1.\" $NetBSD: ccd.4,v 1.5 1995/10/09 06:09:09 thorpej Exp $ 2.\" 3.\" Copyright (c) 1994 Jason Downs. 4.\" Copyright (c) 1994, 1995 Jason R. Thorpe. 5.\" All rights reserved. 6.\" 7.\" Redistribution and use in source and binary forms, with or without 8.\" modification, are permitted provided that the following conditions 9.\" are met: 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 the 14.\" documentation and/or other materials provided with the distribution. 15.\" 3. All advertising materials mentioning features or use of this software 16.\" must display the following acknowledgement: 17.\" This product includes software developed for the NetBSD Project 18.\" by Jason Downs and Jason R. Thorpe. 19.\" 4. Neither the name of the author nor the names of its contributors 20.\" may be used to endorse or promote products derived from this software 21.\" without specific prior written permission. 22.\" 23.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 24.\" IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 25.\" OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 26.\" IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 27.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 28.\" BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 29.\" LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 30.\" AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 31.\" OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33.\" SUCH DAMAGE. 34.\" 35.\" $FreeBSD$ 36.\" 37.Dd August 9, 1995 38.Dt CCD 4 39.Os 40.Sh NAME 41.Nm ccd 42.Nd Concatenated Disk driver 43.Sh SYNOPSIS 44.Cd "device ccd" 45.Sh DESCRIPTION 46The 47.Nm 48driver provides the capability of combining one or more disks/partitions 49into one virtual disk. 50.Pp 51This document assumes that you are familiar with how to generate kernels, 52how to properly configure disks and devices in a kernel 53configuration file, and how to partition disks. 54.Pp 55In order to compile in support for the 56.Nm , 57you must add a line similar 58to the following to your kernel configuration file: 59.Pp 60.Dl "device ccd # concatenated disk devices" 61.Pp 62As of the 63.Fx 3.0 64release, you do not need to 65configure your kernel with 66.Nm 67but may instead use it as a kernel loadable 68module. 69Simply running 70.Xr ccdconfig 8 71will load the module into the kernel. 72.Pp 73A 74.Nm 75may be either serially concatenated or interleaved. 76To serially 77concatenate the partitions, specify the interleave factor of 0. 78Note that mirroring may not be used with an interleave factor of 0. 79.Pp 80There is a run-time utility that is used for configuring 81.Nm Ns s . 82See 83.Xr ccdconfig 8 84for more information. 85.Ss The Interleave Factor 86If a 87.Nm 88is interleaved correctly, a 89.Dq striping 90effect is achieved, which can increase sequential read/write 91performance. 92The interleave factor is expressed in units of 93.Dv DEV_BSIZE 94(usually 512 bytes). 95For large writes, the optimum interleave factor 96is typically the size of a track, while for large reads, it is about a 97quarter of a track. 98(Note that this changes greatly depending on the 99number and speed of disks.) 100For instance, with eight 7,200 RPM drives 101on two Fast-Wide SCSI buses, this translates to about 128 for writes 102and 32 for reads. 103A larger interleave tends to work better when the 104disk is taking a multitasking load by localizing the file I/O from 105any given process onto a single disk. 106You lose sequential performance when 107you do this, but sequential performance is not usually an issue with a 108multitasking load. 109.Pp 110An interleave factor must be specified when using a mirroring configuration, 111even when you have only two disks (i.e., the layout winds up being the same 112no matter what the interleave factor). 113The interleave factor will determine 114how I/O is broken up, however, and a value 128 or greater is recommended. 115.Pp 116.Nm 117has an option for a parity disk, but does not currently implement it. 118.Pp 119The best performance is achieved if all component disks have the same 120geometry and size. 121Optimum striping cannot occur with different 122disk types. 123.Pp 124For random-access oriented workloads, such as news servers, a larger 125interleave factor (e.g., 65,536) is more desirable. 