*** UNIX MANUAL PAGE BROWSER ***

WORM(8) System Manager's Manual WORM(8) NAME worm, jukebox - optical disk utilities SYNOPSIS worm mkfs [ -fdevice ] [ -ccomments ] [ -bblksz ] [ -nnblks ] [ -vnewvol_id ] vol_id worm stat [ -fdevice ] [ -Fn ] [ -v ] [ vol_id ] worm ls [ -fdevice ] [ -l ] [ file ... ] worm rm [ -fdevice ] vol_id [ file ... ] worm mv [ -fdevice ] vol_id src dest worm write [ -fdevice ] vol_id [ file ... ] worm read [ -fdevice ] [ -dm ] vol_id [ file ... ] worm cat [ -fdevice ] vol_id file worm copy [ -v ] [ -mmin_free ] [ -fsrc_dev ] src_vol_id dest_dev dest_vol_id worm offline [ -fdevice ] worm btree [ -fdevice ] vol_id worm dir [ -fdevice ] vol_id worm tmpdir [ -fdevice ] vol_id worm mount [ -wsecs ] [ vol_id ] jukebox [ -aemprsuU ] [ -wsecs ] [ vol_id ] DESCRIPTION The worm programs manipulate arbitrary files. They are intended for use with the raw device associated with a Write-Once Read-Many (WORM) optical disk. The default device is Other devices are specified by -fdevice and a device name of a single digit n is taken as an abbrevia- tion for Most of the commands implement a simple file system. Programs just wanting a raw device should still use worm mkfs so that the disk is properly labeled. The vol_id, or label, should be unique and by convention, the vol_id's for the A and B sides of a disk should be the same string suffixed by a lowercase a and b respectively. Worm mkfs labels an optical disk. The comments field is limited to 256 chars. It is purely descriptive and is printed by worm stat -v. The (default) blocksize is 1024 for our SONY disks. The number of blocks on a disk can be found by ra(4) or scsish(8); the default size (1,600,000 for single density, 3,250,000 for double density) sets aside 30MB or so as a hedge against oversights. If the disk has already been initialised, its vol_id must match vol_id. A new vol_id can be set with -v. Worm stat prints out labeling information including the amount of free space left on the disk. Option vol_id turns off all output except exit status: zero if vol_id matches that of the disk, one otherwise. Option -F similarly exits with status zero if the disk has more than n free blocks, otherwise three. Option -v produces more output. Worm ls simulates an emasculated ls(1). Worm rm makes the specifed files unavailable to the rest of the worm commands. Worm mv renames src to dest. Worm write copies files onto the WORM. If no file arguments are given, filenames are read one per line from standard input. The total number of files and bytes is printed on standard output. Worm read restores files from the WORM. If no file arguments are given, filenames are read one per line from standard input. Option -d causes directories to be created as needed. Option -m restores the original modification times. Worm cat copies the named file from the WORM to the standard output. Worm copy copies files directly from one disk to another. The names of the files to be copied are taken from standard input; groups (separated by blank lines) will be kept together. The names are typically gener- ated by worm ls. The -v option prints out progress and summary infor- mation. The copy will terminate before copying a group that would leave the destination volume with less than minfree (deafult value is 40000) blocks free. Worm offline makes the WORM go offline, ready for ejecting. This com- mand is harmless; accessing an offline drive will cause it to spin up and go online without operator intervention. Worm offline only takes effect after the last close of the WORM and as a bonus, applies to any MSCP device such as an RA81. Worm tmpdir saves a copy of the directory in /usr/worm/tmp/vol_id if the directory exists. This will speed up subsequent access substan- tially, although it will still be slower than worm btree below. On the other hand, worm tmpdir typically takes 5 minutes to run (on a VAX 11/750) whereas worm btree takes about 45 minutes. Worm btree constructs a new directory for the whole disk (in the form of a cbt(1) database). The new superblock is at zero. All the worm commands go faster with such an index but it is intended to be done just once, after the disk is complete. The directory occupies of the order of 10MB but may be more. If you really have to add more files to the disk, you need to write zeros on the first 1K block of the WORM be- fore using worm write. Worm dir takes the btree directory from the disk and stores in Future uses of the disk will be much faster. Worm mount returns the device on which the disk labelled vol_id is mounted. If the drive(s) are busy and you have a jukebox, the -ws op- tion tells how many seconds to wait before failing. The default is wait forever. If no vol_id is given, print the drive status. Jukebox manages the disks in the SONY jukebox. There are several op- tions (default is -s): -a Allocate a blank disk and label it vol_id. Use worm mkfs to change any fields from their default value. -e Eject the disk labeled vol_id. To physically retrieve the disk, press the OUT button (the OUT READY light should be on). Repeat until the IN READY light goes on. -m Mount the disk labelled vol_id in some drive and print the drive number on standard output. -p Print the list of disks in the jukebox. -r Rebuild the list of disks by examining each disk in the juke- box. Do not do this unless you are sure you need to. If vol_id is given, it should be one of the following letters and governs how disks are assigned shelf numbers. The de- fault is to leave the shelf number unchanged. Other options (mainly useful for demos) are c (compresses the disks in the jukebox towards the bottom or lower numbered shelves), r (distributes the disks randomly), and s (sorts the disks by vol_id). -s Print the status of the jukebox. -u Unload offline disks back onto their shelves. -U Unload all disks (offline or not) back onto their shelves. -wsecs This option only affects the behavior of -m. If all drives are busy, try again for secs seconds before failing. To load a disk into the jukebox, press the IN button on the jukebox when the IN READY light is on. After the shutter opens, push the disk in firmly. The disk (blank or initialised) is not examined immediately but on demand. Etiquette Vol_ids should be unique as discussed above. The file contains known vol_ids. The commands for reading and writing require vol_id's to guard against accessing the wrong disk. The recommended protocol for changing disks is if no one appears to be using the drive (by using ps(1)), execute worm offline and go to the drive. If, and only if, the drive has the DRIVE OFF (middle) light on, hit the EJECT button and change disks. If the light is not on, then someone is still using the disk and you should wait until they are done before hitting EJECT. Programming considerations Programs should not depend on writing any block more than once; how- ever, our SONY optical disks implement a small number of multiple writes via bad block replacement. A read(2) of an unwritten block re- turns with an errno of ENXIO. On Vaxes, the WORM is an MSCP device; thus geometry information can be fetched as in ra(4). For maximum speed, read and write in large blocks (preferably 63K) and avoid seeks. A seek across the whole disk takes about 1 second. The device is simply an appropriate raw ra(4) device, partition 7 (the whole disk). EXAMPLES worm mkfs -c"512x512x24 movies" tdmoviesa worm write tdmoviesa < filenames worm read -d tdmoviesa bumblebee/act2/frame1 FILES SEE ALSO backup(8), scsish(8), backup(1) BUGS The output of worm ls is not necessarily sorted. wild WORM(8)