Search the FAQ Archives

3 - A - B - C - D - E - F - G - H - I - J - K - L - M
N - O - P - Q - R - S - T - U - V - W - X - Y - Z - Internet FAQ Archives

comp.unix.sco Technical FAQ (1/7)

( Part1 - Part2 - Part3 - Part4 - Part5 - Part6 - Part7 )
[ Usenet FAQs | Web FAQs | Documents | RFC Index | Sex offenders ]
Archive-Name: comp.unix.sco Technical FAQ 1/7
Posting-Frequency: Monthly (mid month)
Last-modified: Oct 12

comp.unix.sco Technical FAQ 1/7

See reader questions & answers on this topic! - Help others by sharing your knowledge
   Questions and Answers Common to Unix, Xenix and ODT
   FAQ Starting Page
   These FAQS were developed and maintained for years by (Stephen M. Dunn). Steve no longer has the time to
   maintain them, and has asked me to take them over. Please remember the
   debt all of us owe to Steve for his efforts- I myself spent many hours
   learning from these very documents, and I'm sure many of us can say
   similar things.
   Because Steve has not been able to maintain these for a while now,
   some of the information herein is outdated. I am working to correct
   that, but it's a lot to catch up on, so if you spot something, please
   let me know. For the moment, I'm just marking some of it as probably
   being useless; as I have time, I'll check further to be certain before
   I remove anything.
   Suggestion: Use my to find what you
   are looking for.
   Are there any screen savers?

   Unix (and Xenix 2.3.4) have a built-in screen saver for VGA only. You
   have to reconfigure the kernel for this to work. It doesn't work with
   all hardware, but try it first. Also, it has been reported that VP/ix
   may not be compatible with this screen saver. To enable the
   screensaver, set the kernel variable TBLNK to the number of seconds of
   inactivity which should trigger the screensaver, relink, and reboot.
   Unixware users may use "vtblank" to dynamically set and adjust the
   screen saver time
   Roberto Zini:
   If you're operating under X-windows you can use the PD program
   xscreensaver; it comes with dozens of very nice screen savers you can
   install on your system. You can find the 'xscreensaver' program on the
   Skunkware CD-ROM; alternatively you can use the xlockmore program,
   also available on the same CD-ROM.
   [Table of Contents]
   Is tar/cpio a good backup program?

   tar is not; cpio is, to some degree. tar will not back up things like
   device nodes (and, prior to OpenServer Release 5, it will also not
   back up empty directories), so a tar backup will not catch anything in
   /dev, for example, and you will find your device nodes missing when
   you do your restore. cpio will catch these things.
   Neither one is very good at verification. You can dd the tape to make
   sure you can read the whole thing, and run it through tar or cpio ...
   but they'll just check the file headers to make sure they make some
   sense. If you need better verification, try one of the products listed
   below. Most third-party backup programs do many things better than the
   standard utilities included with the OS, including things like making
   much better emergency recovery diskettes, byte-for-byte verification
   (if you want), compression, more options for things like
   nondestructive restore, etc. Many of us swear by them.
   gnu tar is a significantly better backup utility, and is available on
   many archive sites listed in the Administrative FAQ. There is also a
   shareware tar/cpio archive checker called tapechk, written by Nigel
   Horne <>. A demonstration version is available
   Commercial programs provide better solutions. The following vendors
   offer backup programs for SCO, Linux and many other platforms:
     * CTAR
     * Lone Tar
     * Microlite Edge
   Also see
   [Table of Contents]
   How do I compress my backups?

   Well, you could just run the output of tar, cpio, or whatever through
   compress, but if even one bit of your tape or diskette goes bad,
   you'll lose the rest of the backup. Not recommended at all, unless of
   course you don't actually care if your backups work - but if you
   didn't care, you wouldn't be doing any, right?
   A better solution would be a third-party product. The next answer
   lists a few ... if you produce, market, or use one that's not listed
   below but which you feel should be, please send me the information.
   [Table of Contents]
   What are some third-party backup/recovery products?

