130.It Fl q 131Decrease verboseness. 132May be specified multiple times. 133.It Fl r 134Specify recursion for those commands which support it. 135.It Fl t Ar seconds 136When pruning, rebalancing or reblocking you can tell the utility to stop 137after a certain period of time. 138This option is used along with the 139.Fl c Ar cyclefile 140option to prune, rebalance or reblock incrementally. 141.It Fl v 142Increase verboseness. 143May be specified multiple times. 144.It Fl y 145Force "yes" for any interactive question. 146.It Fl B 147Bulk transfer. 148.Cm Mirror-stream 149will not attempt to break-up large initial bulk transfers into smaller 150pieces. 151This can save time but if the link is lost in the middle of the 152initial bulk transfer you will have to start over from scratch. 153This option is not recommended. 154For more information see the 155.Fl S 156option. 157.It Fl C Ar cachesize Ns Op Ns Cm \&: Ns Ar readahead 158Set the memory cache size for any raw 159.Tn I/O . 160The default is 16MB. 161A suffix of 162.Cm k 163for kilobytes and 164.Cm m 165for megabytes is allowed, 166else the cache size is specified in bytes. 167.Pp 168The read-behind/read-ahead defaults to 4 169.Nm HAMMER 170blocks. 171.Pp 172This option is typically only used with diagnostic commands 173as kernel-supported commands will use the kernel's buffer cache. 174.It Fl S Ar splitsize 175Specify the bulk splitup size in bytes for mirroring streams. 176When a 177.Cm mirror-stream 178is first started 179.Nm 180will do an initial run-through of the data to calculate good 181transaction ids to cut up the bulk transfers, creating 182restart points in case the stream is interrupted. 183If we don't do this and the stream is interrupted it might 184have to start all over again. 185The default is a splitsize of 4G. 186.Pp 187At the moment the run-through is disk-bandwidth-heavy but some 188future version will limit the run-through to just the B-Tree 189records and not the record data. 190.Pp 191The splitsize may be suffixed with 192.Cm k , m , 193or 194.Cm g 195to specify values in kilobytes, megabytes, or gigabytes. 196If no suffix is specified, bytes is assumed. 197.Pp 198When mirroring very large filesystems the minimum recommended 199split side is 4G. 200A small split size may wind up generating a great deal of overhead 201but very little actual incremental data and is not recommended. 202.It Fl X 203Enable compression for any remote ssh specifications. 204Unfortunately the 205.Fl C 206option has already been reserved for other purposes so we had to use 207a different letter. 208This option is typically used with the mirroring directives. 209.It Fl y 210Force "yes" for any interactive question. 211.El 212.Pp 213The commands are as follows: 214.Bl -tag -width indent 215.\" ==== synctid ==== 216.It Cm synctid Ar filesystem Op Cm quick 217Generates a guaranteed, formal 64 bit transaction id representing the 218current state of the specified 219.Nm HAMMER 220file system. 221The file system will be synced to the media. 222.Pp 223If the 224.Cm quick 225keyword is specified the file system will be soft-synced, meaning that a 226crash might still undo the state of the file system as of the transaction 227id returned but any new modifications will occur after the returned 228transaction id as expected. 229.Pp 230This operation does not create a snapshot. 231It is meant to be used 232to track temporary fine-grained changes to a subset of files and 233will only remain valid for 234.Ql @@ 235snapshot access purposes for the 236.Cm prune-min 237period configured for the PFS. 238If you desire a real snapshot then the 239.Cm snapq 240directive may be what you are looking for. 241.\" ==== bstats ==== 242.It Cm bstats Op Ar interval 243Output 244.Nm HAMMER 245B-Tree statistics until interrupted. 246Pause 247.Ar interval 248seconds between each display. 249The default interval is one second. 250.\" ==== iostats ==== 251.It Cm iostats Op Ar interval 252Output 253.Nm HAMMER 254.Tn I/O 255statistics until interrupted. 256Pause 257.Ar interval 258seconds between each display. 259The default interval is one second. 260.\" ==== history ==== 261.It Cm history Ar path ... 262Show the modification history for 263.Nm HAMMER 264file's inode and data. 265.\" ==== blockmap ==== 266.It Cm blockmap 267Dump the blockmap for the file system. 268The 269.Nm HAMMER 270blockmap is two-layer 271blockmap representing the maximum possible file system size of 1 Exabyte. 272Needless to say the second layer is only present for blocks which exist. 273.Nm HAMMER Ns 's 274blockmap represents 8-Megabyte blocks, called big-blocks. 275Each big-block has an append 276point, a free byte count, and a typed zone id which allows content to be 277reverse engineered to some degree. 278.Pp 279In 280.Nm HAMMER 281allocations are essentially appended to a selected big-block using 282the append offset and deducted from the free byte count. 283When space is freed the free byte count is adjusted but 284.Nm HAMMER 285does not track holes in big-blocks for reallocation. 286A big-block must be completely freed, either 287through normal file system operations or through reblocking, before 288it can be reused. 289.Pp 290Data blocks can be shared by deducting the space used from the free byte 291count for each shared references. 292This means the free byte count can legally go negative. 293.Pp 294This command needs the 295.Fl f 296flag. 297.\" ==== checkmap ==== 298.It Cm checkmap 299Check the blockmap allocation count. 300.Nm 301will scan the B-Tree, collect allocation information, and 302construct a blockmap in-memory. It will then check that blockmap 303against the on-disk blockmap. 304.Pp 305This command needs the 306.Fl f 307flag. 308.\" ==== show ==== 309.It Cm show Op Ar lo Ns Cm \&: Ns Ar objid 310Dump the B-Tree. 311By default this command will validate all B-Tree 312linkages and CRCs, including data CRCs, and will report the most verbose 313information it can dig up. 314Any errors will show up with a 315.Ql B 316in column 1 along with various 317other error flags. 318.Pp 319If you specify a localization field or a localization:obj_id field, 320.Ar lo Ns Cm \&: Ns Ar objid , 321the dump will 322search for the key printing nodes as it recurses down, and then 323will iterate forwards. 324These fields are specified in HEX. 325Note that the pfsid is the top 16 bits of the 32 bit localization 326field so PFS #1 would be 00010000. 