1.200 Some info about tape backups 1.201 How do I do remote backup? 1.202 How do I backup a multi-disk volume group? 1.203 How do I put multiple backups on a single 8mm tape? 1.204 How can I make an exact duplicate of a tape over the network? 1.205 What is tape block size of 0? 1.206 Resetting a hung tape drive... 1.207 How do I read a mksysb tape with tar?
1.200: Some info about tape backups
From: Craig Anderson The following supplements the information on rmt devices in InfoExplorer. It is based on my own personal experience with IBM tape drives running on AIX 3.1. No warranty is expressed or implied. CONFIGURING THROUGH SMIT: BLOCK size (0=variable length) (ALL) Sets the tape block size. When reading, the block size must be set to the block size set when the tape was written. When using some commands, tapes written with ANY block size can be read if the block size is set to 0 (variable length) (see "BLOCK SIZES" below). Use DEVICE BUFFERS during writes (ALL) Set to yes, the device will buffer data internally on writes. This greatly improves performance, but under certain cases may be undesirable since the data is not written to tape before returning a good indication. Use EXTENDED file marks (8mm only) Extended file marks take up much more space than short (or non-extended) file marks. But extended file marks can be overwritten, allowing data not at the beginning of tape to be overwritten (see "FILE MARKS" below). RETENSION on tape change or reset (1/4" only) If set to "no" then the tape will not be retentioned automatically when the tape is inserted. Note that this will take effect only after the device is used. FILE MARKS: Tape devices support multiple tape files. Tape files are the result of a backup/cpio/tar/dd type command, where the device is opened, written to, and closed. Because tapes allow large quantities of data to be written on a single tape, several backups (that is, tape files), may be combined on one physical tape. Between each tape file is a "tape file mark" or simply "file mark". These file marks are used by the device driver to indicate where one tape file ends and another begins. B E <------- O O -------> T T __ ____________________________ _______________ physical | \ | | \ |physical beginning| \ | tape | \ | end of | \ | file | \ | of tape | \ | mark | \ | tape |_____\________|_______|__________\_________| Note that there is a distinction between the beginning of tape (BOT) side of a file mark and the end of tape (EOT) side of a file mark. If the head is on the BOT side of a file- mark, "tctl fsf 1" command will move only to the EOT side of the same file mark. With the 1/4" tape drive, writing can only take place sequentially, or after blank tape has been detected. You cannot write over data on the tape (except at BOT). If you wish to add data to a tape which has been written and then rewound you should space forward file mark until an error occurs. Only then can you start writing again. With an 8mm tape drive, writing can only take place before blank tape, an EXTENDED file mark, or at BOT. Thus if several backups have been made on one tape and you wish to overwrite one of the backups, position the tape to the place you wish to start writing and issue the following commands: tctl bsf 1 tctl eof 1 The first command skips back to the BOT side of the same file mark. The second command rewrites the file mark (writing is allowed before extended file marks). The erase head will erase data ahead of the write head, so that after writing the file mark the head will be positioned before blank tape. Only after this may you start writing over data in the middle of the tape. (All data beyond where you are currently writing will be lost). Note that you cannot write over short file marks. In order for this to work, the tape must have been written with extended file marks (use smit to change this). With the 9-track drive writing can take place anywhere on the tape although overwriting single blocks of data is not supported. On the 8mm drive extended filemarks use 2.2 megabytes of tape and can take up to 8.5 seconds to write. Short filemarks use 184K and take up to 1.5 seconds to write. BLOCK SIZES: When data is written to tape it is written in blocks. The blocks on a tape are separated by inter-record gaps. It is important to understand the structure of the written tape in order to understand the problems which can occur with changing block sizes. In fixed