   There are a couple of categories here - products which are mostly
   aimed at one or a small number of Unix machines, and those which are
   aimed at enterprise-wide, multiplatform backup. The following two
   lists are NOT meant to be all-inclusive, but merely a sample of some
   of the better-known products.
   First, the ones aimed at one or a few Unix machines:
     * BackupEDGE (Microlite Corp., 2315 Mill Street, Aliquippa, PA
       15001-2228;; (888) BKP-EDGE or (724) 375-6711;
     * BRU (Enhanced Software Technologies Inc., 5016 S. Ash Avenue Suite
       109, Tempe, AZ 85282;; (800) 998-8649 or (602)
     * Lone-Tar (Lone Star Software Inc., 13987 W. Annapolis Court, Mt.
       Airy, MD 21771;; (301) 829-1622 or (800)
     * Ctar (Unitrends Software Corp., 1601 Oak Street, Suite 201, Myrtle
       Beach, SC 29677;; (800) 648-2827 or (803)
   These products tend to be fast and robust, generally offer data
   compression, and tend to be able to handle errors on the backup media.
   Many also include, or can optionally be purchased with, utilities to
   create automated emergency recovery diskettes which are much
   friendlier and easier to use than the ones you can produce with
   standard SCO utilities.
   Now, a few for those with more ambitious backup plans ... this section
   is under construction and hopefully I'll have some more contact info
     * The Backup Professional (Lone Star Software Inc., 13987 W.
       Annapolis Court, Mt. Airy, MD 21771;; (301)
       829-1622 or (800) LONE-TAR;
     * ARCserve/Open (The Santa Cruz Operation Inc., 400 Encinal Street,
       Santa Cruz, CA 95061;; (800) SCO-UNIX or (408)
     * Legato (415) 812-6000
   A variety of backup products was reviewed in the September 1997 issue
   of SCO World Magazine.
   [Table of Contents]
   I don't like being restricted to 14 character filenames

   If you're running Xenix, or a version of Unix prior to 3.2v4, I'm
   afraid you're stuck. Unix 3.2v4, however, includes long filename
   support on all EAFS filesystems. OSR5 adds two new filesystems, HTFS
   and DTFS, which also support long filenames. Unixware 7 also supports
   long file names. More information on long filenames can be found in
   the section dealing with Unix.
   [Table of Contents]
   How do I get a copy of adb?

   If you have the Development System, you already have /bin/adb. If not,
   you may need to grab a copy from your distribution, or it may already
   have been installed, depending on your OS and version. It could be
   called /bin/adb (older Xenix) or /etc/_fst (Unix and recent versions
   of Xenix). If you don't have either of these, look through the files
   in /etc/perms for them; in Xenix 2.3.4, you will find one of each,
   which will be in fact the exact same file but on two different
   diskettes. If the volume on which the file you want is mountable (you
   can check this in the manual, or use the dtype command), then mount it
   and copy the file off. Otherwise, use tar to extract the file, keeping
   in mind that the filenames on your diskettes are all written with
   relative paths (i.e. ./bin/adb, not /bin/adb). Note that if you look
   in the Unix documentation, it may well tell you that you need
   /bin/adb, when in fact it's called /etc/_fst.
   [Table of Contents]
   I can't find crypt

   Most (all?) of SCO's release notes state that due to American
   government restrictions aimed at trying to prevent unfriendly nations
   from having access to data encryption technology, SCO does not ship
   crypt with their products. If you live in the States and would like
   crypt(C) and the crypt(S) libraries, contact SCO support. This is also
   worth trying in Canada, as the particular regulation in question
   permits export of such technology to Canada; however, I don't know if
   SCO will honour such requests. There is also an international version
   of crypt available from the usual places as lng225b.
   Recently, some of the cryptographic restrictions have changed- for the
   better, we hope, though plenty of stupidity still remains.
   [Table of Contents]
   What do I need to compile programs?

