|
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 http://aplawrence.com/SCOFAQ/index.html These FAQS were developed and maintained for years by steved@ussinc.com (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 Because Steve has not been able to maintain these for a while now, some of the information herein is outdated.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 http://aplawrence.com/search.html 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 <njh@smsltd.demon.co.uk>. A demonstration version is available at ftp://garbo.uwasa.fi/unix/util/tapechk.sco.tar.Z 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 http://aplawrence.com/Reviews/supertars.html [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; info@microlite.com; (888) BKP-EDGE or (724) 375-6711; http://www.microlite.com/) * BRU (Enhanced Software Technologies Inc., 5016 S. Ash Avenue Suite 109, Tempe, AZ 85282; swinfo@estinc.com; (800) 998-8649 or (602) 820-0042; http://www.estinc.com/) * Lone-Tar (Lone Star Software Inc., 13987 W. Annapolis Court, Mt. Airy, MD 21771; sales@cactus.com; (301) 829-1622 or (800) LONE-TAR; http://www.cactus.com/) * Ctar (Unitrends Software Corp., 1601 Oak Street, Suite 201, Myrtle Beach, SC 29677; sales@unitrends.com; (800) 648-2827 or (803) 626-2878; http://www.unitrends.com/) 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 shortly. * The Backup Professional (Lone Star Software Inc., 13987 W. Annapolis Court, Mt. Airy, MD 21771; sales@cactus.com; (301) 829-1622 or (800) LONE-TAR; http://www.cactus.com/bp.html) * ARCserve/Open (The Santa Cruz Operation Inc., 400 Encinal Street, Santa Cruz, CA 95061; info@sco.com; (800) SCO-UNIX or (408) 425-7222; http://www.sco.com/) * 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 http://www.sco.com/skunkware/faq.html or read the section below entitled "What is Skunkware?" gcc sources and binaries are also available on Robert Lipe's home page: http://www.dgii.com/people/robertl/ or ftp://ftp.dgii.com/users/robertl/scods/ These are mirrored by SCO at http://www.sco.com/skunkware/gds/ and ftp://ftp.sco.com/skunkware/gds/ You can also look at a different version at ftp://ftp.sco.com/skunkware/osr5/devtools/gcc/ and http://www.sco.com/skunkware/osr5/devtools/gcc/ 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 root:encryptedpasswordgoeshere:0:0:God,Everywhere:/:/bin/sh Edit out the encrypted password (don't touch anything else!) so that the line reads something like root::0:0:God,Everywhere:/:/bin/sh 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 http://aplawrence.com/Boot/defs.html#bootfloppy. [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" passwd 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 tools 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 http://www.sco.com/cgi-bin/ssl_reference?110414 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 script: for a in 0 1 2 3 4 5 6 7 8 9 do for b in 0 1 2 3 4 5 6 7 8 9 do > /lost+found/dummy$a$b done done 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 12345). 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 480563. 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 http://www.cruzio.com/~jeffl/sco/versions.txt [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, http://www.sarcheck.com/, 74013.1625@compuserve.com), 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, olympus@olysoft.com), 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: http://www.bccs.com 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. default:\ :d_name=default:\ :u_pwd=*:\ :u_priority#0:u_cmdpriv=audittrail,su,queryspace,printqueue,mem,termina l:\ :u_syspriv=execsuid,nopromain,chmodsugid,chown:\ :u_minchg#0:u_maxlen#80:u_exp#0:u_life#0:\ :u_pickpw:u_genpwd:u_restrict@:u_nullpw:\ :u_suclog#0:u_unsuclog#0:u_maxtries#99:u_lock:\ :u_singleuserpswd:u_secclass=c2:u_integrity@:u_tcbpw@:\ :u_pwseg#2:\ :t_logdelay#1:t_maxtries#99:t_login_timeout#60:\ :chkent: 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 rebooted. 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` MYPROC=$$ NEXTPROC=$MYPROC while [ $NEXTPROC != 0 ] do ps -lp $NEXTPROC MYPROC=$NEXTPROC NEXTPROC=`ps -p $MYPROC -o "ppid=" ` done } | 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 machine? 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 http://aplawrence.com/Unixart/printing.html 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: