| blob | d60e18582ca2f18f67f1d1b2e8d5eae776302d18 |
1 .\" $NetBSD: raidctl.8,v 1.55 2008/08/26 21:08:08 oster Exp $
2 .\"
3 .\" Copyright (c) 1998, 2002 The NetBSD Foundation, Inc.
4 .\" All rights reserved.
5 .\"
6 .\" This code is derived from software contributed to The NetBSD Foundation
7 .\" by Greg Oster
8 .\"
9 .\" Redistribution and use in source and binary forms, with or without
10 .\" modification, are permitted provided that the following conditions
11 .\" are met:
12 .\" 1. Redistributions of source code must retain the above copyright
13 .\" notice, this list of conditions and the following disclaimer.
14 .\" 2. Redistributions in binary form must reproduce the above copyright
15 .\" notice, this list of conditions and the following disclaimer in the
16 .\" documentation and/or other materials provided with the distribution.
17 .\"
18 .\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
19 .\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
20 .\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
21 .\" PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
22 .\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23 .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24 .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25 .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26 .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27 .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28 .\" POSSIBILITY OF SUCH DAMAGE.
29 .\"
30 .\"
31 .\" Copyright (c) 1995 Carnegie-Mellon University.
32 .\" All rights reserved.
33 .\"
34 .\" Author: Mark Holland
35 .\"
36 .\" Permission to use, copy, modify and distribute this software and
37 .\" its documentation is hereby granted, provided that both the copyright
38 .\" notice and this permission notice appear in all copies of the
39 .\" software, derivative works or modified versions, and any portions
40 .\" thereof, and that both notices appear in supporting documentation.
41 .\"
42 .\" CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
43 .\" CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
44 .\" FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
45 .\"
46 .\" Carnegie Mellon requests users of this software to return to
47 .\"
48 .\" Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
49 .\" School of Computer Science
50 .\" Carnegie Mellon University
51 .\" Pittsburgh PA 15213-3890
52 .\"
53 .\" any improvements or extensions that they make and grant Carnegie the
54 .\" rights to redistribute these changes.
55 .\"
56 .Dd August 26, 2008
57 .Dt RAIDCTL 8
58 .Os
59 .Sh NAME
60 .Nm raidctl
61 .Nd configuration utility for the RAIDframe disk driver
62 .Sh SYNOPSIS
63 .Nm
64 .Op Fl v
65 .Fl a Ar component Ar dev
66 .Nm
67 .Op Fl v
68 .Fl A Op yes | no | root
69 .Ar dev
70 .Nm
71 .Op Fl v
72 .Fl B Ar dev
73 .Nm
74 .Op Fl v
75 .Fl c Ar config_file Ar dev
76 .Nm
77 .Op Fl v
78 .Fl C Ar config_file Ar dev
79 .Nm
80 .Op Fl v
81 .Fl f Ar component Ar dev
82 .Nm
83 .Op Fl v
84 .Fl F Ar component Ar dev
85 .Nm
86 .Op Fl v
87 .Fl g Ar component Ar dev
88 .Nm
89 .Op Fl v
90 .Fl G Ar dev
91 .Nm
92 .Op Fl v
93 .Fl i Ar dev
94 .Nm
95 .Op Fl v
96 .Fl I Ar serial_number Ar dev
97 .Nm
98 .Op Fl v
99 .Fl p Ar dev
100 .Nm
101 .Op Fl v
102 .Fl P Ar dev
103 .Nm
104 .Op Fl v
105 .Fl r Ar component Ar dev
106 .Nm
107 .Op Fl v
108 .Fl R Ar component Ar dev
109 .Nm
110 .Op Fl v
111 .Fl s Ar dev
112 .Nm
113 .Op Fl v
114 .Fl S Ar dev
115 .Nm
116 .Op Fl v
117 .Fl u Ar dev
118 .Sh DESCRIPTION
119 .Nm
120 is the user-land control program for
121 .Xr raid 4 ,
122 the RAIDframe disk device.
123 .Nm
124 is primarily used to dynamically configure and unconfigure RAIDframe disk
125 devices.
126 For more information about the RAIDframe disk device, see
127 .Xr raid 4 .
128 .Pp
129 This document assumes the reader has at least rudimentary knowledge of
130 RAID and RAID concepts.
131 .Pp
132 The command-line options for
133 .Nm
134 are as follows:
135 .Bl -tag -width indent
136 .It Fl a Ar component Ar dev
137 Add
138 .Ar component
139 as a hot spare for the device
140 .Ar dev .
141 Component labels (which identify the location of a given
142 component within a particular RAID set) are automatically added to the
143 hot spare after it has been used and are not required for
144 .Ar component
145 before it is used.
146 .It Fl A Ic yes Ar dev
147 Make the RAID set auto-configurable.
148 The RAID set will be automatically configured at boot
149 .Ar before
150 the root file system is mounted.
151 Note that all components of the set must be of type
152 .Dv RAID
153 in the disklabel.
154 .It Fl A Ic no Ar dev
155 Turn off auto-configuration for the RAID set.
156 .It Fl A Ic root Ar dev
157 Make the RAID set auto-configurable, and also mark the set as being
158 eligible to be the root partition.
159 A RAID set configured this way will
160 .Ar override
161 the use of the boot disk as the root device.
162 All components of the set must be of type
163 .Dv RAID
164 in the disklabel.
165 Note that only certain architectures
166 .Pq currently alpha, i386, pmax, sparc, sparc64, and vax
167 support booting a kernel directly from a RAID set.
168 .It Fl B Ar dev
169 Initiate a copyback of reconstructed data from a spare disk to
170 its original disk.
171 This is performed after a component has failed,
172 and the failed drive has been reconstructed onto a spare drive.
173 .It Fl c Ar config_file Ar dev
174 Configure the RAIDframe device
175 .Ar dev
176 according to the configuration given in
177 .Ar config_file .
178 A description of the contents of
179 .Ar config_file
180 is given later.
181 .It Fl C Ar config_file Ar dev
182 As for
183 .Fl c ,
184 but forces the configuration to take place.
185 This is required the first time a RAID set is configured.
186 .It Fl f Ar component Ar dev
187 This marks the specified
188 .Ar component
189 as having failed, but does not initiate a reconstruction of that component.
190 .It Fl F Ar component Ar dev
191 Fails the specified
192 .Ar component
193 of the device, and immediately begin a reconstruction of the failed
194 disk onto an available hot spare.
195 This is one of the mechanisms used to start
196 the reconstruction process if a component does have a hardware failure.
197 .It Fl g Ar component Ar dev
198 Get the component label for the specified component.
199 .It Fl G Ar dev
200 Generate the configuration of the RAIDframe device in a format suitable for
201 use with the
202 .Fl c
203 or
204 .Fl C
205 options.
206 .It Fl i Ar dev
207 Initialize the RAID device.
208 In particular, (re-)write the parity on the selected device.
209 This
210 .Em MUST
211 be done for
212 .Em all
213 RAID sets before the RAID device is labeled and before
214 file systems are created on the RAID device.
215 .It Fl I Ar serial_number Ar dev
216 Initialize the component labels on each component of the device.
217 .Ar serial_number
218 is used as one of the keys in determining whether a
219 particular set of components belong to the same RAID set.
220 While not strictly enforced, different serial numbers should be used for
221 different RAID sets.
222 This step
223 .Em MUST
224 be performed when a new RAID set is created.
225 .It Fl p Ar dev
226 Check the status of the parity on the RAID set.
227 Displays a status message,
228 and returns successfully if the parity is up-to-date.
229 .It Fl P Ar dev
230 Check the status of the parity on the RAID set, and initialize
231 (re-write) the parity if the parity is not known to be up-to-date.
232 This is normally used after a system crash (and before a
233 .Xr fsck 8 )
234 to ensure the integrity of the parity.
235 .It Fl r Ar component Ar dev
236 Remove the spare disk specified by
237 .Ar component
238 from the set of available spare components.
239 .It Fl R Ar component Ar dev
240 Fails the specified
241 .Ar component ,
242 if necessary, and immediately begins a reconstruction back to
243 .Ar component .
244 This is useful for reconstructing back onto a component after
245 it has been replaced following a failure.
246 .It Fl s Ar dev
247 Display the status of the RAIDframe device for each of the components
248 and spares.
249 .It Fl S Ar dev
250 Check the status of parity re-writing, component reconstruction, and
251 component copyback.
252 The output indicates the amount of progress
253 achieved in each of these areas.
254 .It Fl u Ar dev
255 Unconfigure the RAIDframe device.
256 This does not remove any component labels or change any configuration
257 settings (e.g. auto-configuration settings) for the RAID set.
258 .It Fl v
259 Be more verbose.
260 For operations such as reconstructions, parity
261 re-writing, and copybacks, provide a progress indicator.
262 .El
263 .Pp
264 The device used by
265 .Nm
266 is specified by
267 .Ar dev .
268 .Ar dev
269 may be either the full name of the device, e.g.,
270 .Pa /dev/rraid0d ,
271 for the i386 architecture, or
272 .Pa /dev/rraid0c
273 for many others, or just simply
274 .Pa raid0
275 (for
276 .Pa /dev/rraid0[cd] ) .
277 It is recommended that the partitions used to represent the
278 RAID device are not used for file systems.
279 .Ss Configuration file
280 The format of the configuration file is complex, and
281 only an abbreviated treatment is given here.
282 In the configuration files, a
283 .Sq #
284 indicates the beginning of a comment.
285 .Pp
286 There are 4 required sections of a configuration file, and 2
287 optional sections.
288 Each section begins with a
289 .Sq START ,
290 followed by the section name,
291 and the configuration parameters associated with that section.
292 The first section is the
293 .Sq array
294 section, and it specifies
295 the number of rows, columns, and spare disks in the RAID set.
296 For example:
297 .Bd -literal -offset indent
298 START array
299 1 3 0
300 .Ed
301 .Pp
302 indicates an array with 1 row, 3 columns, and 0 spare disks.
303 Note that although multi-dimensional arrays may be specified, they are
304 .Em NOT
305 supported in the driver.
306 .Pp
307 The second section, the
308 .Sq disks
309 section, specifies the actual components of the device.
310 For example:
311 .Bd -literal -offset indent
312 START disks
313 /dev/sd0e
314 /dev/sd1e
315 /dev/sd2e
316 .Ed
317 .Pp
318 specifies the three component disks to be used in the RAID device.
319 If any of the specified drives cannot be found when the RAID device is
320 configured, then they will be marked as
321 .Sq failed ,
322 and the system will operate in degraded mode.
323 Note that it is
324 .Em imperative
325 that the order of the components in the configuration file does not
326 change between configurations of a RAID device.
327 Changing the order of the components will result in data loss
328 if the set is configured with the
329 .Fl C
330 option.
331 In normal circumstances, the RAID set will not configure if only
332 .Fl c
333 is specified, and the components are out-of-order.
334 .Pp
335 The next section, which is the
336 .Sq spare
337 section, is optional, and, if present, specifies the devices to be used as
338 .Sq hot spares
339 \(em devices which are on-line,
340 but are not actively used by the RAID driver unless
341 one of the main components fail.
342 A simple
343 .Sq spare
344 section might be:
345 .Bd -literal -offset indent
346 START spare
347 /dev/sd3e
348 .Ed
349 .Pp
350 for a configuration with a single spare component.
351 If no spare drives are to be used in the configuration, then the
352 .Sq spare
353 section may be omitted.
354 .Pp
355 The next section is the
356 .Sq layout
357 section.
358 This section describes the general layout parameters for the RAID device,
359 and provides such information as
360 sectors per stripe unit,
361 stripe units per parity unit,
362 stripe units per reconstruction unit,
363 and the parity configuration to use.
364 This section might look like:
365 .Bd -literal -offset indent
366 START layout
367 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level
368 32 1 1 5
369 .Ed
370 .Pp
371 The sectors per stripe unit specifies, in blocks, the interleave
372 factor; i.e., the number of contiguous sectors to be written to each
373 component for a single stripe.
374 Appropriate selection of this value (32 in this example)
375 is the subject of much research in RAID architectures.
376 The stripe units per parity unit and
377 stripe units per reconstruction unit are normally each set to 1.
378 While certain values above 1 are permitted, a discussion of valid
379 values and the consequences of using anything other than 1 are outside
380 the scope of this document.
381 The last value in this section (5 in this example)
382 indicates the parity configuration desired.
383 Valid entries include:
384 .Bl -tag -width inde
385 .It 0
386 RAID level 0.
387 No parity, only simple striping.
388 .It 1
389 RAID level 1.
390 Mirroring.
391 The parity is the mirror.
392 .It 4
393 RAID level 4.
394 Striping across components, with parity stored on the last component.
395 .It 5
396 RAID level 5.
397 Striping across components, parity distributed across all components.
398 .El
399 .Pp
400 There are other valid entries here, including those for Even-Odd
401 parity, RAID level 5 with rotated sparing, Chained declustering,
402 and Interleaved declustering, but as of this writing the code for
403 those parity operations has not been tested with
404 .Nx .