   If you have free OpenServer, you already have a license to install the
   development system; the Web page on which you license free OpenServer
   gave you several keys and codes, including one to license the
   development system.
   Xenix, Unix and ODT do not ship with program development tools. These
   are unbundled into packages known as Development Systems. The
   rationale behind this is that many users of SCO systems are using
   off-the-shelf software and never need to write a line of C code. If
   everyone was forced to buy the development system whether they needed
   it or not, some of the customers might get upset. There is a periodic
   flame war about this; this is not the place to discuss it.
   You can buy the Development System for any of the three environments
   listed above as a separate package including the compiler, header
   files, libraries, lex, yacc, linker, and other tools. Additionally,
   development systems are available for other packages such as TCP/IP;
   these development systems add the include files, libraries, etc.
   required to program for the package in question. The ODT Dev Sys
   includes the development tools for all of the other goodies (e.g.
   TCP/IP, X) that are bundled into ODT. Since OSR5 generally bundles the
   various runtime packages (e.g. TCP/IP) with the OS, it also bundles
   the same development packages, so there are not the same development
   system add-ons in OSR5 that there were in previous versions.
   The "lxrun" package allows you to run many Linux programs on OSR5 and
   Unixware; you could always compile on a Linux machine.
   There are versions of gcc (the Gnu C Compiler) freely available for
   SCO systems. On older SCO operating systems, however, you will
   probably need the development system, as the header and library files
   you need are part of it and not part of the operating system itself.
   This problem has been alleviated in OpenServer Release 5, as the
   headers and libraries are now shipped as part of the base operating
   system and are available even if you have not purchased the
   development system.
   gcc sources and binaries for OpenServer Release 5 only are on the free
   Skunkware family of CD-ROMs; for more info, see or read the section below
   entitled "What is Skunkware?"
   gcc sources and binaries are also available on Robert Lipe's home
   page: or These are mirrored by SCO at and
   You can also look at a different version at and
   For those who want to find this based on a keyword search: programming
   programmer library libraries developer source
   [Table of Contents]
   What does the NCALL kernel parameter affect?

   NCALL controls the size of the kernel callout table. The kernel has
   the ability to schedule some action at a given real time; this is
   often used by device drivers and by the nap(S) system call. The size
   of this table is set by NCALL. If the system message "timeout table
   overflow" appears on your console, NCALL should be increased.
   Increasing NCALL is not expensive in terms of memory or CPU overhead,
   as the structure is small (16 bytes per entry) and stored sorted, so
   it is best to be generous with these entries.
   [Table of Contents]
   How do I reset the root password if I forget it? (part 1)

   This procedure will work for Xenix, and for Unix as well if you are
   using a very relaxed security level (one which stores encrypted
   passwords directly in /etc/passwd). If you're using a higher security
   level on Unix, look for part 2 below.
   Boot the system from your emergency boot diskettes (if you didn't make
   these and keep them up to date, shame on you, but you should be able
   to use N1/N2 instead, and see the entry on crashing out of these
   diskettes below). Next,

   mount /dev/hd0root /mnt

   ; this will mount your hard drive's root filesystem on /mnt. Edit
   /mnt/etc/passwd. The first line will be your root line, such as
   Edit out the encrypted password (don't touch anything else!) so that
   the line reads something like
   Save the file and shut down. Reboot from the hard drive. Your root
   password has now been removed, and you can reset it normally.
   Also see
   [Table of Contents]
   How do I reset the root password if I forget it? (part 2)

   This is another procedure involving manually editing files, and is
   specific to SCO Unix 3.2v4.0 through 3.2v4.2. The location of the
   encrypted passwords depends on the security settings. Look in
   /etc/passwd, /etc/shadow, and /tcb/files/auth/r/root; one or more of
   these will be used depending on how you have security configured.
   Follow the procedure in part 1 above; instead of editing /etc/passwd,
   edit the appropriate file(s) from the above list, and delete the
   encrypted password field. Note that formatting is critical; while you
   can delete the contents of the field, you must not remove separators,
   and making seemingly minor errors such as leaving blank lines can
   cause problems. Save, shut down, and reboot. C2 security will complain
   about what you've done; to make it happy, run /etc/fixmog. You may
   also want to run /tcb/bin/integrity and /etc/tcbck.
   [Table of Contents]
   How do I reset the root password if I forget it? (part 3)

   This procedure will work for any variant of SCO Xenix or Unix. As
   above, boot from your emergency boot diskettes and

   mount /dev/hd0root /mnt

   to gain access to your hard drive's root filesystem. Now, run
   /mnt/bin/chroot /mnt "/mnt/bin/passwd root"

   (check the chroot man page for more info on how it works). As before,
   shut down and reboot. It has been reported that on 3.2v4.2 (and
   possibly others), this must be done in two steps:

   /mnt/bin/chroot /mnt "/bin/su root"

   If it doesn't work with the quotes, try it without.
   [Table of Contents]
   How do I crash out of the install script?