block size mode all blocks on the tape are the same size. They are the size of the block size set in the device configuration. All read()s and write()s to the tape drive must be a multiple of the fixed block size. In fixed block mode a read() will return as many blocks as needed to satisfy the read() request. If a file mark is encountered while reading the tape only the data up until the file mark will be returned. It is not possible for the tape drive to read a tape whose block size is not the same as the block size in the device configuration. (Unless the device configuration is in variable size blocks.) In variable block size (0) mode, the blocks written on the tape are the size of the read() and write() requests to the device driver. In this case, the actual block sizes on the tape can be changed using the options to the backup commands (tar -C, cpio -C, backup -C). In variable mode, read() requests greater than size of the block on the tape will return only the data from the next block on the tape. It is this feature that allows tapes written in any block size (fixed or variable) to read with the dd command (the output from the dd command may be piped to restore, tar, or cpio for example.) Note that backup, tar, and cpio cannot read all tapes by using a large block size because they assume there is an error if they get a short read(). dd ibs=128k obs=16k if=/dev/rmt0 | ... The tape head is always positioned at an inter-record gap, file mark, or blank tape after reading or writing. With the 8mm tape drive, using a fixed block size which is not a multiple of 1K is inefficient. The 8mm tape drive always writes internally in 1K blocks. It simulates the effect of variable block sizes, but, for example, using a fixed block size of 512 bytes (or using variable block size and write()ing 512 bytes at a time) wastes one half of the tape capacity and gives only one half the maximum transfer rate. To figure out a tape's actual block size try: 1). Set the tape to variable block size. 2). "dd if=<tape> of=/tmp/dummy bs=128k count=1" 3). "ls -l /tmp/dummy" 4). The number of bytes in "/tmp/dummy" is the physical block size. EXCHANGING DATA WITH NON-UNIX AND OTHER VENDORS MACHINES: Many tape drives support both variable and fixed block sizes. Variable block mode writes block sizes the size of the write command issued (tar and backup specify this with the -b option). In fixed mode, block sizes are fixed and all writes must be a multiple of the fixed block size. Unix often internally chops larger reads and writes up into manageable pieces (often 65535, 65534, or 65532 bytes) before doing the actual reads and writes. This means reads and writes of 64K bytes are often broken up into a 65535 byte record and a 1 byte record (In fixed mode the write will fail). Block sizes >= 64K (-C128 and greater) should be avoided for this reason. AIX does not break up read and write requests, but be aware of the situation on other machines. If the tape is written in an unknown block size then set the device configuration in smit to use variable size blocks, use the "dd" command with a large input block size, and pipe it to the restore command. For example: chdev -l rmt0 -a block_size=0 dd if=/dev/rmt0 ibs=128k obs=16k | tar -tvf-
1.201: How do I do remote backup?
There seems to be several ways of doing this. I found the following works: tar -b1 -cf - . | rsh REMOTEHOST "dd ibs=512 obs=1024 of=/dev/TAPEDEVICE" From: kraemerf@franvm3.VNET.IBM.COM (Frank Kraemer) mksysb will not back up to remote tape devices. The following script will do remote backups. [Ed.: I've verified this script works fine. However, it may be slow for large filesystems since it creates a temp file of filenames in /tmp.]#!/bin/ksh # @(#) Create a backup tape of the private user data. #=================================================================# # Script : usave.sh # # Author : F. Kraemer # # Date : 92/02/19 # # Update : 92/10/29 # # Info : the ultimative backup script # # Example: usave.sh /dev/rmt0 - save to local tape # # usave.sh /save/save.me - save to local file # # usave.sh /tmp/pipe - save to remote tape # #-----------------------------------------------------------------# PS4="(+) " #set -x PROG=$(basename $0) HOST=$(hostname) TODAY=$(date +%H:%M:%S) # # cleanup # cleanup () { ec=$1 error=$2 case "$ec" in "$USAGE_EC") # usage error error="Usage:\t$PROG DeviceName\n" 1>&2 ;; "$NOTAP_EC") # Tape error error="error:\t$PROG: $DEVICE is not available on the system.