405 .Pp
406 The next required section is the
407 .Sq queue
408 section.
409 This is most often specified as:
410 .Bd -literal -offset indent
411 START queue
412 fifo 100
413 .Ed
414 .Pp
415 where the queuing method is specified as fifo (first-in, first-out),
416 and the size of the per-component queue is limited to 100 requests.
417 Other queuing methods may also be specified, but a discussion of them
418 is beyond the scope of this document.
419 .Pp
420 The final section, the
421 .Sq debug
422 section, is optional.
423 For more details on this the reader is referred to
424 the RAIDframe documentation discussed in the
425 .Sx HISTORY
426 section.
427 .Pp
428 See
429 .Sx EXAMPLES
430 for a more complete configuration file example.
431 .Sh FILES
432 .Bl -tag -width /dev/XXrXraidX -compact
433 .It Pa /dev/{,r}raid*
434 .Cm raid
435 device special files.
436 .El
437 .Sh EXAMPLES
438 It is highly recommended that before using the RAID driver for real
439 file systems that the system administrator(s) become quite familiar
440 with the use of
441 .Nm ,
442 and that they understand how the component reconstruction process works.
443 The examples in this section will focus on configuring a
444 number of different RAID sets of varying degrees of redundancy.
445 By working through these examples, administrators should be able to
446 develop a good feel for how to configure a RAID set, and how to
447 initiate reconstruction of failed components.
448 .Pp
449 In the following examples
450 .Sq raid0
451 will be used to denote the RAID device.
452 Depending on the architecture,
453 .Pa /dev/rraid0c
454 or
455 .Pa /dev/rraid0d
456 may be used in place of
457 .Pa raid0 .
458 .Ss Initialization and Configuration
459 The initial step in configuring a RAID set is to identify the components
460 that will be used in the RAID set.
461 All components should be the same size.
462 Each component should have a disklabel type of
463 .Dv FS_RAID ,
464 and a typical disklabel entry for a RAID component might look like:
465 .Bd -literal -offset indent
466 f: 1800000 200495 RAID # (Cyl. 405*- 4041*)
467 .Ed
468 .Pp
469 While
470 .Dv FS_BSDFFS
471 will also work as the component type, the type
472 .Dv FS_RAID
473 is preferred for RAIDframe use, as it is required for features such as
474 auto-configuration.
475 As part of the initial configuration of each RAID set,
476 each component will be given a
477 .Sq component label .
478 A
479 .Sq component label
480 contains important information about the component, including a
481 user-specified serial number, the row and column of that component in
482 the RAID set, the redundancy level of the RAID set, a
483 .Sq modification counter ,
484 and whether the parity information (if any) on that
485 component is known to be correct.
486 Component labels are an integral part of the RAID set,
487 since they are used to ensure that components
488 are configured in the correct order, and used to keep track of other
489 vital information about the RAID set.
490 Component labels are also required for the auto-detection
491 and auto-configuration of RAID sets at boot time.
492 For a component label to be considered valid, that
493 particular component label must be in agreement with the other
494 component labels in the set.
495 For example, the serial number,
496 .Sq modification counter ,
497 number of rows and number of columns must all be in agreement.
498 If any of these are different, then the component is
499 not considered to be part of the set.
500 See
501 .Xr raid 4
502 for more information about component labels.
503 .Pp
504 Once the components have been identified, and the disks have
505 appropriate labels,
506 .Nm
507 is then used to configure the
508 .Xr raid 4
509 device.
510 To configure the device, a configuration file which looks something like:
511 .Bd -literal -offset indent
512 START array
513 # numRow numCol numSpare
514 1 3 1
516 START disks
517 /dev/sd1e
518 /dev/sd2e
519 /dev/sd3e
521 START spare
522 /dev/sd4e
524 START layout
525 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_5
526 32 1 1 5
528 START queue
529 fifo 100
530 .Ed
531 .Pp
532 is created in a file.
533 The above configuration file specifies a RAID 5
534 set consisting of the components
535 .Pa /dev/sd1e ,
536 .Pa /dev/sd2e ,
537 and
538 .Pa /dev/sd3e ,
539 with
540 .Pa /dev/sd4e
541 available as a
542 .Sq hot spare
543 in case one of the three main drives should fail.
544 A RAID 0 set would be specified in a similar way:
545 .Bd -literal -offset indent
546 START array
547 # numRow numCol numSpare
548 1 4 0
550 START disks
551 /dev/sd10e
552 /dev/sd11e
553 /dev/sd12e
554 /dev/sd13e
556 START layout
557 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_0
558 64 1 1 0
560 START queue
561 fifo 100
562 .Ed
563 .Pp
564 In this case, devices
565 .Pa /dev/sd10e ,
566 .Pa /dev/sd11e ,
567 .Pa /dev/sd12e ,
568 and
569 .Pa /dev/sd13e
570 are the components that make up this RAID set.
571 Note that there are no hot spares for a RAID 0 set,
572 since there is no way to recover data if any of the components fail.
573 .Pp
574 For a RAID 1 (mirror) set, the following configuration might be used:
575 .Bd -literal -offset indent
576 START array
577 # numRow numCol numSpare
578 1 2 0
580 START disks
581 /dev/sd20e
582 /dev/sd21e
584 START layout
585 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_1
586 128 1 1 1
588 START queue
589 fifo 100
590 .Ed
591 .Pp
592 In this case,
593 .Pa /dev/sd20e
594 and
595 .Pa /dev/sd21e
596 are the two components of the mirror set.
597 While no hot spares have been specified in this
598 configuration, they easily could be, just as they were specified in
599 the RAID 5 case above.
600 Note as well that RAID 1 sets are currently limited to only 2 components.
601 At present, n-way mirroring is not possible.
602 .Pp
603 The first time a RAID set is configured, the
604 .Fl C
605 option must be used:
606 .Bd -literal -offset indent
607 raidctl -C raid0.conf raid0
608 .Ed
609 .Pp
610 where
611 .Pa raid0.conf
612 is the name of the RAID configuration file.
613 The
614 .Fl C
615 forces the configuration to succeed, even if any of the component
616 labels are incorrect.
617 The
618 .Fl C
619 option should not be used lightly in
620 situations other than initial configurations, as if
621 the system is refusing to configure a RAID set, there is probably a
622 very good reason for it.