   On OpenServer Release 5, boot from the boot diskette, and at the Boot:
   prompt, type


   This is not an undocumented option to the boot command, but rather a
   special line in /etc/default/boot on the installation diskette - so
   you can't use it from anywhere but your installation boot diskette.
   And you can't use it anymore at all. Tom Melvin pointed out:
   The 'tools' option broke around 5.0.4 time - I know it does not work
   in 5.0.5 or 5.0.6 Don't have a 5.0.4 box around to test it with. 
   Tom's right: I booted "tools" on a 5.0.6 install disk. Part of the
   functionality of "tools" is still there (and in fact "tools" still is
   a boot option. If you press F8, you get the same screen that "tools"
   gave you directly, and you can do a shell escape. Unfortunately, so
   far I haven't been able to see how you can mount the existing hard
   drive. The old "hd0root" device is not there, and attempting "mount
   /dev/hd0a /mnt" didn't work.
   For older SCO Unix/Xenix/ODT releases, wait until the question early
   in the process that asks you what your keyboard type is. For
   character-mode installations, this is a regular textual prompt; for
   ODT, it's a box in a curses-style installation program. How to break
   out at this point depends on the OS. Under Xenix, press Del. Under
   Unix, type shell and press enter. Under ODT, press Control-A.
   Roberto Zini:
   See also
   How can I generate and save a debug logfile for an SCO OpenServer 5
   installation or upgrade (not strictly related but worth reading :-)
   [Table of Contents]
   Why can't fsck find my lost+found directory?

   Because you don't have one. It's possible someone deleted it, but the
   more likely cause is that you didn't use mkdev fs to create it.
   One of the things that fsck looks for is inodes which are marked as
   used (i.e. not in the free list) but do not have a directory entry
   pointing to them. fsck will ask if you wish to reconnect these; if you
   say yes, it will try to create a file entry in the /lost+found
   directory on that filesystem. If there is no free space in
   /lost+found, it is not safe to expand it because the rest of the
   filesystem may still be corrupt; for information on this one, see
   below. If there is not /lost+found directory, fsck will tell you that
   it can't reconnect the file and the data in that file will be lost.
   [Table of Contents]
   I want more space in my lost+found directory

   By default, the mkdev fs script creates 62 empty entries in
   lost+found. If you'd like to make more, use a variant of the following
   for a in 0 1 2 3 4 5 6 7 8 9
  for b in 0 1 2 3 4 5 6 7 8 9
    > /lost+found/dummy$a$b
   rm -f /lost+found/*

   The above will create 100 entries. Season to taste.
   [Table of Contents]
   How do I find out serial numbers of my various components?

   For the OS itself, you can use uname. For Unix, use uname -X; it will
   print (among other items) the serial number. For Xenix (at least
   2.3.3, and probably other releases), uname -u will print the numerical
   portion (e.g. if your serial number is sco012345, it will print
   Starting in Unix 3.2v4.2/ODT 3.0, SCO added /etc/getserno. To find out
   the serial number of a package, first find out what files are
   serialized in that package using grep ser= /etc/perms/* (or
   /etc/perms/packagename if you know it). Then, run /etc/getserno
   filename, where filename is the name of one of the files that is
   serialized during installation. Note that not all files listed may
   actually contain a textual representation of the serial number (for
   example, none of the binaries in the Unix dev sys do).
   As a special case, the serial number of the OS itself can be found
   simply by watching the kernel ID it displays at boot time (or look
   through /usr/adm/messages for it).
   Roberto Zini:
   If one's interested in finding out the original activation key issued
   during the OpenServer 5 installation (eg, widely used here before
   submitting a commercial upgrade order), one could retrieve it by using
   the following command:

   grep IQM_ACTIVATION_KEY /usr/adm/ISL/iqm_file

   Alternatively, try with the following:
   grep IQM_ACTIVATION_KEY /opt/softmgmt/profiles/standard/SCO/Unix/*/iqm_file

   where '*' is the actual OS5 version (eg, 5.0.5Eb)
   Under SCO Unix 3.2v4.2:
   grep IQM_ACTIVATION_KEY /install/iqm_file

   Under SCO UnixWare 7:
   grep ActKey /vad/adm/isl/ifile

   [Table of Contents]
   How do I solve an "arglist too long"?