\n" 1>&2 ;; "$LISTE_EC") # list error error="error:\t$PROG: could not create tar list for $LOGNAME.\n" 1>&2 ;; "$NOTAR_EC") # tar command error error="error:\t$PROG: tar command failed.\n" 1>&2 ;; "$PIPEP_EC") # pipe error error="error:\t$PROG: mknod command failed.\n" 1>&2 ;; "$NORSH_EC") # rsh error error="error:\t$PROG: rsh - Remote Shell command failed.\n" 1>&2 ;; "$RHOST_EC") # remote host error error="error:\t$PROG: Remote Host unknown.\n" 1>&2 ;; *) ;; esac case "$DEVICE" in # # Fix the block size if $DEVICE is a tape device # /dev/rmt[0-9]*) echo "\n\t$PROG: Rewinding tape to begin.........(please wait)\n" tctl -f $DEVICE rewind 2>/dev/null ;; *) ;; esac rm -f ${LIST} ${PIPE} 2>/dev/null [ -n "$error" ] && echo "\n${error}\n" trap '' 0 1 2 15 exit "$ec" } # # Variables # USAGE_EC=1 # exit code for usage error NOMNT_EC=2 # exit code wrong device name NOTAP_EC=3 # exit code no tape available LISTE_EC=4 # exit code backup list error NOTAR_EC=5 # exit code for wrong tar TRAPP_EC=6 # exit code for trap PIPEP_EC=7 # exit code for pipe RHOST_EC=8 # exit code for bad ping NORSH_EC=9 # exit code for bad rsh DEVICE="$1" # device to tar into LIST="/tmp/.tar.$LOGNAME.$$" # REMOTEH="" # Remote host for backup REMOTET="" # Remote tape for backup tapedev= # PIPE="/tmp/pipe" # Pipe for remote backup # # main() # tput clear echo "\n\t$PROG started from $LOGNAME@$HOST on $TERM at $TODAY.\n" rm -f $LIST 2>/dev/null # # Trap on exit/interrupt/break to clean up # trap "cleanup $TRAPP_EC \"Abnormal program termination. $PROG"\" 0 1 2 15 # # Check command options # [ "$#" -ne 1 ] && cleanup "$USAGE_EC" "" # # Check device name # [ `expr "$DEVICE" : "[/]"` -eq 0 ] && cleanup "$NOMNT_EC" \ "$PROG: Backup device or file name must start with a '/'." # # Check tape device # case "$DEVICE" in # # Fix the block size if $DEVICE is a tape device # /dev/rmt[0-9]*) # echo "\n\t$PROG: Verify backup media ($DEVICE)............\n" # # see if a low or high density tape device was specified # (eg rmt0.1) density="`expr $DEVICE : \ "/dev/rmt[0-9]*\.\([0-9]*\)"`" # # strip /dev/ from device name and # get the base name (eg translate: # /dev/rmt0.2 to rmt0) # tapedev="`expr $DEVICE : \ "/dev/\(rmt[0-9]*\)[\.]*[0-9]*"`" # # Check if the tape is defined in the system. lsdev -C -c tape -S Available -F "name" | grep $tapedev >/dev/null 2>&1 rc=$? [ "$rc" -ne 0 ] && cleanup "$NOTAP_EC" "" # # Restore old tape name. # [ "${density:-1}" -lt 4 ] && density=1 || density=5 DEVICE="/dev/${tapedev}.${density}" echo "\n\t$PROG: Insert a tape in ($DEVICE)........(press enter)\n" read TEMP echo "\n\t$PROG: Rewinding tape to begin...........(please wait)\n" tctl -f $DEVICE rewind 2>/dev/null ;; # # Backup is done on remote host. The remote shell facility # must be set up and running. # ${PIPE}*) # echo "\n\t$PROG: Assuming remote backup via network.\n" echo "\t$PROG: Enter name of Remote Host ===> \c" read REMOTEH echo "\n\t$PROG: Pinging Remote Host to test connection.\n" ping ${REMOTEH} 1 1 >/dev/null 2>&1 rc=$? # give up unknown host [ "$rc" -ne 0 ] && cleanup "$RHOST_EC" "" JUNK=$(rsh ${REMOTEH} "/usr/sbin/lsdev -C -c tape -S Available") rc=$? # give up rsh failed [ "$rc" -ne 0 ] && cleanup "$NORSH_EC" "" echo "\t$PROG: Available Tapes on ${REMOTEH} are :\n\n\t\t${JUNK}\n" echo "\t$PROG: Enter name of Remote Tape (e.g. /dev/rmt0) ===> \c" read REMOTET echo "\n\t$PROG: Insert tape on ${REMOTEH} in ${REMOTET}..(press enter)" read TEMP echo "\t$PROG: Rewinding Remote Tape ${REMOTET} on ${REMOTEH}.\n" rsh ${REMOTEH} "tctl -f ${REMOTET} rewind" rc=$? # give up rsh failed [ "$rc" -ne 0 ] && cleanup "$NOTAP_EC" "" rm -f ${PIPE} 2>/dev/null mknod ${PIPE} p rc=$? # give up mknod failed [ "$rc" -ne 0 ] && cleanup "$PIPEP_EC" "" cat ${DEVICE} | rsh ${REMOTEH} "dd of=${REMOTET} obs=100b 2>/dev/null" & ;; *) ;; esac # # Prepare the list # echo "\n\t$PROG: Create list of files to be saved...." find $HOME -print > $LIST rc=$? [ "$rc" -ne 0 ] && cleanup "$LISTE_EC" "" # # tar the files # echo "\n\t$PROG: Changing current directory to (/)...." cd / > /dev/null 2>&1 echo "\n\t$PROG: Running tar format backup from user ($LOGNAME)...." tar -cvf "$DEVICE" -L "$LIST" rc="$?" [ "$rc" -ne 0 ] && cleanup "$NOTAR_EC" "" # # Backup completed # TODAY=$(date +%H:%M:%S) echo "\n\t$PROG ended at $TODAY............................\n\n" cleanup 0
1.202: How do I backup a multi-disk volume group?