327.Pp 328If you use 329.Fl q 330the command will report less information about the inode contents. 331.Pp 332If you use 333.Fl qq 334the command will not report the content of the inode or other typed 335data at all. 336.Pp 337If you use 338.Fl qqq 339the command will not report volume header information, big-block fill 340ratios, mirror transaction ids, or report or check data CRCs. 341B-Tree CRCs and linkages are still checked. 342.Pp 343This command needs the 344.Fl f 345flag. 346.\" ==== show-undo ==== 347.It Cm show-undo 348.Nm ( HAMMER 349VERSION 4+) 350Dump the UNDO map. 351.Pp 352This command needs the 353.Fl f 354flag. 355.\" .It Ar blockmap 356.\" Dump the B-Tree, record, large-data, and small-data blockmaps, showing 357.\" physical block assignments and free space percentages. 358.\" ==== recover ==== 359.It Cm recover Ar targetdir 360This is a low level command which operates on the filesystem image and 361attempts to locate and recover files from a corrupted filesystem. The 362entire image is scanned linearly looking for B-Tree nodes. Any node 363found which passes its crc test is scanned for file, inode, and directory 364fragments and the target directory is populated with the resulting data. 365files and directories in the target directory are initially named after 366the object id and are renamed as fragmentary information is processed. 367.Pp 368This command keeps track of filename/objid translations and may eat a 369considerably amount of memory while operating. 370.Pp 371This command is literally the last line of defense when it comes to 372recovering data from a dead filesystem. 373.\" ==== namekey1 ==== 374.It Cm namekey1 Ar filename 375Generate a 376.Nm HAMMER 37764 bit directory hash for the specified file name, using 378the original directory hash algorithm in version 1 of the file system. 379The low 32 bits are used as an iterator for hash collisions and will be 380output as 0. 381.\" ==== namekey2 ==== 382.It Cm namekey2 Ar filename 383Generate a 384.Nm HAMMER 38564 bit directory hash for the specified file name, using 386the new directory hash algorithm in version 2 of the file system. 387The low 32 bits are still used as an iterator but will start out containing 388part of the hash key. 389.\" ==== namekey32 ==== 390.It Cm namekey32 Ar filename 391Generate the top 32 bits of a 392.Nm HAMMER 39364 bit directory hash for the specified file name. 394.\" ==== info ==== 395.It Cm info 396Shows extended information about all the mounted 397.Nm HAMMER 398file systems. 399The information is divided into sections: 400.Bl -tag -width indent 401.It Volume identification 402General information, like the label of the 403.Nm HAMMER 404filesystem, the number of volumes it contains, the FSID, and the 405.Nm HAMMER 406version being used. 407.It Big block information 408Big block statistics, such as total, used, reserved and free big blocks. 409.It Space information 410Information about space used on the filesystem. 411Currently total size, used, reserved and free space are displayed. 412.It PFS information 413Basic information about the PFSs currently present on a 414.Nm HAMMER 415filesystem. 416.Pp 417.Dq PFS ID 418is the ID of the PFS, with 0 being the root PFS. 419.Dq Snaps 420is the current snapshot count on the PFS. 421.Dq Mounted on 422displays the mount point of the PFS is currently mounted on (if any). 423.El 424.\" ==== cleanup ==== 425.It Cm cleanup Op Ar filesystem ... 426This is a meta-command which executes snapshot, prune, rebalance, dedup 427and reblock commands on the specified 428.Nm HAMMER 429file systems. 430If no 431.Ar filesystem 432is specified this command will clean-up all 433.Nm HAMMER 434file systems in use, including PFS's. 435To do this it will scan all 436.Nm HAMMER 437and 438.Nm null 439mounts, extract PFS id's, and clean-up each PFS found. 440.Pp 441This command will access a snapshots 442directory and a configuration file for each 443.Ar filesystem , 444creating them if necessary. 445.Bl -tag -width indent 446.It Nm HAMMER No version 2- 447The configuration file is 448.Pa config 449in the snapshots directory which defaults to 450.Pa <pfs>/snapshots . 451.It Nm HAMMER No version 3+ 452The configuration file is saved in file system meta-data, see 453.Nm 454.Cm config . 455The snapshots directory defaults to 456.Pa /var/hammer/<pfs> 457.Pa ( /var/hammer/root 458for root mount). 459.El 460.Pp 461The format of the configuration file is: 462.Bd -literal -offset indent 463snapshots <period> <retention-time> [any] 464prune <period> <max-runtime> 465rebalance <period> <max-runtime> 466dedup <period> <max-runtime> 467reblock <period> <max-runtime> 468recopy <period> <max-runtime> 469.Ed 470.Pp 471Defaults are: 472.Bd -literal -offset indent 473snapshots 1d 60d # 0d 0d for PFS /tmp, /var/tmp, /usr/obj 474prune 1d 5m 475rebalance 1d 5m 476dedup 1d 5m 477reblock 1d 5m 478recopy 30d 10m 479.Ed 480.Pp 481Time is given with a suffix of 482.Cm d , 483.Cm h , 484.Cm m 485or 486.Cm s 487meaning day, hour, minute and second. 488.Pp 489If the 490.Cm snapshots 491directive has a period of 0 and a retention time of 0 492then snapshot generation is disabled, removal of old snapshots are 493disabled, and prunes will use 494.Cm prune-everything . 495.Pp 496If the 497.Cm snapshots 498directive has a period of 0 but a non-zero retention time 499then this command will not create any new snapshots but will remove old 500snapshots it finds based on the retention time. This form should be 501used on PFS masters where you are generating your own snapshot softlinks 502manually and on PFS slaves when all you wish to do is prune away existing 503snapshots inherited via the mirroring stream. 504.Pp 505By default only snapshots in the form 506.Ql snap- Ns Ar yyyymmdd Ns Op - Ns Ar HHMM 507are processed. 508If the 509.Cm any 510directive is specified as a third argument on the 511.Cm snapshots 512config line then any softlink of the form 513.Ql *- Ns Ar yyyymmdd Ns Op - Ns Ar HHMM 514or 515.Ql *. Ns Ar yyyymmdd Ns Op - Ns Ar HHMM 516will be processed. 517.Pp 518A prune max-runtime of 0 means unlimited. 519.Pp 520If period hasn't passed since the previous 521.Cm cleanup 522run nothing is done. 523For example a day has passed when midnight is passed (localtime). 524By default, 525.Dx 526is set up to run 527.Nm Ar cleanup 528nightly via 529.Xr periodic 8 . 