   Wildcard expansion (globbing) is performed by the shell. There is a
   limit of 5120 bytes (5k) for the environment and command line
   arguments put together, in all versions of SCO Xenix and SCO Unix
   versions prior to OpenServer 5; more on OSR5 later. See also TA
   This is particularly likely to be a problem under X, as it has a habit
   of using a lot of environment space. It is also a problem when running
   a command such as ls *.c in a directory with a large number of files
   which match the filespec.
   The general solution is to construct your command in such a way that
   it does not have to include all of the filenames on the command line.
   You can use the echo command, which is built into the shells and
   therefore is not subject to the 5k limit. For example, rather than rm
   V*, you might try echo V* | xargs rm. A similar, but somewhat more
   complex solution, might involve using the ls command to generate a
   list of filenames, and then using a command such as grep to filter
   them; ls | grep '^V' | xargs rm will perform the same task as the
   above example.
   You may also find the find command to be useful in this; however, it
   works recursively so it may not be appropriate in a directory with
   subdirectories. Please consult the man pages for each of these
   commands to identify any unexpected side effects they may cause.
   Another alternative, in cases where the environment is unnecessarily
   large, is to reduce its size. If you have some environment variables
   that you never use (be careful with this, as the system or some
   commands may use things you don't realize), you can permanently remove
   them in your .profile (or .login for C Shell users). You could also
   temporarily remove some manually. To run a subshell without any of the
   environment being passed to it, try running env - sh -c 'command'
   OpenServer Release 5 makes two changes to cure this problem. The
   default limit has been increased substantially (to 100k), which should
   by itself fix almost all instances of "arglist too long". As well, it
   is now a tunable kernel parameter, so if the default isn't adequate,
   you can adjust it. One exception: /bin/csh still has a hard-coded
   limit to the length of a line. If you are using csh, you may wish to
   replace it with tcsh (discussed below).
   [Table of Contents]
   What versions/configurations am I using?

   WARNING: Many of these commands have different options under different
   versions of different operating systems, and not all of them are
   available under all versions of Unix, Xenix, and ODT. I've tried to
   note such differences but I'm sure many have escaped my attention.
   Take the following with a grain of salt. Unless noted otherwise, these
   entries should be applicable to most/all systems.
     * Kernel Configuration: configure -x | more (for Xenix, run this
       from /usr/sys/conf; for Unix, run it from /etc/conf/cf.d). This
       lists the current and default values for tunable kernel
       parameters. Under Unix, /etc/sysdef prints information including
       BTLDs (Boot Time Loadable Drivers).
     * Software Installed: /usr/bin/swconfig -p and /usr/bin/swconfig -a
       (both for Unix) print various information on installed software.
       You can look at the permissions lists in /etc/perms/* but you
       cannot tell from here which parts are installed; use custom for
       that. Use /usr/bin/displaypkg to display software installed using
       installpkg. Note that swconfig is not a terribly accurate guide.
     * Hardware configuration: /etc/hwconfig -h shows most of the
       installed hardware but not all of it; generally, things like
       multiport cards don't show up here. Use /etc/hwconfig -hc on Unix
       3.2v4.x or later and on Xenix 2.3.4
     * System name, version, etc.: uname -X (Unix and Xenix 2.3.4) or
       uname -a (Xenix 2.3.3 and earlier)
     * Printer configuration: lpstat -t
   Tom Melvin notes:
   A new program with 5.0.6 is /etc/sysinfo
   See also
   [Table of Contents]
   I have a bad block on my hard drive