623 After the initial configuration is done (and
624 appropriate component labels are added with the
625 .Fl I
626 option) then raid0 can be configured normally with:
627 .Bd -literal -offset indent
628 raidctl -c raid0.conf raid0
629 .Ed
630 .Pp
631 When the RAID set is configured for the first time, it is
632 necessary to initialize the component labels, and to initialize the
633 parity on the RAID set.
634 Initializing the component labels is done with:
635 .Bd -literal -offset indent
636 raidctl -I 112341 raid0
637 .Ed
638 .Pp
639 where
640 .Sq 112341
641 is a user-specified serial number for the RAID set.
642 This initialization step is
643 .Em required
644 for all RAID sets.
645 As well, using different serial numbers between RAID sets is
646 .Em strongly encouraged ,
647 as using the same serial number for all RAID sets will only serve to
648 decrease the usefulness of the component label checking.
649 .Pp
650 Initializing the RAID set is done via the
651 .Fl i
652 option.
653 This initialization
654 .Em MUST
655 be done for
656 .Em all
657 RAID sets, since among other things it verifies that the parity (if
658 any) on the RAID set is correct.
659 Since this initialization may be quite time-consuming, the
660 .Fl v
661 option may be also used in conjunction with
662 .Fl i :
663 .Bd -literal -offset indent
664 raidctl -iv raid0
665 .Ed
666 .Pp
667 This will give more verbose output on the
668 status of the initialization:
669 .Bd -literal -offset indent
670 Initiating re-write of parity
671 Parity Re-write status:
672 10% |**** | ETA: 06:03 /
673 .Ed
674 .Pp
675 The output provides a
676 .Sq Percent Complete
677 in both a numeric and graphical format, as well as an estimated time
678 to completion of the operation.
679 .Pp
680 Since it is the parity that provides the
681 .Sq redundancy
682 part of RAID, it is critical that the parity is correct as much as possible.
683 If the parity is not correct, then there is no
684 guarantee that data will not be lost if a component fails.
685 .Pp
686 Once the parity is known to be correct, it is then safe to perform
687 .Xr disklabel 8 ,
688 .Xr newfs 8 ,
689 or
690 .Xr fsck 8
691 on the device or its file systems, and then to mount the file systems
692 for use.
693 .Pp
694 Under certain circumstances (e.g., the additional component has not
695 arrived, or data is being migrated off of a disk destined to become a
696 component) it may be desirable to configure a RAID 1 set with only
697 a single component.
698 This can be achieved by using the word
699 .Dq absent
700 to indicate that a particular component is not present.
701 In the following:
702 .Bd -literal -offset indent
703 START array
704 # numRow numCol numSpare
705 1 2 0
707 START disks
708 absent
709 /dev/sd0e
711 START layout
712 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_1
713 128 1 1 1
715 START queue
716 fifo 100
717 .Ed
718 .Pp
719 .Pa /dev/sd0e
720 is the real component, and will be the second disk of a RAID 1 set.
721 The first component is simply marked as being absent.
722 Configuration (using
723 .Fl C
724 and
725 .Fl I Ar 12345
726 as above) proceeds normally, but initialization of the RAID set will
727 have to wait until all physical components are present.
728 After configuration, this set can be used normally, but will be operating
729 in degraded mode.
730 Once a second physical component is obtained, it can be hot-added,
731 the existing data mirrored, and normal operation resumed.
732 .Pp
733 The size of the resulting RAID set will depend on the number of data
734 components in the set.
735 Space is automatically reserved for the component labels, and
736 the actual amount of space used
737 for data on a component will be rounded down to the largest possible
738 multiple of the sectors per stripe unit (sectPerSU) value.
739 Thus, the amount of space provided by the RAID set will be less
740 than the sum of the size of the components.
741 .Ss Maintenance of the RAID set
742 After the parity has been initialized for the first time, the command:
743 .Bd -literal -offset indent
744 raidctl -p raid0
745 .Ed
746 .Pp
747 can be used to check the current status of the parity.
748 To check the parity and rebuild it necessary (for example,
749 after an unclean shutdown) the command:
750 .Bd -literal -offset indent
751 raidctl -P raid0
752 .Ed
753 .Pp
754 is used.
755 Note that re-writing the parity can be done while
756 other operations on the RAID set are taking place (e.g., while doing a
757 .Xr fsck 8
758 on a file system on the RAID set).
759 However: for maximum effectiveness of the RAID set, the parity should be
760 known to be correct before any data on the set is modified.
761 .Pp
762 To see how the RAID set is doing, the following command can be used to
763 show the RAID set's status:
764 .Bd -literal -offset indent
765 raidctl -s raid0
766 .Ed
767 .Pp
768 The output will look something like:
769 .Bd -literal -offset indent
770 Components:
771 /dev/sd1e: optimal
772 /dev/sd2e: optimal
773 /dev/sd3e: optimal
774 Spares:
775 /dev/sd4e: spare
776 Component label for /dev/sd1e:
777 Row: 0 Column: 0 Num Rows: 1 Num Columns: 3
778 Version: 2 Serial Number: 13432 Mod Counter: 65
779 Clean: No Status: 0
780 sectPerSU: 32 SUsPerPU: 1 SUsPerRU: 1
781 RAID Level: 5 blocksize: 512 numBlocks: 1799936
782 Autoconfig: No
783 Last configured as: raid0
784 Component label for /dev/sd2e:
785 Row: 0 Column: 1 Num Rows: 1 Num Columns: 3
786 Version: 2 Serial Number: 13432 Mod Counter: 65
787 Clean: No Status: 0
788 sectPerSU: 32 SUsPerPU: 1 SUsPerRU: 1
789 RAID Level: 5 blocksize: 512 numBlocks: 1799936
790 Autoconfig: No
791 Last configured as: raid0
792 Component label for /dev/sd3e:
793 Row: 0 Column: 2 Num Rows: 1 Num Columns: 3
794 Version: 2 Serial Number: 13432 Mod Counter: 65
795 Clean: No Status: 0
796 sectPerSU: 32 SUsPerPU: 1 SUsPerRU: 1
797 RAID Level: 5 blocksize: 512 numBlocks: 1799936
798 Autoconfig: No
799 Last configured as: raid0
800 Parity status: clean
801 Reconstruction is 100% complete.
802 Parity Re-write is 100% complete.
803 Copyback is 100% complete.
804 .Ed
805 .Pp
806 This indicates that all is well with the RAID set.
807 Of importance here are the component lines which read
808 .Sq optimal ,
809 and the
810 .Sq Parity status
811 line.