530.Pp 531The default configuration file will create a daily snapshot, do a daily 532pruning, rebalancing, deduping and reblocking run and a monthly recopy run. 533Reblocking is defragmentation with a level of 95%, 534and recopy is full defragmentation. 535.Pp 536By default prune and rebalance operations are time limited to 5 minutes, 537dedup and reblock operations to a bit over 5 minutes, 538and recopy operations to a bit over 10 minutes. 539Reblocking and recopy runs are each broken down into four separate functions: 540btree, inodes, dirs and data. 541Each function is time limited to the time given in the configuration file, 542but the btree, inodes and dirs functions usually does not take very long time, 543full defragmentation is always used for these three functions. 544Also note that this directive will by default disable snapshots on 545the following PFS's: 546.Pa /tmp , 547.Pa /var/tmp 548and 549.Pa /usr/obj . 550.Pp 551The defaults may be adjusted by modifying the configuration file. 552The pruning and reblocking commands automatically maintain a cyclefile 553for incremental operation. 554If you interrupt (^C) the program the cyclefile will be updated, 555but a sub-command 556may continue to run in the background for a few seconds until the 557.Nm HAMMER 558ioctl detects the interrupt. 559The 560.Cm snapshots 561PFS option can be set to use another location for the snapshots directory. 562.Pp 563Work on this command is still in progress. 564Expected additions: 565An ability to remove snapshots dynamically as the 566file system becomes full. 567.\" ==== config ==== 568.It Cm config Op Ar filesystem Op Ar configfile 569.Nm ( HAMMER 570VERSION 3+) 571Show or change configuration for 572.Ar filesystem . 573If zero or one arguments are specified this function dumps the current 574configuration file to stdout. 575Zero arguments specifies the PFS containing the current directory. 576This configuration file is stored in file system meta-data. 577If two arguments are specified this function installs a new config file. 578.Pp 579In 580.Nm HAMMER 581versions less than 3 the configuration file is by default stored in 582.Pa <pfs>/snapshots/config , 583but in all later versions the configuration file is stored in file system 584meta-data. 585.\" ==== viconfig ==== 586.It Cm viconfig Op Ar filesystem 587.Nm ( HAMMER 588VERSION 3+) 589Edit the configuration file and reinstall into file system meta-data when done. 590Zero arguments specifies the PFS containing the current directory. 591.\" ==== volume-add ==== 592.It Cm volume-add Ar device Ar filesystem 593This command will format 594.Ar device 595and add all of its space to 596.Ar filesystem . 597.Pp 598.Em NOTE! 599All existing data contained on 600.Ar device 601will be destroyed by this operation! 602If 603.Ar device 604contains a valid 605.Nm HAMMER 606file system, formatting will be denied. 607You can overcome this sanity check 608by using 609.Xr dd 1 610to erase the beginning sectors of the device. 611Also remember that you have to specify 612.Ar device , 613together with any other device that make up the file system, 614colon-separated to 615.Pa /etc/fstab 616and 617.Xr mount_hammer 8 . 618.\" ==== volume-del ==== 619.It Cm volume-del Ar device Ar filesystem 620This command will remove volume 621.Ar device 622from 623.Ar filesystem . 624.Pp 625Remember that you have to remove 626.Ar device 627from the colon-separated list in 628.Pa /etc/fstab 629and 630.Xr mount_hammer 8 . 631.\" ==== volume-list ==== 632.It Cm volume-list Ar filesystem 633This command will list the volumes that make up 634.Ar filesystem . 635.\" ==== snapshot ==== 636.It Cm snapshot Oo Ar filesystem Oc Ar snapshot-dir 637.It Cm snapshot Ar filesystem Ar snapshot-dir Op Ar note 638Takes a snapshot of the file system either explicitly given by 639.Ar filesystem 640or implicitly derived from the 641.Ar snapshot-dir 642argument and creates a symlink in the directory provided by 643.Ar snapshot-dir 644pointing to the snapshot. 645If 646.Ar snapshot-dir 647is not a directory, it is assumed to be a format string passed to 648.Xr strftime 3 649with the current time as parameter. 650If 651.Ar snapshot-dir 652refers to an existing directory, a default format string of 653.Ql snap-%Y%d%m-%H%M 654is assumed and used as name for the newly created symlink. 655.Pp 656Snapshot is a per PFS operation, so a 657.Nm HAMMER 658file system and each PFS in it have to be snapshot separately. 659.Pp 660Example, assuming that 661.Pa /mysnapshots 662is on file system 663.Pa / 664and that 665.Pa /obj 666and 667.Pa /usr 668are file systems on their own, the following invocations: 669.Bd -literal -offset indent 670hammer snapshot /mysnapshots 671 672hammer snapshot /mysnapshots/%Y-%m-%d 673 674hammer snapshot /obj /mysnapshots/obj-%Y-%m-%d 675 676hammer snapshot /usr /my/snaps/usr "note" 677.Ed 678.Pp 679Would create symlinks similar to: 680.Bd -literal -offset indent 681/mysnapshots/snap-20080627-1210 -> /@@0x10d2cd05b7270d16 682 683/mysnapshots/2008-06-27 -> /@@0x10d2cd05b7270d16 684 685/mysnapshots/obj-2008-06-27 -> /obj@@0x10d2cd05b7270d16 686 687/my/snaps/usr/snap-20080627-1210 -> /usr@@0x10d2cd05b7270d16 688.Ed 689.Pp 690When run on a 691.Nm HAMMER 692version 3+ file system the snapshot is also recorded in file system meta-data 693along with the optional 694.Ar note . 695See the 696.Cm snapls 697directive. 698.\" ==== snap* ==== 699.It Cm snap Ar path Op Ar note 700.Nm ( HAMMER 701VERSION 3+) 702Create a snapshot for the PFS containing 703.Ar path 704and create a snapshot softlink. 705If the path specified is a 706directory a standard snapshot softlink will be created in the directory. 707The snapshot softlink points to the base of the mounted PFS. 708.It Cm snaplo Ar path Op Ar note 709.Nm ( HAMMER 710VERSION 3+) 711Create a snapshot for the PFS containing 712.Ar path 713and create a snapshot softlink. 714If the path specified is a 715directory a standard snapshot softlink will be created in the directory. 716The snapshot softlink points into the directory it is contained in. 717.It Cm snapq Ar dir Op Ar note 718.Nm ( HAMMER 719VERSION 3+) 720Create a snapshot for the PFS containing the specified directory but do 721not create a softlink. 722Instead output a path which can be used to access 723the directory via the snapshot. 724.Pp 725An absolute or relative path may be specified. 726The path will be used as-is as a prefix in the path output to stdout. 