   You will see error messages going by giving you the sector, cylinder,
   head, and other nifty information regarding the error. If you can jot
   this down, it makes it much easier to find the bad block without
   having to scan the entire drive for it.
   Shut the system down cleanly (using shutdown). If the error is on the
   root filesystem, boot from emergency floppies; otherwise, you can boot
   from the hard drive and enter single-user mode. The rule here is that
   the filesystem on which the error is located must not be mounted while
   you try to fix it.
   If you have a SCSI hard drive, use scsibadblk. It ships with Unix
   3.2v4.1 and 3.2v4.2, and ODT 2.0 and 3.0. For Unix 3.2v4.0, install
   the 4.1 maintenance supplement or upgrade to 4.2 (not a bad idea
   anyway). For Unix 3.2.2 or ODT 1.1, install unx347a (no longer
   available). For Xenix 2.3.4, install xnx348a. For OSR5, scsibadblk was
   rolled into badtrk, so just use badtrk. For older versions of Xenix or
   Unix, you're out of luck. One other note about SCSI drives; many of
   them will automatically remap bad blocks, so when you go to run
   scsibadblk you will not actually find any bad blocks - even if you run
   a thorough scan of the area where the bad block was reported. This
   capability is called AWR/ARR. If you see a menu option called
   something like "Modify target parameters", you can enable and disable
   AWR and ARR.
   If you're using a standard drive type (MFM, RLL, ATA, ESDI), use
   /etc/badtrk. I'd recommend doing a thorough, nondestructive scan of
   the area where the error message said there was a bad block.
   Before doing this stuff, have a look at the manual for your specific
   operating system to see any notes or recommendations made by SCO. If
   you're not careful here, you might make things worse than they already
   are (such as by doing a destructive scan, which will wipe out all data
   on the area you scan).
   [Table of Contents]
   My system is slow

   First things first - make sure that somebody didn't accidentally turn
   the Turbo switch off. Don't laugh - I have a client who regularly
   manages this one. At some sites, it may be wise to disconnect this
   switch entirely. It might also be wise to run the system's CMOS setup
   program and ensure that primary and secondary cache is turned on,
   unless you know for a fact that there's something in your system that
   won't work properly that way. Turning on BIOS shadowing will generally
   only speed things up at boot time; with the exception of vbiosd (used
   to call real-mode video BIOS routines for video mode switching on some
   video cards in SCO's X11R5 implementation), the BIOS is not used after
   this point. If you gain the use of extra RAM by disabling BIOS
   shadowing, you should certainly do so; even if you don't, there may be
   cases where BIOS shadowing may lead to weird problems (I've even seen
   a host adapter which wouldn't work at all if its BIOS was shadowed or
   cached, for example).
   Under both Unix and Xenix, you can use vmstat to give you an overview
   of system performance. One problem is that it won't show you what
   percent of the system's time was spent waiting on I/O devices, and
   what percent was spent idle; these are both lumped together as idle
   time. vmstat can be helpful in diagnosing excessive swapping, and in
   finding if your system is CPU-bound.
   Unix also offers sar, which is far more advanced than vmstat. It
   reports on a wide range of system statistics including CPU utilization
   (system, user, idle, waiting for I/O), memory use, disk cache
   effectiveness, swapping/paging, and things you've never even thought
   of. Note that under MPX, it may not be reliable; check your MPX
   release notes for info (and for information on the mpstat and mpsar
   programs). One third-party program which may be useful in conjunction
   with sar is sarcheck (Aurora Software Inc., P. O. Box 1033, Plaistow
   NH 03865, (603) 382-4200,,, which translates sar's results into
   English to identify system performance bottlenecks and suggest
   possible resolutions for these problems. sarcheck also works on
   multi-processor systems.
   There are some other utilities you may wish to use. Some freely-
   available ones include u386mon, bcw, and cpuhog/iohog/memhog, all of
   which are available in various TLSes (tls518 for OSR5, tls018d for
   older versions- but not Xenix). u386mon is a general performance
   monitoring utility which watches about as many different things as sar
   (but presents the information in a full-screen display format); bcw is
   the Buffer Cache Watch, which can help you see how well your cache
   buffers are tuned for your system's actual needs; the hog programs
   show you processes which are hogging those respective resources.
   Another commercial product which may be of use is Olympus Tuneup
   (Olympus Software, (408) 426-7582,, which will
   monitor how your system is making use of tunable kernel resources and
   can perform tuning for you.
   Multiuser/multitasking/etc. operating systems love extra memory. Xenix
   will use up to 16 MB; Unix will use much more (how much depends on
   what version; check your release notes). There are several ways that
   extra memory is used; here are three of the most important. First,
   disk buffers; the system uses these for disk cache, and in general,
   the more, the better. Second, to avoid swapping; while a virtual
   memory system allows you to access more memory than you actually have,
   doing so involves the hard drive, which is several orders of magnitude
   slower than memory. Third, the system keeps recently-used programs in
   memory; if you access one again, it doesn't have to be reloaded from
   disk. There are tradeoffs between #1 and #2+#3; the more memory you
   have, the more generously each can be configured. Note that adding
   more memory will not cure CPU-bound processes, and will only cure
   I/O-bound processes if it can be used effectively as a disk cache
   (often it can, but not always).
   Roberto Zini: I seem to remember that some "old" systems could start
   crawling after adding more RAM; if I remember correctly, that was due
   to the fact the CPU could not cope with the additional RAM since it
   had too little internal cache. I'm not an hardware expert so the above
   could be plain wrong nowadays; could you confirm that ?
   Yes. CPU cache is still important- Tony Lawrence
   Also, double check the "netstat -m" output; we're currently fighting
   against a problem under SCO OpenServer 5.0.5 (fully patched) which
   causes it to crawl when STREAMS resources get low. If you notice
   non-zero values under the "fail" column, it's time to add more STREAMS
   buffer by making use of the configure utility under /etc/conf/cf.d
   (NSTRPAGES is the parameter to boost).
   [Table of Contents]
   Why did my region table overflow?