126Note that there 127is not much 128.Nm 129can do to speed up applications that are seek-time limited. 130Larger 131interleave factors will at least reduce the chance of having to seek 132two disk-heads to read one directory or a file. 133.Ss Disk Mirroring 134You can configure the 135.Nm 136to 137.Dq mirror 138any even number of disks. 139See 140.Xr ccdconfig 8 141for how to specify the necessary flags. 142For example, if you have a 143.Nm 144configuration specifying four disks, the first two disks will be mirrored with 145the second two disks. 146A write will be run to both sides of 147the mirror. 148A read will be run to either side of the mirror depending 149on what the driver believes to be most optimal. 150If the read fails, 151the driver will automatically attempt to read the same sector from the 152other side of the mirror. 153Currently 154.Nm 155uses a dual seek zone model to optimize reads for a multi-tasking load 156rather than a sequential load. 157.Pp 158In an event of a disk 159failure, you can use 160.Xr dd 1 161to recover the failed disk. 162.Pp 163Note that a one-disk 164.Nm 165is not the same as the original partition. 166In particular, this means 167if you have a file system on a two-disk mirrored 168.Nm 169and one of the disks fail, you cannot mount and use the remaining 170partition as itself; you have to configure it as a one-disk 171.Nm . 172You cannot replace a disk in a mirrored 173.Nm 174partition without first backing up the partition, then replacing the disk, 175then restoring the partition. 176.Ss Linux Compatibility 177The 178.Tn Linux 179compatibility mode does not try to read the label that 180.Tn Linux Ns ' 181.Xr md 4 182driver leaves on the raw devices. 183You will have to give the order 184of devices and the interleave factor on your own. 185When in 186.Tn Linux 187compatibility mode, 188.Nm 189will convert the interleave factor from 190.Tn Linux 191terminology. 192That means you give the same interleave factor that you 193gave as chunk size in 194.Tn Linux . 195.Pp 196If you have a 197.Tn Linux 198.Xr md 4 199device in 200.Dq legacy 201mode, do not use the 202.Dv CCDF_LINUX 203flag in 204.Xr ccdconfig 8 . 205Use the 206.Dv CCDF_NO_OFFSET 207flag instead. 208In that case you have to convert 209the interleave factor on your own, usually it is 210.Tn Linux Ns ' 211chunk size multiplied by two. 212.Pp 213Using a 214.Tn Linux 215RAID this way is potentially dangerous and can destroy 216the data in there. 217Since 218.Fx 219does not read the label used by 220.Tn Linux , 221changes in 222.Tn Linux 223might invalidate the compatibility layer. 224.Pp 225However, using this is reasonably safe if you test the compatibility 226before mounting a RAID read-write for the first time. 227Just using 228.Xr ccdconfig 8 229without mounting does not write anything to the 230.Tn Linux 231RAID. 232Then you do a 233.Nm fsck.ext2fs Pq Pa ports/sysutils/e2fsprogs 234on the 235.Nm 236device using the 237.Fl n 238flag. 239You can mount the file system read-only to check files in there. 240If all this works, it is unlikely that there is a problem with 241.Nm . 242Keep in mind that even when the 243.Tn Linux 244compatibility mode in 245.Nm 246is working correctly, bugs in 247.Fx Ap s 248.Nm ex2fs 249implementation would still destroy 250your data. 251.Sh WARNINGS 252If just one (or more) of the disks in a 253.Nm 254fails, the entire 255file system will be lost unless you are mirroring the disks. 256.Pp 257If one of the disks in a mirror is lost, you should still 258be able to back up your data. 259If a write error occurs, however, data 260read from that sector may be non-deterministic. 261It may return the data 262prior to the write or it may return the data that was written. 263When a 264write error occurs, you should recover and regenerate the data as soon 265as possible. 266.Pp 267Changing the interleave or other parameters for a 268.Nm 269disk usually destroys whatever data previously existed on that disk. 270.Sh FILES 271.Bl -tag -width ".Pa /dev/ccd*" 272.It Pa /dev/ccd* 273.Nm 274device special files 275.El 276.Sh SEE ALSO 277.Xr dd 1 , 278.Xr ccdconfig 8 , 279.Xr config 8 , 280.Xr disklabel 8 , 281.Xr fsck 8 , 282.Xr gvinum 8 , 283.Xr mount 8 , 284.Xr newfs 8 285.Sh HISTORY 286The concatenated disk driver was originally written at the University of 287Utah. 288