727As with the other 728snap and snapshot directives the snapshot transaction id will be registered 729in the file system meta-data. 730.It Cm snaprm Ar path Ar ... 731.It Cm snaprm Ar transid Ar ... 732.It Cm snaprm Ar filesystem Ar transid Ar ... 733.Nm ( HAMMER 734VERSION 3+) 735Remove a snapshot given its softlink or transaction id. 736If specifying a transaction id 737the snapshot is removed from file system meta-data but you are responsible 738for removing any related softlinks. 739.Pp 740If a softlink path is specified the filesystem and transaction id 741is derived from the contents of the softlink. 742If just a transaction id is specified it is assumed to be a snapshot 743in the HAMMER filesystem you are currently chdir'd into. 744You can also specify the filesystem and transaction id explicitly. 745.It Cm snapls Op Ar path ... 746.Nm ( HAMMER 747VERSION 3+) 748Dump the snapshot meta-data for PFSs containing each 749.Ar path 750listing all available snapshots and their notes. 751If no arguments are specified snapshots for the PFS containing the 752current directory are listed. 753This is the definitive list of snapshots for the file system. 754.\" ==== prune ==== 755.It Cm prune Ar softlink-dir 756Prune the file system based on previously created snapshot softlinks. 757Pruning is the act of deleting file system history. 758The 759.Cm prune 760command will delete file system history such that 761the file system state is retained for the given snapshots, 762and all history after the latest snapshot. 763By setting the per PFS parameter 764.Cm prune-min , 765history is guaranteed to be saved at least this time interval. 766All other history is deleted. 767.Pp 768The target directory is expected to contain softlinks pointing to 769snapshots of the file systems you wish to retain. 770The directory is scanned non-recursively and the mount points and 771transaction ids stored in the softlinks are extracted and sorted. 772The file system is then explicitly pruned according to what is found. 773Cleaning out portions of the file system is as simple as removing a 774snapshot softlink and then running the 775.Cm prune 776command. 777.Pp 778As a safety measure pruning only occurs if one or more softlinks are found 779containing the 780.Ql @@ 781snapshot id extension. 782Currently the scanned softlink directory must contain softlinks pointing 783to a single 784.Nm HAMMER 785mount. 786The softlinks may specify absolute or relative paths. 787Softlinks must use 20-character 788.Ql @@0x%016llx 789transaction ids, as might be returned from 790.Nm Cm synctid Ar filesystem . 791.Pp 792Pruning is a per PFS operation, so a 793.Nm HAMMER 794file system and each PFS in it have to be pruned separately. 795.Pp 796Note that pruning a file system may not immediately free-up space, 797though typically some space will be freed if a large number of records are 798pruned out. 799The file system must be reblocked to completely recover all available space. 800.Pp 801Example, lets say your that you didn't set 802.Cm prune-min , 803and snapshot directory contains the following links: 804.Bd -literal -offset indent 805lrwxr-xr-x 1 root wheel 29 May 31 17:57 snap1 -> 806/usr/obj/@@0x10d2cd05b7270d16 807 808lrwxr-xr-x 1 root wheel 29 May 31 17:58 snap2 -> 809/usr/obj/@@0x10d2cd13f3fde98f 810 811lrwxr-xr-x 1 root wheel 29 May 31 17:59 snap3 -> 812/usr/obj/@@0x10d2cd222adee364 813.Ed 814.Pp 815If you were to run the 816.Cm prune 817command on this directory, then the 818.Nm HAMMER 819.Pa /usr/obj 820mount will be pruned to retain the above three snapshots. 821In addition, history for modifications made to the file system older than 822the oldest snapshot will be destroyed and history for potentially fine-grained 823modifications made to the file system more recently than the most recent 824snapshot will be retained. 825.Pp 826If you then delete the 827.Pa snap2 828softlink and rerun the 829.Cm prune 830command, 831history for modifications pertaining to that snapshot would be destroyed. 832.Pp 833In 834.Nm HAMMER 835file system versions 3+ this command also scans the snapshots stored 836in the file system meta-data and includes them in the prune. 837.\" ==== prune-everything ==== 838.It Cm prune-everything Ar filesystem 839This command will remove all historical records from the file system. 840This directive is not normally used on a production system. 841.Pp 842This command does not remove snapshot softlinks but will delete all 843snapshots recorded in file system meta-data (for file system version 3+). 844The user is responsible for deleting any softlinks. 845.Pp 846Pruning is a per PFS operation, so a 847.Nm HAMMER 848file system and each PFS in it have to be pruned separately. 849.\" ==== rebalance ==== 850.It Cm rebalance Ar filesystem Op Ar saturation_percentage 851This command will rebalance the B-Tree, nodes with small number of 852elements will be combined and element counts will be smoothed out 853between nodes. 854.Pp 855The saturation percentage is between 50% and 100%. 856The default is 85% (the 857.Sq % 858suffix is not needed). 859.Pp 860Rebalancing is a per PFS operation, so a 861.Nm HAMMER 862file system and each PFS in it have to be rebalanced separately. 863.\" ==== dedup ==== 864.It Cm dedup Ar filesystem 865.Nm ( HAMMER 866VERSION 5+) 867Perform offline (post-process) deduplication. Deduplication occurs at 868the block level, currently only data blocks of the same size can be 869deduped, metadata blocks can not. The hash function used for comparing 870data blocks is CRC-32 (CRCs are computed anyways as part of 871.Nm HAMMER 872data integrity features, so there's no additional overhead). Since CRC 873is a weak hash function a byte-by-byte comparison is done before actual 874deduping. In case of a CRC collision (two data blocks have the same CRC 875but different contents) the checksum is upgraded to SHA-256. 876.Pp 877Currently 878.Nm HAMMER 879reblocker may partially blow up (re-expand) dedup (reblocker's normal 880operation is to reallocate every record, so it's possible for deduped 881blocks to be re-expanded back). 882.Pp 883Deduplication is a per PFS operation, so a 884.Nm HAMMER 885file system and each PFS in it have to be deduped separately. This also 886means that if you have duplicated data in two different PFSs that data 887won't be deduped, however the addition of such feature is planned. 888.