   Each process generally consists of several (usually, but not always,
   three) regions - typically code, data, and stack. Two copies of the
   same program running at the same time will often share code, reducing
   the number of regions required; however, there's nothing to stop a
   program from using more than three regions, either.
   There is a tunable kernel parameter, NREGION, which specifies the
   maximum number of regions available. This should always be set to at
   least three times the number of processes (NPROC), and if you want to
   be on the safe side, use four times NPROC. Note that in OSR5, by
   default, both NREGION and NPROC are allocated dynamically.
   [Table of Contents]
   How do I solve "fork failed: no more processes"?

   This is usually one of two things. There is a tunable kernel
   parameter, NPROC, which determines the maximum number of processes
   that may be running at any time. You may have exceeded this limit. The
   usual method of solving this is to increase it a fair bit and see if
   the problem goes away. If you are running on OSR5, this is unlikely to
   be the reason, as NPROC is allocated dynamically.
   There is another tunable kernel parameter, MAXUPRC, which determines
   the maximum number of processes any one user may have running at one
   time. Under Unix, for example, a large number of mail messages being
   processed at once may cause this to be exceeded by MMDF, usually
   resulting in "uux failure - pipe broke" or similar messages. Once
   again, increase it and see if the problem goes away.
   Also, have a look at the console and/or /usr/adm/messages for any
   system messages which appeared at the same time the user got this
   message. They may point to another potential reason, such as being out
   of swap space or exceeding NREGION (see the previous topic).
   [Table of Contents]
   How are minor device numbers assigned by mkdev hd?

   Basically, they start at 64 (the major device number is 1) and go up
   by 64 each time you run mkdev hd. Don't expect them to be in the same
   order as your SCSI IDs for the drives unless that's the order you
   added them in. Also, if you being running mkdev hd but do not complete
   the process, it will generally already have assigned the next number;
   the next time you run mkdev hd, it will add another 64 even though you
   aren't actually using the last drive you started to create. This isn't
   a problem; it just looks weird.
   [Table of Contents]
   How do I install a BTLD on a running system?

   Mount the floppy and run "btldinstall":
   mount /dev/fd0 /mnt
   btldinstall /mnt

   [Table of Contents]
   How do I add a Zip or Jaz drive?

   You need to have the SCSI versions- I don't know of any drivers for
   the parallel port types. The SCSI versions are added as any SCSI hard
   drive is: "mkdev hd". I'd suggest that you name the filesystem
   something like "zip" or "jazz".
   When you want to mount this, just "mount /dev/jaz /mnt" etc. If you
   have enabled Dos filesystem support (on OSR5 "mdev dos" ), you can
   mount Dos filesystems on these cartridges.
   To create a file system on a blank or Dos cartridge, leave it
   unmounted and run "divvy /dev/jaz" (or whatever you called it). Follow
   the prompts to create a new filesystem.
   [Table of Contents]
   Where can I get older versions of SCO products?

   You really should be upgrading. Even if you can find what you need
   (usually it's TCP/IP that folks are looking for), it's going to be
   very expensive- if you find it at all..
   These folks MAY be able to help you: Blue Chip Computer Systems: 310-410-0126.
   [Table of Contents]
   How do I capture program screens to a file?