812 .Sq Parity status: clean
813 indicates that the parity is up-to-date for this RAID set,
814 whether or not the RAID set is in redundant or degraded mode.
815 .Sq Parity status: DIRTY
816 indicates that it is not known if the parity information is
817 consistent with the data, and that the parity information needs
818 to be checked.
819 Note that if there are file systems open on the RAID set,
820 the individual components will not be
821 .Sq clean
822 but the set as a whole can still be clean.
823 .Pp
824 To check the component label of
825 .Pa /dev/sd1e ,
826 the following is used:
827 .Bd -literal -offset indent
828 raidctl -g /dev/sd1e raid0
829 .Ed
830 .Pp
831 The output of this command will look something like:
832 .Bd -literal -offset indent
833 Component label for /dev/sd1e:
834 Row: 0 Column: 0 Num Rows: 1 Num Columns: 3
835 Version: 2 Serial Number: 13432 Mod Counter: 65
836 Clean: No Status: 0
837 sectPerSU: 32 SUsPerPU: 1 SUsPerRU: 1
838 RAID Level: 5 blocksize: 512 numBlocks: 1799936
839 Autoconfig: No
840 Last configured as: raid0
841 .Ed
842 .Ss Dealing with Component Failures
843 If for some reason
844 (perhaps to test reconstruction) it is necessary to pretend a drive
845 has failed, the following will perform that function:
846 .Bd -literal -offset indent
847 raidctl -f /dev/sd2e raid0
848 .Ed
849 .Pp
850 The system will then be performing all operations in degraded mode,
851 where missing data is re-computed from existing data and the parity.
852 In this case, obtaining the status of raid0 will return (in part):
853 .Bd -literal -offset indent
854 Components:
855 /dev/sd1e: optimal
856 /dev/sd2e: failed
857 /dev/sd3e: optimal
858 Spares:
859 /dev/sd4e: spare
860 .Ed
861 .Pp
862 Note that with the use of
863 .Fl f
864 a reconstruction has not been started.
865 To both fail the disk and start a reconstruction, the
866 .Fl F
867 option must be used:
868 .Bd -literal -offset indent
869 raidctl -F /dev/sd2e raid0
870 .Ed
871 .Pp
872 The
873 .Fl f
874 option may be used first, and then the
875 .Fl F
876 option used later, on the same disk, if desired.
877 Immediately after the reconstruction is started, the status will report:
878 .Bd -literal -offset indent
879 Components:
880 /dev/sd1e: optimal
881 /dev/sd2e: reconstructing
882 /dev/sd3e: optimal
883 Spares:
884 /dev/sd4e: used_spare
885 [...]
886 Parity status: clean
887 Reconstruction is 10% complete.
888 Parity Re-write is 100% complete.
889 Copyback is 100% complete.
890 .Ed
891 .Pp
892 This indicates that a reconstruction is in progress.
893 To find out how the reconstruction is progressing the
894 .Fl S
895 option may be used.
896 This will indicate the progress in terms of the
897 percentage of the reconstruction that is completed.
898 When the reconstruction is finished the
899 .Fl s
900 option will show:
901 .Bd -literal -offset indent
902 Components:
903 /dev/sd1e: optimal
904 /dev/sd2e: spared
905 /dev/sd3e: optimal
906 Spares:
907 /dev/sd4e: used_spare
908 [...]
909 Parity status: clean
910 Reconstruction is 100% complete.
911 Parity Re-write is 100% complete.
912 Copyback is 100% complete.
913 .Ed
914 .Pp
915 At this point there are at least two options.
916 First, if
917 .Pa /dev/sd2e
918 is known to be good (i.e., the failure was either caused by
919 .Fl f
920 or
921 .Fl F ,
922 or the failed disk was replaced), then a copyback of the data can
923 be initiated with the
924 .Fl B
925 option.
926 In this example, this would copy the entire contents of
927 .Pa /dev/sd4e
928 to
929 .Pa /dev/sd2e .
930 Once the copyback procedure is complete, the
931 status of the device would be (in part):
932 .Bd -literal -offset indent
933 Components:
934 /dev/sd1e: optimal
935 /dev/sd2e: optimal
936 /dev/sd3e: optimal
937 Spares:
938 /dev/sd4e: spare
939 .Ed
940 .Pp
941 and the system is back to normal operation.
942 .Pp
943 The second option after the reconstruction is to simply use
944 .Pa /dev/sd4e
945 in place of
946 .Pa /dev/sd2e
947 in the configuration file.
948 For example, the configuration file (in part) might now look like:
949 .Bd -literal -offset indent
950 START array
951 1 3 0
953 START drives
954 /dev/sd1e
955 /dev/sd4e
956 /dev/sd3e
957 .Ed
958 .Pp
959 This can be done as
960 .Pa /dev/sd4e
961 is completely interchangeable with
962 .Pa /dev/sd2e
963 at this point.
964 Note that extreme care must be taken when
965 changing the order of the drives in a configuration.
966 This is one of the few instances where the devices and/or
967 their orderings can be changed without loss of data!
968 In general, the ordering of components in a configuration file should
969 .Em never
970 be changed.
971 .Pp
972 If a component fails and there are no hot spares
973 available on-line, the status of the RAID set might (in part) look like:
974 .Bd -literal -offset indent
975 Components:
976 /dev/sd1e: optimal
977 /dev/sd2e: failed
978 /dev/sd3e: optimal
979 No spares.
980 .Ed
981 .Pp
982 In this case there are a number of options.
983 The first option is to add a hot spare using:
984 .Bd -literal -offset indent
985 raidctl -a /dev/sd4e raid0
986 .Ed
987 .Pp
988 After the hot add, the status would then be:
989 .Bd -literal -offset indent
990 Components:
991 /dev/sd1e: optimal
992 /dev/sd2e: failed
993 /dev/sd3e: optimal
994 Spares:
995 /dev/sd4e: spare
996 .Ed
997 .Pp
998 Reconstruction could then take place using
999 .Fl F
1000 as describe above.
1001 .Pp
1002 A second option is to rebuild directly onto
1003 .Pa /dev/sd2e .
1004 Once the disk containing
1005 .Pa /dev/sd2e
1006 has been replaced, one can simply use:
1007 .Bd -literal -offset indent
1008 raidctl -R /dev/sd2e raid0
1009 .Ed
1010 .Pp
1011 to rebuild the
1012 .Pa /dev/sd2e
1013 component.