\" ==== dedup-simulate ==== 889.It Cm dedup-simulate Ar filesystem 890.Nm ( HAMMER 891VERSION 5+) 892Shows potential space savings (simulated dedup ratio) one can get after 893running 894.Cm dedup 895command. If the estimated dedup ratio is greater than 1.00 you will see 896dedup space savings. Remember that this is an estimated number, in 897practice real dedup ratio will be slightly smaller because of 898.Nm HAMMER 899bigblock underflows, B-Tree locking issues and other factors. 900.Pp 901Note that deduplication currently works only on bulk data so if you 902try to run 903.Cm dedup-simulate 904or 905.Cm dedup 906commands on a PFS that contains metadata only (directory entries, 907softlinks) you will get a 0.00 dedup ratio. 908.\" ==== reblock* ==== 909.It Cm reblock Ar filesystem Op Ar fill_percentage 910.It Cm reblock-btree Ar filesystem Op Ar fill_percentage 911.It Cm reblock-inodes Ar filesystem Op Ar fill_percentage 912.It Cm reblock-dirs Ar filesystem Op Ar fill_percentage 913.It Cm reblock-data Ar filesystem Op Ar fill_percentage 914Attempt to defragment and free space for reuse by reblocking a live 915.Nm HAMMER 916file system. 917Big-blocks cannot be reused by 918.Nm HAMMER 919until they are completely free. 920This command also has the effect of reordering all elements, effectively 921defragmenting the file system. 922.Pp 923The default fill percentage is 100% and will cause the file system to be 924completely defragmented. 925All specified element types will be reallocated and rewritten. 926If you wish to quickly free up space instead try specifying 927a smaller fill percentage, such as 90% or 80% (the 928.Sq % 929suffix is not needed). 930.Pp 931Since this command may rewrite the entire contents of the disk it is 932best to do it incrementally from a 933.Xr cron 8 934job along with the 935.Fl c Ar cyclefile 936and 937.Fl t Ar seconds 938options to limit the run time. 939The file system would thus be defragmented over long period of time. 940.Pp 941It is recommended that separate invocations be used for each data type. 942B-Tree nodes, inodes, and directories are typically the most important 943elements needing defragmentation. 944Data can be defragmented over a longer period of time. 945.Pp 946Reblocking is a per PFS operation, so a 947.Nm HAMMER 948file system and each PFS in it have to be reblocked separately. 949.\" ==== pfs-status ==== 950.It Cm pfs-status Ar dirpath ... 951Retrieve the mirroring configuration parameters for the specified 952.Nm HAMMER 953file systems or pseudo-filesystems (PFS's). 954.\" ==== pfs-master ==== 955.It Cm pfs-master Ar dirpath Op Ar options 956Create a pseudo-filesystem (PFS) inside a 957.Nm HAMMER 958file system. 959Up to 65535 such file systems can be created. 960Each PFS uses an independent inode numbering space making it suitable 961for use as a replication source or target. 962.Pp 963The 964.Cm pfs-master 965directive creates a PFS that you can read, write, and use as a mirroring 966source. 967.Pp 968It is recommended to use a 969.Nm null 970mount to access a PFS, for more information see 971.Xr HAMMER 5 . 972.\" ==== pfs-slave ==== 973.It Cm pfs-slave Ar dirpath Op Ar options 974Create a pseudo-filesystem (PFS) inside a 975.Nm HAMMER 976file system. 977Up to 65535 such file systems can be created. 978Each PFS uses an independent inode numbering space making it suitable 979for use as a replication source or target. 980.Pp 981The 982.Cm pfs-slave 983directive creates a PFS that you can use as a mirroring target. 984You will not be able to access a slave PFS until you have completed the 985first mirroring operation with it as the target (its root directory will 986not exist until then). 987.Pp 988Access to the pfs-slave via the special softlink, as described in the 989.Sx PFS NOTES 990below, allows 991.Nm HAMMER 992to 993dynamically modify the snapshot transaction id by returning a dynamic result 994from 995.Xr readlink 2 996calls. 997.Pp 998A PFS can only be truly destroyed with the 999.Cm pfs-destroy 1000directive. 1001Removing the softlink will not destroy the underlying PFS. 1002.Pp 1003It is recommended to use a 1004.Nm null 1005mount to access a PFS, for more information see 1006.Xr HAMMER 5 . 1007.\" ==== pfs-update ==== 1008.It Cm pfs-update Ar dirpath Op Ar options 1009Update the configuration parameters for an existing 1010.Nm HAMMER 1011file system or pseudo-filesystem. 1012Options that may be specified: 1013.Bl -tag -width indent 1014.It Cm sync-beg-tid= Ns Ar 0x16llx 1015This is the automatic snapshot access starting transaction id for 1016mirroring slaves. 1017This parameter is normally updated automatically by the 1018.Cm mirror-write 1019directive. 1020.Pp 1021It is important to note that accessing a mirroring slave 1022with a transaction id greater than the last fully synchronized transaction 1023id can result in an unreliable snapshot since you will be accessing 1024data that is still undergoing synchronization. 1025.Pp 1026Manually modifying this field is dangerous and can result in a broken mirror. 1027.It Cm sync-end-tid= Ns Ar 0x16llx 1028This is the current synchronization point for mirroring slaves. 1029This parameter is normally updated automatically by the 1030.Cm mirror-write 1031directive. 1032.Pp 1033Manually modifying this field is dangerous and can result in a broken mirror. 1034.It Cm shared-uuid= Ns Ar uuid 1035Set the shared UUID for this file system. 1036All mirrors must have the same shared UUID. 1037For safety purposes the 1038.Cm mirror-write 1039directives will refuse to operate on a target with a different shared UUID. 1040.Pp 1041Changing the shared UUID on an existing, non-empty mirroring target, 1042including an empty but not completely pruned target, 1043can lead to corruption of the mirroring target. 1044.It Cm unique-uuid= Ns Ar uuid 1045Set the unique UUID for this file system. 1046This UUID should not be used anywhere else, 1047even on exact copies of the file system. 1048.It Cm label= Ns Ar string 1049Set a descriptive label for this file system. 1050.It Cm snapshots= Ns Ar string 1051Specify the snapshots directory which 1052.Nm 1053.Cm cleanup 1054will use to manage this PFS. 1055.Bl -tag -width indent 1056.It Nm HAMMER No version 2- 1057The snapshots directory does not need to be configured for 1058PFS masters and will default to 1059.Pa <pfs>/snapshots . 