   To capture text output, use "script". By default, the file will be
   called "typescript", but "script myfile" will use that name instead.
   Press CTRL-D when you are finished recording.
   To capture X screens, the "xwd" program is simple to use. Open an
   xterm, type "xwd > file" and (after pressing ENTER), just click on the
   window you want to capture. The resulting file can be printed using
   "xpr" or manipulated by graphic programs like "Gimp" and others if you
   need to save it as .gif or some other format.
   Tom Melvin contributed this:
   No idea where I got this from, it's pretty old (Xenix days). Works
   only on the console.

   # Shell script to screen dump
   # This will only work at present on the  main console

   oldstty=`stty -g`
   stty -echo ixon ixoff -ixany
   /bin/echo '\033[2i\c'
   head -25 > /tmp/screen.$$
   stty $oldstty

   # Ok now output the file to the printer
   echo "\f" >>/tmp/screen.$$
   lp /tmp/screen.$$
   rm /tmp/screen.$$

   Or get Chip Rosenthal's prtscrn from any of the 'comp.sources.misc'
   archive sites it's in Volume 22 - This works very well on the console.
   [Table of Contents]
   How can I increase the number of characters that are significant in passwords?

   (This applies to 3.2v4.x and up)
   Two factors control passwords: the maximum length that the password
   can be, and how much of that is significant.
   Both parameters are in /etc/auth/system/default.

   In the above example, u_maxlen#80 means that passwords can be up to 80
   characters long, but u_pwseg#2 limits the significance to 2 segments
   or 16 bytes (2 * 8). If you wanted 24 characters to be significant,
   you'd change it to u_pwseg#3.
   [Table of Contents]
   How do I mount a CDROM?

   Of course you've already run "mkdev cdrom", relinked the kernel and
   In most cases, you can just do:
 mount -r /dev/cd0 /mnt

   You may want to add:
 mount -o lower -r /dev/cd0 /mnt

   To avoid getting everything in UPPER CASE.
   If you get "No such device", you have not correctly identified the
   device. Try "sconf -v" if it's SCSI; if IDE you are probably confused
   as to it being master or slave or whether it's primary or secondary.
   DO NOT assume that just because your system installed from CD that it
   now has a clue where to look- it does not.
   See the other CDROM related entries in this FAQ also.
   [Table of Contents]
   How do I find out who or what halted my system?

   First, look in crontab for a call to haltsys or init. Someone may have
   added this for silly reasons.
   If you think some privileged user or process has run /etc/haltsys, add
   these lines to it right after the PATH= line

 echo $0 `tty` `id`
 while [ $NEXTPROC != 0 ]
 ps -lp $NEXTPROC
 NEXTPROC=`ps -p $MYPROC -o "ppid=" `
 } | mail -s "haltsys was run" root

   This will give you a full trace of where it was called from. You can
   use a similar technique with /etc/shutdown.
   You might also write a "K" script and put it in /etc/rc0.d.
   Unfortunately, by that time there isn't as much information to glean
   from the system. Adding to /etc/rc0 doesn't gain you much either, but
   at least you know it was not a crash and you *might* still see a
   suspect process in a ps listing.
   [Table of Contents]
   How can I regularly and reliably transfer data to a program running on another

   You could write a client on the sending machine and a server on the
   receiving end. Such network programs are not difficult, and you can
   find many examples on the web in C, Perl and other languages.
   However, consider that there are existing client/server programs
   already: mail and remote printing.
   A "printer" on the receiving end can instead process the data it
   receives: see for more
   details on that sort of method.
   With mail, you can set up an alias on the receiving end that runs the
   program you need. That's done by including an alias that might look
   like this:

 mydata: |/usr/local/bin/myprog

   Mail sent to "mydata" gets piped to /usr/local/bin/myprog. If you
   don't have access to the alias file on the remote machine, use
   "procmail" (available from Skunkware).
   The "mail" method has the additional advantage of including mail
   header information that might be of use in some circumstances.
   [Table of Contents]

User Contributions:

Comment about this article, ask questions, or add new information about this topic:

Part1 - Part2 - Part3 - Part4 - Part5 - Part6 - Part7

[ Usenet FAQs | Web FAQs | Documents | RFC Index ]

Send corrections/additions to the FAQ Maintainer: (Tony Lawrence)

Last Update March 27 2014 @ 02:12 PM