From: pack@acd.ucar.edu (Daniel Packman) [ Ed.: I have not verified this procedure. I would actually recommend NOT to have one volume group span multiple disks unless you really need such big logical volumes. ] 1. If you have a set of three or more disks in a volume group (typically 3 for 5xx machines with three internal drives; with only two, the procedures outlined here have to be modified to ignore the fact that you don't have a quorum in the volume group) 2. If one drive has failed (usually only one fails at a time :-) ) It is possible to go through a service boot (the volume group is called rootvg and one of the 2 good disks on it is called hdisk0): importvg -y rootvg hdisk0 varyonvg -f -n -m1 rootvg These commands will work, but give error messages. If you wish to mount a user filesystem, say /u on logical volume /dev/lv00, then mount -f /dev/lv00 /v will work only if jfslog, the journaled file system log device, is not on the damaged disk. If it is, you must (and can in any case) mount the filesystem read-only: mount -f -r /dev/lv00 /v This crucial and rather obvious point baffled several level 3 support personnel at Austin as well as myself for almost a week. Once the file system(s) of interest are available, they can be saved to tape for restoration later. Of course, one can expect only about two thirds of a filesystem to be recoverable if it spans all 3 physical disks. One other point to remember is that the standard boot procedure from floppy includes the restore command but does not include the backup command. ***************************************************************************** * If you do not have other RS6000 machines at your site it is imperative * * that you either build a bootable tape which includes either restore or * * tar or cpio (a bootable floppy set will not have enough space) or at the * * very least copy onto a spare floppy backup, cpio, or tar. The floppy * * should be created with backup -ivq so that its contents can be read into * * the memory resident system after booting. * ***************************************************************************** All is not lost if tar, cpio or backup are available on an undamaged disk that can be mounted. Since tar and cpio are in /bin, they may both very well be unavailable. It is a very good idea for those who have tape devices to build a bootable tape with their desired extra commands in it. Follow the instructions from IBM but add your desired commands to the following three files: /usr/lpp/bosinst/tape2 /usr/lpp/bosinst/diskette/boot2 /usr/lpp/bosinst/diskette/inslist If you have anything other than a minimum memory configuration, you should be able to add many commands.
1.203: How do I put multiple backups on a single 8mm tape?