1014 As the rebuilding is in progress, the status will be:
1015 .Bd -literal -offset indent
1016 Components:
1017 /dev/sd1e: optimal
1018 /dev/sd2e: reconstructing
1019 /dev/sd3e: optimal
1020 No spares.
1021 .Ed
1022 .Pp
1023 and when completed, will be:
1024 .Bd -literal -offset indent
1025 Components:
1026 /dev/sd1e: optimal
1027 /dev/sd2e: optimal
1028 /dev/sd3e: optimal
1029 No spares.
1030 .Ed
1031 .Pp
1032 In circumstances where a particular component is completely
1033 unavailable after a reboot, a special component name will be used to
1034 indicate the missing component.
1035 For example:
1036 .Bd -literal -offset indent
1037 Components:
1038 /dev/sd2e: optimal
1039 component1: failed
1040 No spares.
1041 .Ed
1042 .Pp
1043 indicates that the second component of this RAID set was not detected
1044 at all by the auto-configuration code.
1045 The name
1046 .Sq component1
1047 can be used anywhere a normal component name would be used.
1048 For example, to add a hot spare to the above set, and rebuild to that hot
1049 spare, the following could be done:
1050 .Bd -literal -offset indent
1051 raidctl -a /dev/sd3e raid0
1052 raidctl -F component1 raid0
1053 .Ed
1054 .Pp
1055 at which point the data missing from
1056 .Sq component1
1057 would be reconstructed onto
1058 .Pa /dev/sd3e .
1059 .Pp
1060 When more than one component is marked as
1061 .Sq failed
1062 due to a non-component hardware failure (e.g., loss of power to two
1063 components, adapter problems, termination problems, or cabling issues) it
1064 is quite possible to recover the data on the RAID set.
1065 The first thing to be aware of is that the first disk to fail will
1066 almost certainly be out-of-sync with the remainder of the array.
1067 If any IO was performed between the time the first component is considered
1068 .Sq failed
1069 and when the second component is considered
1070 .Sq failed ,
1071 then the first component to fail will
1072 .Em not
1073 contain correct data, and should be ignored.
1074 When the second component is marked as failed, however, the RAID device will
1075 (currently) panic the system.
1076 At this point the data on the RAID set
1077 (not including the first failed component) is still self consistent,
1078 and will be in no worse state of repair than had the power gone out in
1079 the middle of a write to a file system on a non-RAID device.
1080 The problem, however, is that the component labels may now have 3 different
1081 .Sq modification counters
1082 (one value on the first component that failed, one value on the second
1083 component that failed, and a third value on the remaining components).
1084 In such a situation, the RAID set will not autoconfigure,
1085 and can only be forcibly re-configured
1086 with the
1087 .Fl C
1088 option.
1089 To recover the RAID set, one must first remedy whatever physical
1090 problem caused the multiple-component failure.
1091 After that is done, the RAID set can be restored by forcibly
1092 configuring the raid set
1093 .Em without
1094 the component that failed first.
1095 For example, if
1096 .Pa /dev/sd1e
1097 and
1098 .Pa /dev/sd2e
1099 fail (in that order) in a RAID set of the following configuration:
1100 .Bd -literal -offset indent
1101 START array
1102 1 4 0
1104 START drives
1105 /dev/sd1e
1106 /dev/sd2e
1107 /dev/sd3e
1108 /dev/sd4e
1110 START layout
1111 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_5
1112 64 1 1 5
1114 START queue
1115 fifo 100
1117 .Ed
1118 .Pp
1119 then the following configuration (say "recover_raid0.conf")
1120 .Bd -literal -offset indent
1121 START array
1122 1 4 0
1124 START drives
1125 absent
1126 /dev/sd2e
1127 /dev/sd3e
1128 /dev/sd4e
1130 START layout
1131 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_5
1132 64 1 1 5
1134 START queue
1135 fifo 100
1136 .Ed
1137 .Pp
1138 can be used with
1139 .Bd -literal -offset indent
1140 raidctl -C recover_raid0.conf raid0
1141 .Ed
1142 .Pp
1143 to force the configuration of raid0.
1144 A
1145 .Bd -literal -offset indent
1146 raidctl -I 12345 raid0
1147 .Ed
1148 .Pp
1149 will be required in order to synchronize the component labels.
1150 At this point the file systems on the RAID set can then be checked and
1151 corrected.
1152 To complete the re-construction of the RAID set,
1153 .Pa /dev/sd1e
1154 is simply hot-added back into the array, and reconstructed
1155 as described earlier.
1156 .Ss RAID on RAID
1157 RAID sets can be layered to create more complex and much larger RAID sets.
1158 A RAID 0 set, for example, could be constructed from four RAID 5 sets.
1159 The following configuration file shows such a setup:
1160 .Bd -literal -offset indent
1161 START array
1162 # numRow numCol numSpare
1163 1 4 0
1165 START disks
1166 /dev/raid1e
1167 /dev/raid2e
1168 /dev/raid3e
1169 /dev/raid4e
1171 START layout
1172 # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_0
1173 128 1 1 0
1175 START queue
1176 fifo 100
1177 .Ed
1178 .Pp
1179 A similar configuration file might be used for a RAID 0 set
1180 constructed from components on RAID 1 sets.
1181 In such a configuration, the mirroring provides a high degree
1182 of redundancy, while the striping provides additional speed benefits.
1183 .Ss Auto-configuration and Root on RAID
1184 RAID sets can also be auto-configured at boot.
1185 To make a set auto-configurable,
1186 simply prepare the RAID set as above, and then do a:
1187 .Bd -literal -offset indent
1188 raidctl -A yes raid0
1189 .Ed
1190 .Pp
1191 to turn on auto-configuration for that set.
1192 To turn off auto-configuration, use:
1193 .Bd -literal -offset indent
1194 raidctl -A no raid0
1195 .Ed
1196 .Pp
1197 RAID sets which are auto-configurable will be configured before the
1198 root file system is mounted.
1199 These RAID sets are thus available for
1200 use as a root file system, or for any other file system.
1201 A primary advantage of using the auto-configuration is that RAID components
1202 become more independent of the disks they reside on.
1203 For example, SCSI ID's can change, but auto-configured sets will always be
1204 configured correctly, even if the SCSI ID's of the component disks
1205 have become scrambled.
1206 .Pp
1207 Having a system's root file system
1208 .Pq Pa /
1209 on a RAID set is also allowed, with the
1210 .Sq a
1211 partition of such a RAID set being used for
1212 .Pa / .