1060.Pp 1061PFS slaves are mirroring slaves so you cannot configure a snapshots 1062directory on the slave itself to be managed by the slave's machine. 1063In fact, the slave will likely have a 1064.Pa snapshots 1065sub-directory mirrored 1066from the master, but that directory contains the configuration the master 1067is using for its copy of the file system, not the configuration that we 1068want to use for our slave. 1069.Pp 1070It is recommended that 1071.Pa <fs>/var/slaves/<name> 1072be configured for a PFS slave, where 1073.Pa <fs> 1074is the base 1075.Nm HAMMER 1076file system, and 1077.Pa <name> 1078is an appropriate label. 1079.It Nm HAMMER No version 3+ 1080The snapshots directory does not need to be configured for PFS masters or 1081slaves. 1082The snapshots directory defaults to 1083.Pa /var/hammer/<pfs> 1084.Pa ( /var/hammer/root 1085for root mount). 1086.El 1087.Pp 1088You can control snapshot retention on your slave independent of the master. 1089.It Cm snapshots-clear 1090Zero out the 1091.Cm snapshots 1092directory path for this PFS. 1093.It Cm prune-min= Ns Ar N Ns Cm d 1094.It Cm prune-min= Ns Oo Ar N Ns Cm d/ Oc Ns \ 1095Ar hh Ns Op Cm \&: Ns Ar mm Ns Op Cm \&: Ns Ar ss 1096Set the minimum fine-grained data retention period. 1097.Nm HAMMER 1098always retains fine-grained history up to the most recent snapshot. 1099You can extend the retention period further by specifying a non-zero 1100pruning minimum. 1101Any snapshot softlinks within the retention period are ignored 1102for the purposes of pruning (the fine grained history is retained). 1103Number of days, hours, minutes and seconds are given as 1104.Ar N , hh , mm 1105and 1106.Ar ss . 1107.Pp 1108Because the transaction id in the snapshot softlink cannot be used 1109to calculate a timestamp, 1110.Nm HAMMER 1111uses the earlier of the 1112.Fa st_ctime 1113or 1114.Fa st_mtime 1115field of the softlink to 1116determine which snapshots fall within the retention period. 1117Users must be sure to retain one of these two fields when manipulating 1118the softlink. 1119.El 1120.\" ==== pfs-upgrade ==== 1121.It Cm pfs-upgrade Ar dirpath 1122Upgrade a PFS from slave to master operation. 1123The PFS will be rolled back to the current end synchronization transaction id 1124(removing any partial synchronizations), and will then become writable. 1125.Pp 1126.Em WARNING! 1127.Nm HAMMER 1128currently supports only single masters and using 1129this command can easily result in file system corruption 1130if you don't know what you are doing. 1131.Pp 1132This directive will refuse to run if any programs have open descriptors 1133in the PFS, including programs chdir'd into the PFS. 1134.\" ==== pfs-downgrade ==== 1135.It Cm pfs-downgrade Ar dirpath 1136Downgrade a master PFS from master to slave operation. 1137The PFS becomes read-only and access will be locked to its 1138.Cm sync-end-tid . 1139.Pp 1140This directive will refuse to run if any programs have open descriptors 1141in the PFS, including programs chdir'd into the PFS. 1142.\" ==== pfs-destroy ==== 1143.It Cm pfs-destroy Ar dirpath 1144This permanently destroys a PFS. 1145.Pp 1146This directive will refuse to run if any programs have open descriptors 1147in the PFS, including programs chdir'd into the PFS. 1148.\" ==== mirror-read ==== 1149.It Cm mirror-read Ar filesystem Op Ar begin-tid 1150Generate a mirroring stream to stdout. 1151The stream ends when the transaction id space has been exhausted. 1152.\" ==== mirror-read-stream ==== 1153.It Cm mirror-read-stream Ar filesystem Op Ar begin-tid 1154Generate a mirroring stream to stdout. 1155Upon completion the stream is paused until new data is synced to the 1156.Ar filesystem , 1157then resumed. 1158Operation continues until the pipe is broken. 1159See the 1160.Cm mirror-stream 1161command for more details. 1162.\" ==== mirror-write ==== 1163.It Cm mirror-write Ar filesystem 1164Take a mirroring stream on stdin. 1165.Pp 1166This command will fail if the 1167.Cm shared-uuid 1168configuration field for the two file systems do not match. 1169See the 1170.Cm mirror-copy 1171command for more details. 1172.Pp 1173If the target PFS does not exist this command will ask you whether 1174you want to create a compatible PFS slave for the target or not. 1175.\" ==== mirror-dump ==== 1176.It Cm mirror-dump 1177A 1178.Cm mirror-read 1179can be piped into a 1180.Cm mirror-dump 1181to dump an ASCII representation of the mirroring stream. 1182.\" ==== mirror-copy ==== 1183.\".It Cm mirror-copy Ar [[user@]host:]filesystem [[user@]host:]filesystem 1184.It Cm mirror-copy \ 1185Oo Oo Ar user Ns Cm @ Oc Ns Ar host Ns Cm \&: Oc Ns Ar filesystem \ 1186Oo Oo Ar user Ns Cm @ Oc Ns Ar host Ns Cm \&: Oc Ns Ar filesystem 1187This is a shortcut which pipes a 1188.Cm mirror-read 1189command to a 1190.Cm mirror-write 1191command. 1192If a remote host specification is made the program forks a 1193.Xr ssh 1 1194and execs the 1195.Cm mirror-read 1196and/or 1197.Cm mirror-write 1198on the appropriate host. 1199The source may be a master or slave PFS, and the target must be a slave PFS. 1200.Pp 1201This command also establishes full duplex communication and turns on 1202the 2-way protocol feature 1203.Fl ( 2 ) 1204which automatically negotiates transaction id 1205ranges without having to use a cyclefile. 1206If the operation completes successfully the target PFS's 1207.Cm sync-end-tid 1208will be updated. 1209Note that you must re-chdir into the target PFS to see the updated information. 1210If you do not you will still be in the previous snapshot. 1211.Pp 1212If the target PFS does not exist this command will ask you whether 1213you want to create a compatible PFS slave for the target or not. 1214.\" ==== mirror-stream ==== 1215.\".It Cm mirror-stream Ar [[user@]host:]filesystem [[user@]host:]filesystem 1216.It Cm mirror-stream \ 1217Oo Oo Ar user Ns Cm @ Oc Ns Ar host Ns Cm \&: Oc Ns Ar filesystem \ 1218Oo Oo Ar user Ns Cm @ Oc Ns Ar host Ns Cm \&: Oc Ns Ar filesystem 1219This is a shortcut which pipes a 1220.Cm mirror-read-stream 1221command to a 1222.Cm mirror-write 1223command. 1224This command works similarly to 1225.Cm mirror-copy 1226but does not exit after the initial mirroring completes. 1227The mirroring operation will resume as changes continue to be made to the 1228source. 1229The command is commonly used with 1230.Fl i Ar delay 1231and 1232.Fl b Ar bandwidth 1233options to keep the mirroring target in sync with the source on a continuing 1234basis. 