From: kerm@mcnc.org (Cary E. Burnette) There are two possible solutions to this, both of which use /dev/rmt0.1 which is non-rewinding. SOLUTION #1 ----------- To put multiple backups on a single tape, use /dev/rmt0.1, which is a no-rewind device, using either rdump or backup (both by name & inode work). Using rdump or backup "byinode" both generate the message that the tape is rewinding but actually do not. This is an example that works on my system: # rsh remote1 -l root /etc/rdump host:/dev/rmt0.1 -Level -u /u # rsh remote2 -l root /etc/rdump host:/dev/rmt0.1 -Level -u /u # tctl -f /dev/rmt0.1 rewind # rewinds the tape where I am implementing the command from host. To restore a table of contents of the first I would use # restore -f /dev/rmt0.1 -s1 -tv where the -s1 flag tells restore to go to the first record on the tape. Type the exact command again to get the second record. The -s(Number) means go to Number record from this spot. It works pretty well. SOLUTION #2 ----------- Steve Knodle Educational Resources Center Clarkson University I use:------------------- Dump.sh -------------------- CONTENTSFILE=`date |dd conv=lcase |sed -e 's/19//' |awk '{print $6 $2 $3}'` set -x LEVEL=$1 shift backup -c -b 56 -$LEVEL -uf /dev/rmt0.1 / backup -c -b 56 -$LEVEL -uf /dev/rmt0.1 /usr backup -c -b 56 -$LEVEL -uf /dev/rmt0.1 /u tctl -f /dev/rmt0 rewind touch /usr/local/dumps/Contents.$CONTENTSFILE echo "Dumping /" >>/usr/local/dumps/Contents.$CONTENTSFILE restore -t -s 1 -f /dev/rmt0.1 >>/usr/local/dumps/Contents.$CONTENTSFILE echo "Dumping /usr" >>/usr/local/dumps/Contents.$CONTENTSFILE restore -t -q -s 1 -f /dev/rmt0.1 >>/usr/local/dumps/Contents.$CONTENTSFILE echo "Dumping /u" >>/usr/local/dumps/Contents.$CONTENTSFILE restore -t -q -s 1 -f /dev/rmt0.1 >>/usr/local/dumps/Contents.$CONTENTSFILE tctl -f /dev/rmt0 rewindI process the table-of-contents first by a little program that does common prefix encoding, and then compress. This gives a table of contents file I can keep on-line until the tape is reused.
1.204: How can I make an exact duplicate of a tape over the network?
The challenge here is not to have to create a temporary file (disk space limitation) and work across heterogeneous networks. This script might work: LOCAL=/dev/tape_dev REMOTE=/dev/tape_dev dd if=$LOCAL ibs=64k obs=512 | rsh remote_host dd ibs=512 obs=64k of=$REMOTE From: pack@acd.ucar.edu (Daniel Packman) Daniel provides the following perl script to convert from the known world's function codes to AIX for compatibility.#!/bin/perl # Wrapper to convert input rmt requests to # AIX 3.2 ioctl numbers. We pass on all commands we don't understand # I0 MTWEOF -> I10 STWEOF write and end-of-file record # I1 MTFSF -> I11 STFSF forward space file # I2 MTBSF -> I12 STRSF reverse space file # I3 MTFSR -> I13 STFSR forward space record # I4 MTBSR -> I14 STRSR reverse space record # I5 MTREW -> I6 STREW rewind # I6 MTOFFL -> I5 STOFFL rewind and unload tape # I7 MTNOP -> I0 (no-op? should ignore following count) # I8 MTRETEN-> I8 STRETEN retension tape, leave at load point # I9 MTERASE-> I7 STERASE erase tape, leave at load point #I10 MTEOM (position to end of media ... no ibm equivalent?) #I11 MTNBSF (backward space file to BOF ... no ibm equivalent?) @iocs = (10,11,12,13,14,6,5,0,8,7); open(RMT,"|/usr/sbin/rmt") || die "Can't open pipe to rmt\n"; select(RMT); $| = 1; while (<STDIN>) { s/(^I)(\d$)/I$iocs[$2]/; exit 0 if $_ =~ /^[Qq]/; print RMT $_ ; } exit 0;
1.205: What is tape block size of 0?
From: benson@odi.com (Benson I. Margulies) Tape devices are generally split into two categories: fixed block and variable block. 1/4" tape is the fixed block, and 8mm is variable. On a fixed block size device, the kernel always sends data to the device in suitable block size lumps, and varying the size passed to write(2) (e.g., via the bs option to dd) gives the kernel more data to stream. On a variable block size device, the kernel writes to the device whatever passed to it. On an 8mm, it had better be a multiple of 1024 to get efficient tape usage. AIX has the World's Only Variable Block Size 1/4" tape drive. If you use SMIT to set the block size to a nonzero value, AIX treats the device as fixed block size, whether it is or not. By default, 8mm drives are set to the same size as 1/4", 512 bytes. This is wasteful, but otherwise mksysb and installp would fail. If you set the block size to 0, the device is treated as variable block size, and the size passed to write becomes the physical block size. Then if you use a sensible block size to dd, all should be wonderful.