1213 To use raid0a as the root file system, simply use:
1214 .Bd -literal -offset indent
1215 raidctl -A root raid0
1216 .Ed
1217 .Pp
1218 To return raid0a to be just an auto-configuring set simply use the
1219 .Fl A Ar yes
1220 arguments.
1221 .Pp
1222 Note that kernels can only be directly read from RAID 1 components on
1223 architectures that support that
1224 .Pq currently alpha, i386, pmax, sparc, sparc64, and vax .
1225 On those architectures, the
1226 .Dv FS_RAID
1227 file system is recognized by the bootblocks, and will properly load the
1228 kernel directly from a RAID 1 component.
1229 For other architectures, or to support the root file system
1230 on other RAID sets, some other mechanism must be used to get a kernel booting.
1231 For example, a small partition containing only the secondary boot-blocks
1232 and an alternate kernel (or two) could be used.
1233 Once a kernel is booting however, and an auto-configuring RAID set is
1234 found that is eligible to be root, then that RAID set will be
1235 auto-configured and used as the root device.
1236 If two or more RAID sets claim to be root devices, then the
1237 user will be prompted to select the root device.
1238 At this time, RAID 0, 1, 4, and 5 sets are all supported as root devices.
1239 .Pp
1240 A typical RAID 1 setup with root on RAID might be as follows:
1241 .Bl -enum
1242 .It
1243 wd0a - a small partition, which contains a complete, bootable, basic
1244 .Nx
1245 installation.
1246 .It
1247 wd1a - also contains a complete, bootable, basic
1248 .Nx
1249 installation.
1250 .It
1251 wd0e and wd1e - a RAID 1 set, raid0, used for the root file system.
1252 .It
1253 wd0f and wd1f - a RAID 1 set, raid1, which will be used only for
1254 swap space.
1255 .It
1256 wd0g and wd1g - a RAID 1 set, raid2, used for
1257 .Pa /usr ,
1258 .Pa /home ,
1259 or other data, if desired.
1260 .It
1261 wd0h and wd1h - a RAID 1 set, raid3, if desired.
1262 .El
1263 .Pp
1264 RAID sets raid0, raid1, and raid2 are all marked as auto-configurable.
1265 raid0 is marked as being a root file system.
1266 When new kernels are installed, the kernel is not only copied to
1267 .Pa / ,
1268 but also to wd0a and wd1a.
1269 The kernel on wd0a is required, since that
1270 is the kernel the system boots from.
1271 The kernel on wd1a is also
1272 required, since that will be the kernel used should wd0 fail.
1273 The important point here is to have redundant copies of the kernel
1274 available, in the event that one of the drives fail.
1275 .Pp
1276 There is no requirement that the root file system be on the same disk
1277 as the kernel.
1278 For example, obtaining the kernel from wd0a, and using
1279 sd0e and sd1e for raid0, and the root file system, is fine.
1280 It
1281 .Em is
1282 critical, however, that there be multiple kernels available, in the
1283 event of media failure.
1284 .Pp
1285 Multi-layered RAID devices (such as a RAID 0 set made
1286 up of RAID 1 sets) are
1287 .Em not
1288 supported as root devices or auto-configurable devices at this point.
1289 (Multi-layered RAID devices
1290 .Em are
1291 supported in general, however, as mentioned earlier.)
1292 Note that in order to enable component auto-detection and
1293 auto-configuration of RAID devices, the line:
1294 .Bd -literal -offset indent
1295 options RAID_AUTOCONFIG
1296 .Ed
1297 .Pp
1298 must be in the kernel configuration file.
1299 See
1300 .Xr raid 4
1301 for more details.
1302 .Ss Swapping on RAID
1303 A RAID device can be used as a swap device.
1304 In order to ensure that a RAID device used as a swap device
1305 is correctly unconfigured when the system is shutdown or rebooted,
1306 it is recommended that the line
1307 .Bd -literal -offset indent
1308 swapoff=YES
1309 .Ed
1310 .Pp
1311 be added to
1312 .Pa /etc/rc.conf .
1313 .Ss Unconfiguration
1314 The final operation performed by
1315 .Nm
1316 is to unconfigure a
1317 .Xr raid 4
1318 device.
1319 This is accomplished via a simple:
1320 .Bd -literal -offset indent
1321 raidctl -u raid0
1322 .Ed
1323 .Pp
1324 at which point the device is ready to be reconfigured.
1325 .Ss Performance Tuning
1326 Selection of the various parameter values which result in the best
1327 performance can be quite tricky, and often requires a bit of
1328 trial-and-error to get those values most appropriate for a given system.
1329 A whole range of factors come into play, including:
1330 .Bl -enum
1331 .It
1332 Types of components (e.g., SCSI vs. IDE) and their bandwidth
1333 .It
1334 Types of controller cards and their bandwidth
1335 .It
1336 Distribution of components among controllers
1337 .It
1338 IO bandwidth
1339 .It
1340 file system access patterns
1341 .It
1342 CPU speed
1343 .El
1344 .Pp
1345 As with most performance tuning, benchmarking under real-life loads
1346 may be the only way to measure expected performance.
1347 Understanding some of the underlying technology is also useful in tuning.
1348 The goal of this section is to provide pointers to those parameters which may
1349 make significant differences in performance.
1350 .Pp
1351 For a RAID 1 set, a SectPerSU value of 64 or 128 is typically sufficient.
1352 Since data in a RAID 1 set is arranged in a linear
1353 fashion on each component, selecting an appropriate stripe size is
1354 somewhat less critical than it is for a RAID 5 set.
1355 However: a stripe size that is too small will cause large IO's to be
1356 broken up into a number of smaller ones, hurting performance.
1357 At the same time, a large stripe size may cause problems with
1358 concurrent accesses to stripes, which may also affect performance.
1359 Thus values in the range of 32 to 128 are often the most effective.
1360 .Pp
1361 Tuning RAID 5 sets is trickier.
1362 In the best case, IO is presented to the RAID set one stripe at a time.
1363 Since the entire stripe is available at the beginning of the IO,
1364 the parity of that stripe can be calculated before the stripe is written,
1365 and then the stripe data and parity can be written in parallel.
1366 When the amount of data being written is less than a full stripe worth, the
1367 .Sq small write
1368 problem occurs.
1369 Since a
1370 .Sq small write
1371 means only a portion of the stripe on the components is going to
1372 change, the data (and parity) on the components must be updated
1373 slightly differently.