1235.Pp 1236If the pipe is broken the command will automatically retry after sleeping 1237for a short while. 1238The time slept will be 15 seconds plus the time given in the 1239.Fl i 1240option. 1241.Pp 1242This command also detects the initial-mirroring case and spends some 1243time scanning the B-Tree to find good break points, allowing the initial 1244bulk mirroring operation to be broken down into 100MB pieces. 1245This means that the user can kill and restart the operation and it will 1246not have to start from scratch once it has gotten past the first chunk. 1247The 1248.Fl S 1249option may be used to change the size of pieces and the 1250.Fl B 1251option may be used to disable this feature and perform an initial bulk 1252transfer instead. 1253.\" ==== version ==== 1254.It Cm version Ar filesystem 1255This command returns the 1256.Nm HAMMER 1257file system version for the specified 1258.Ar filesystem 1259as well as the range of versions supported in the kernel. 1260The 1261.Fl q 1262option may be used to remove the summary at the end. 1263.\" ==== version-upgrade ==== 1264.It Cm version-upgrade Ar filesystem Ar version Op Cm force 1265This command upgrades the 1266.Nm HAMMER 1267.Ar filesystem 1268to the specified 1269.Ar version . 1270Once upgraded a file system may not be downgraded. 1271If you wish to upgrade a file system to a version greater or equal to the 1272work-in-progress version number you must specify the 1273.Cm force 1274directive. 1275Use of WIP versions should be relegated to testing and may require wiping 1276the file system as development progresses, even though the WIP version might 1277not change. 1278.Pp 1279.Em NOTE! 1280This command operates on the entire 1281.Nm HAMMER 1282file system and is not a per PFS operation. 1283All PFS's will be affected. 1284.Bl -tag -width indent 1285.It 1 1286.Dx 2.0 1287default version, first 1288.Nm HAMMER 1289release. 1290.It 2 1291.Dx 2.3 . 1292New directory entry layout. 1293This version is using a new directory hash key. 1294.It 3 1295.Dx 2.5 . 1296New snapshot management, using file system meta-data for saving 1297configuration file and snapshots (transaction ids etc.). 1298Also default snapshots directory has changed. 1299.It 4 1300.Dx 2.6 1301default version. 1302New undo/redo/flush, giving HAMMER a much faster sync and fsync. 1303.It 5 1304.Dx 2.9 . 1305Offline deduplication support. 1306.El 1307.El 1308.Sh PSEUDO-FILESYSTEM (PFS) NOTES 1309The root of a PFS is not hooked into the primary 1310.Nm HAMMER 1311file system as a directory. 1312Instead, 1313.Nm HAMMER 1314creates a special softlink called 1315.Ql @@PFS%05d 1316(exactly 10 characters long) in the primary 1317.Nm HAMMER 1318file system. 1319.Nm HAMMER 1320then modifies the contents of the softlink as read by 1321.Xr readlink 2 , 1322and thus what you see with an 1323.Nm ls 1324command or if you were to 1325.Nm cd 1326into the link. 1327If the PFS is a master the link reflects the current state of the PFS. 1328If the PFS is a slave the link reflects the last completed snapshot, and the 1329contents of the link will change when the next snapshot is completed, and 1330so forth. 1331.Pp 1332The 1333.Nm 1334utility employs numerous safeties to reduce user foot-shooting. 1335The 1336.Cm mirror-copy 1337directive requires that the target be configured as a slave and that the 1338.Cm shared-uuid 1339field of the mirroring source and target match. 1340.Sh UPGRADE INSTRUCTIONS HAMMER V1 TO V2 1341This upgrade changes the way directory entries are stored. 1342It is possible to upgrade a V1 file system to V2 in place, but 1343directories created prior to the upgrade will continue to use 1344the old layout. 1345.Pp 1346Note that the slave mirroring code in the target kernel had bugs in 1347V1 which can create an incompatible root directory on the slave. 1348Do not mix a 1349.Nm HAMMER 1350master created after the upgrade with a 1351.Nm HAMMER 1352slave created prior to the upgrade. 1353.Pp 1354Any directories created after upgrading will use a new layout. 1355.Sh UPGRADE INSTRUCTIONS HAMMER V2 TO V3 1356This upgrade adds meta-data elements to the B-Tree. 1357It is possible to upgrade a V2 file system to V3 in place. 1358After issuing the upgrade be sure to run a 1359.Nm 1360.Cm cleanup 1361to perform post-upgrade tasks. 1362.Pp 1363After making this upgrade running a 1364.Nm 1365.Cm cleanup 1366will move the 1367.Pa <pfs>/snapshots 1368directory for each PFS mount into 1369.Pa /var/hammer/<pfs> . 1370A 1371.Nm HAMMER 1372root mount will migrate 1373.Pa /snapshots 1374into 1375.Pa /var/hammer/root . 1376Migration occurs only once and only if you have not specified 1377a snapshots directory in the PFS configuration. 1378If you have specified a snapshots directory in the PFS configuration no 1379automatic migration will occur. 1380.Pp 1381For slaves, if you desire, you can migrate your snapshots 1382config to the new location manually and then clear the 1383snapshot directory configuration in the slave PFS. 1384The new snapshots hierarchy is designed to work with 1385both master and slave PFSs equally well. 1386.Pp 1387In addition, the old config file will be moved to file system meta-data, 1388editable via the new 1389.Nm 1390.Cm viconfig 1391directive. 1392The old config file will be deleted. 1393Migration occurs only once. 1394.Pp 1395The V3 file system has new 1396.Cm snap* 1397directives for creating snapshots. 1398All snapshot directives, including the original, will create 1399meta-data entries for the snapshots and the pruning code will 1400automatically incorporate these entries into its list and 1401expire them the same way it expires softlinks. 1402If you by accident blow away your snapshot softlinks you can use the 1403.Cm snapls 1404directive to get a definitive list from the file system meta-data and 1405regenerate them from that list. 1406.Pp 1407.Em WARNING! 1408If you are using 1409.Nm 1410to backup file systems your scripts may be using the 1411.Cm synctid 1412directive to generate transaction ids. 1413This directive does not create a snapshot. 1414You will have to modify your scripts to use the 1415.Cm snapq 1416directive to generate the linkbuf for the softlink you create, or 1417use one of the other 1418.Cm snap* 1419directives. 1420The older 1421.Cm snapshot 1422directive will continue to work as expected and in V3 it will also 1423record the snapshot transaction id in file system meta-data. 1424You may also want to make use of the new 1425.Ar note 1426tag for the meta-data. 1427.Pp 1428.Em WARNING! 