1374 First, the
1375 .Sq old parity
1376 and
1377 .Sq old data
1378 must be read from the components.
1379 Then the new parity is constructed,
1380 using the new data to be written, and the old data and old parity.
1381 Finally, the new data and new parity are written.
1382 All this extra data shuffling results in a serious loss of performance,
1383 and is typically 2 to 4 times slower than a full stripe write (or read).
1384 To combat this problem in the real world, it may be useful
1385 to ensure that stripe sizes are small enough that a
1386 .Sq large IO
1387 from the system will use exactly one large stripe write.
1388 As is seen later, there are some file system dependencies
1389 which may come into play here as well.
1390 .Pp
1391 Since the size of a
1392 .Sq large IO
1393 is often (currently) only 32K or 64K, on a 5-drive RAID 5 set it may
1394 be desirable to select a SectPerSU value of 16 blocks (8K) or 32
1395 blocks (16K).
1396 Since there are 4 data sectors per stripe, the maximum
1397 data per stripe is 64 blocks (32K) or 128 blocks (64K).
1398 Again, empirical measurement will provide the best indicators of which
1399 values will yeild better performance.
1400 .Pp
1401 The parameters used for the file system are also critical to good performance.
1402 For
1403 .Xr newfs 8 ,
1404 for example, increasing the block size to 32K or 64K may improve
1405 performance dramatically.
1406 As well, changing the cylinders-per-group
1407 parameter from 16 to 32 or higher is often not only necessary for
1408 larger file systems, but may also have positive performance implications.
1409 .Ss Summary
1410 Despite the length of this man-page, configuring a RAID set is a
1411 relatively straight-forward process.
1412 All that needs to be done is the following steps:
1413 .Bl -enum
1414 .It
1415 Use
1416 .Xr disklabel 8
1417 to create the components (of type RAID).
1418 .It
1419 Construct a RAID configuration file: e.g.,
1420 .Pa raid0.conf
1421 .It
1422 Configure the RAID set with:
1423 .Bd -literal -offset indent
1424 raidctl -C raid0.conf raid0
1425 .Ed
1426 .Pp
1427 .It
1428 Initialize the component labels with:
1429 .Bd -literal -offset indent
1430 raidctl -I 123456 raid0
1431 .Ed
1432 .Pp
1433 .It
1434 Initialize other important parts of the set with:
1435 .Bd -literal -offset indent
1436 raidctl -i raid0
1437 .Ed
1438 .Pp
1439 .It
1440 Get the default label for the RAID set:
1441 .Bd -literal -offset indent
1442 disklabel raid0 \*[Gt] /tmp/label
1443 .Ed
1444 .Pp
1445 .It
1446 Edit the label:
1447 .Bd -literal -offset indent
1448 vi /tmp/label
1449 .Ed
1450 .Pp
1451 .It
1452 Put the new label on the RAID set:
1453 .Bd -literal -offset indent
1454 disklabel -R -r raid0 /tmp/label
1455 .Ed
1456 .Pp
1457 .It
1458 Create the file system:
1459 .Bd -literal -offset indent
1460 newfs /dev/rraid0e
1461 .Ed
1462 .Pp
1463 .It
1464 Mount the file system:
1465 .Bd -literal -offset indent
1466 mount /dev/raid0e /mnt
1467 .Ed
1468 .Pp
1469 .It
1470 Use:
1471 .Bd -literal -offset indent
1472 raidctl -c raid0.conf raid0
1473 .Ed
1474 .Pp
1475 To re-configure the RAID set the next time it is needed, or put
1476 .Pa raid0.conf
1477 into
1478 .Pa /etc
1479 where it will automatically be started by the
1480 .Pa /etc/rc.d
1481 scripts.
1482 .El
1483 .Sh SEE ALSO
1484 .Xr ccd 4 ,
1485 .Xr raid 4 ,
1486 .Xr rc 8
1487 .Sh HISTORY
1488 RAIDframe is a framework for rapid prototyping of RAID structures
1489 developed by the folks at the Parallel Data Laboratory at Carnegie
1490 Mellon University (CMU).
1491 A more complete description of the internals and functionality of
1492 RAIDframe is found in the paper "RAIDframe: A Rapid Prototyping Tool
1493 for RAID Systems", by William V. Courtright II, Garth Gibson, Mark
1494 Holland, LeAnn Neal Reilly, and Jim Zelenka, and published by the
1495 Parallel Data Laboratory of Carnegie Mellon University.
1496 .Pp
1497 The
1498 .Nm
1499 command first appeared as a program in CMU's RAIDframe v1.1 distribution.
1500 This version of
1501 .Nm
1502 is a complete re-write, and first appeared in
1503 .Nx 1.4 .
1504 .Sh COPYRIGHT
1505 .Bd -literal
1506 The RAIDframe Copyright is as follows:
1508 Copyright (c) 1994-1996 Carnegie-Mellon University.
1509 All rights reserved.
1511 Permission to use, copy, modify and distribute this software and
1512 its documentation is hereby granted, provided that both the copyright
1513 notice and this permission notice appear in all copies of the
1514 software, derivative works or modified versions, and any portions
1515 thereof, and that both notices appear in supporting documentation.
1517 CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
1518 CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
1519 FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
1521 Carnegie Mellon requests users of this software to return to
1523 Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
1524 School of Computer Science
1525 Carnegie Mellon University
1526 Pittsburgh PA 15213-3890
1528 any improvements or extensions that they make and grant Carnegie the
1529 rights to redistribute these changes.
1530 .Ed
1531 .Sh WARNINGS
1532 Certain RAID levels (1, 4, 5, 6, and others) can protect against some
1533 data loss due to component failure.
1534 However the loss of two components of a RAID 4 or 5 system,
1535 or the loss of a single component of a RAID 0 system will
1536 result in the entire file system being lost.
1537 RAID is
1538 .Em NOT
1539 a substitute for good backup practices.
1540 .Pp
1541 Recomputation of parity
1542 .Em MUST
1543 be performed whenever there is a chance that it may have been compromised.
1544 This includes after system crashes, or before a RAID
1545 device has been used for the first time.
1546 Failure to keep parity correct will be catastrophic should a
1547 component ever fail \(em it is better to use RAID 0 and get the
1548 additional space and speed, than it is to use parity, but
1549 not keep the parity correct.
1550 At least with RAID 0 there is no perception of increased data security.
1551 .Sh BUGS
1552 Hot-spare removal is currently not available.