1429If you used to remove snapshot softlinks with 1430.Nm rm 1431you should probably start using the 1432.Cm snaprm 1433directive instead to also remove the related meta-data. 1434The pruning code scans the meta-data so just removing the 1435softlink is not sufficient. 1436.Sh UPGRADE INSTRUCTIONS HAMMER V3 TO V4 1437This upgrade changes undo/flush, giving faster sync. 1438It is possible to upgrade a V3 file system to V4 in place. 1439This upgrade reformats the UNDO FIFO (typically 1GB), so upgrade might take 1440a minute or two depending. 1441.Pp 1442Version 4 allows the UNDO FIFO to be flushed without also having 1443to flush the volume header, removing 2 of the 4 disk syncs typically 1444required for an 1445.Fn fsync 1446and removing 1 of the 2 disk syncs typically 1447required for a flush sequence. 1448Version 4 also implements the REDO log (see below) which is capable 1449of fsync()ing with either one disk flush or zero disk flushes. 1450.Sh UPGRADE INSTRUCTIONS HAMMER V4 TO V5 1451This upgrade brings in offline deduplication support. It is possible to 1452upgrade a V4 file system to V5 in place. Technically it makes the 1453layer2 1454.Fa bytes_free 1455field a signed value instead of unsigned, allowing it 1456to go negative. A version 5 filesystem is required for dedup operation. 1457.Sh FSYNC FLUSH MODES 1458.Nm HAMMER 1459implements five different fsync flush modes via the 1460.Va vfs.hammer.fsync_mode 1461sysctl, for 1462.Nm HAMMER 1463version 4+ file systems. 1464.Pp 1465As of 1466.Dx 2.6 1467fsync mode 3 is set by default. 1468REDO operation and recovery is enabled by default. 1469.Bl -tag -width indent 1470.It mode 0 1471Full synchronous fsync semantics without REDO. 1472.Pp 1473.Nm HAMMER 1474will not generate REDOs. 1475A 1476.Fn fsync 1477will completely sync 1478the data and meta-data and double-flush the FIFO, including 1479issuing two disk synchronization commands. 1480The data is guaranteed 1481to be on the media as of when 1482.Fn fsync 1483returns. 1484Needless to say, this is slow. 1485.It mode 1 1486Relaxed asynchronous fsync semantics without REDO. 1487.Pp 1488This mode works the same as mode 0 except the last disk synchronization 1489command is not issued. 1490It is faster than mode 0 but not even remotely 1491close to the speed you get with mode 2 or mode 3. 1492.Pp 1493Note that there is no chance of meta-data corruption when using this 1494mode, it simply means that the data you wrote and then 1495.Fn fsync Ns 'd 1496might not have made it to the media if the storage system crashes at a bad 1497time. 1498.Pp 1499.It mode 2 1500Full synchronous fsync semantics using REDO. 1501NOTE: If not running 1502a HAMMER version 4 filesystem or later mode 0 is silently used. 1503.Pp 1504.Nm HAMMER 1505will generate REDOs in the UNDO/REDO FIFO based on a heuristic. 1506If this is sufficient to satisfy the 1507.Fn fsync 1508operation the blocks 1509will be written out and 1510.Nm HAMMER 1511will wait for the I/Os to complete, 1512and then followup with a disk sync command to guarantee the data 1513is on the media before returning. 1514This is slower than mode 3 and can result in significant disk or 1515SSDs overheads, though not as bad as mode 0 or mode 1. 1516.Pp 1517.It mode 3 1518Relaxed asynchronous fsync semantics using REDO. 1519NOTE: If not running 1520a HAMMER version 4 filesystem or later mode 1 is silently used. 1521.Pp 1522.Nm HAMMER 1523will generate REDOs in the UNDO/REDO FIFO based on a heuristic. 1524If this is sufficient to satisfy the 1525.Fn fsync 1526operation the blocks 1527will be written out and 1528.Nm HAMMER 1529will wait for the I/Os to complete, 1530but will 1531.Em NOT 1532issue a disk synchronization command. 1533.Pp 1534Note that there is no chance of meta-data corruption when using this 1535mode, it simply means that the data you wrote and then 1536.Fn fsync Ns 'd 1537might 1538not have made it to the media if the storage system crashes at a bad 1539time. 1540.Pp 1541This mode is the fastest production fsyncing mode available. 1542This mode is equivalent to how the UFS fsync in the 1543.Bx Ns s 1544operates. 1545.Pp 1546.It mode 4 1547fsync is ignored. 1548.Pp 1549Calls to 1550.Fn fsync 1551will be ignored. 1552This mode is primarily designed 1553for testing and should not be used on a production system. 1554.El 1555.Sh RESTORING FROM A SNAPSHOT BACKUP 1556You restore a snapshot by copying it over to live, but there is a caveat. 1557The mtime and atime fields for files accessed via a snapshot is locked 1558to the ctime in order to keep the snapshot consistent, because neither 1559mtime nor atime changes roll any history. 1560.Pp 1561In order to avoid unnecessary copying it is recommended that you use 1562.Nm cpdup 1563.Fl VV 1564.Fl v 1565when doing the copyback. Also make sure you traverse the snapshot softlink 1566by appending a ".", as in "<snapshotpath>/.", and you match up the directory 1567properly. 1568.Sh EXIT STATUS 1569.Ex -std 1570.Sh ENVIRONMENT 1571If the following environment variables exist, they will be used by: 1572.Bl -tag -width ".Ev EDITOR" 1573.It Ev EDITOR 1574The editor program specified in the variable 1575.Ev EDITOR 1576will be invoked instead of the default editor, which is 1577.Xr vi 1 . 1578.It Ev VISUAL 1579Same effect as 1580.Ev EDITOR 1581variable. 1582.El 1583.Sh FILES 1584.Bl -tag -width ".It Pa <fs>/var/slaves/<name>" -compact 1585.It Pa <pfs>/snapshots 1586default per PFS snapshots directory 1587.Nm ( HAMMER 1588VERSION 2-) 1589.It Pa /var/hammer/<pfs> 1590default per PFS snapshots directory (not root) 1591.Nm ( HAMMER 1592VERSION 3+) 1593.It Pa /var/hammer/root 1594default snapshots directory for root directory 1595.Nm ( HAMMER 1596VERSION 3+) 1597.It Pa <snapshots>/config 1598per PFS 1599.Nm 1600.Cm cleanup 1601configuration file 1602.Nm ( HAMMER 1603VERSION 2-) 1604.It Pa <fs>/var/slaves/<name> 1605recommended slave PFS snapshots directory 1606.Nm ( HAMMER 1607VERSION 2-) 1608.El 1609.\".Sh EXAMPLES 1610.Sh SEE ALSO 1611.Xr ssh 1 , 1612.Xr undo 1 , 1613.Xr HAMMER 5 , 1614.Xr periodic.conf 5 , 1615.Xr mount_hammer 8 , 1616.Xr mount_null 8 , 1617.Xr newfs_hammer 8 1618.Sh HISTORY 1619The 1620.Nm 1621utility first appeared in 1622.Dx 1.11 . 1623.Sh AUTHORS 1624.An Matthew Dillon Aq dillon@backplane.com 1625