APLawrence.com - Resources for Unix and Linux Systems, Bloggers and the self-employed
RSS Feeds RSS Feeds









Bela



Scripting crash with awk to find DMAABLE memory usage (SCO Unix) It's normal for processes to get some DMAABLE memory in the course of normal operation. The VM gives out non-DMAABLE memory in preference, but if that's low it'll give out DMAABLE. 1500 DMAABLE pages is about half your total; if those processes are also consuming about 42000 pages of non-DMAABLE (half that total), it's reasonable. If they are somehow consuming a much higher percentage of DMAABLE than non-DMAABLE then I would agree, there's something peculiar going on. Otherwise it is again a simple matter of eating too much memory (and your system being allergic to swapping, for as yet unknown reasons).


Title Date Comments
no such device or address, no such file or directory   1997-2004 
- If the drive is being recognized, you should have two lines in `hwconfig -h` output. You get a "%tape" line at boot time, indicating that the driver is configured to allow access to a drive. Then, the first time you access the drive, you get a "%Stp" line giving more specific details of the actual drive. -
timeout of ping -->Re timeout of ping   1997-2004 
- The new flag, carped from BSD, means 'stop the whole ping session after this much time, regardless of how many packets have been sent or received, or current success status'. -
gunzip File Size Limit Exceeded (core dumped)   1997-2004 
- The size of an individual file is limited to 2GB in OpenServer. Presumably you mean that the unzipped size is ~5GB and the zipped size is smaller? The amount of data that can flow through a pipe is not limited to 2GB. (But it will still fail if any individual file stored in the tar archive is larger than 2GB.) -
sh vs. ksh   1997-2004 
- The Korn shell does not fork a separate shell in this situation; running the same script under `ksh` would make it work. It was a design goal of ksh to eliminate these language oddities where the scope of a variable is constrained by forces outside your control. -
disconnected telnet sessions; keepalive   1997-2004 
- A machine which is going down in a power failure does not send out "I'm dying now" messages. It is normal for sessions from a rebooted machine to stay alive for a while. Once the client is finished rebooting, next time the server sends it anything on the telnet session, it will get a TCP "reset" packet (meaning "What are you talking about? I don't have a session open with you on that socket number!") -
EAGAIN error -->Re Fork failing, but WHY??   1997-2004 
- EAGAIN error -->Re: Fork failing, but WHY?? -
shmget ipcs shared memory -->Re Memory error relating toshmget   1997-2004 
- shmget ipcs shared memory -->Re: Memory error relating toshmget -
divvy filesystem divvy table mkfs -->Re divvy   1997-2004 
- How SCO's 'divvy' program works. Only when you also tell it to 'create a new filesystem' does it do anything. -
5.0.6 syslog to other machines -->Re OpenServer 5.0.6syslogd won't take remote messages   1997-2004 
- 5.0.6 syslog to other machines -->Re: OpenServer 5.0.6syslogd won't take remote messages -
dlvr_audit processes cron   1997-2004 
- dlvr_audit processes cron -
lan managerclient afps samba   1997-2004 
- lan managerclient afps samba -
identify hard drive sco   1997-2004 
- The relevant information (scsi version ,vendor and product) can be obtained fron 'badtrk' on SCO systems. -
 
 
Kerio Connect Mailserver
 
 
drive size incorrect   1997-2004 
- This is an old article about SCO Unix and improper drive geometry and is only left here for historical purposes. -
lost password unixware 2.12   1997-2004 
- lost password unixware 2.12 -
tcp/ip networking loopback on sco host system   1997-2004 
- tcp/ip networking loopback on sco host system -
man pages display incorrectly 5.0.7   1997-2004 
- man pages display incorrectly 5.0.7 -
does baud rate matter for ssh or telnet?   1997-2004 
- does baud rate matter for ssh or telnet? -
How to capture a panic (SCO Unix)   1997-2004 
- Capturing a panic SCO 5.0.7; low swap prevents you from capturing a crash dump. -
osr5 calendar server ERRNO = 22, T_ERRNO = 0, ISERRNO = 117 (SCO Unix)   1997-2004 
- In order for this to have succeeded, the directory /usr/lib/sco/oadb/caldata has to exist, and it has to contain files named event.dat and event.idx. -
SCO_OSR5.0.7 Maintenance Pack 1   1997-2004 
- Before MP1, the kernel was handing out fake IRQ numbers to devices because IRQ numbers were handed out before SMP IRQ routing had been figured out. Later it figured out that the hardware wasn't going to distinguish between those IRQs, so it would have to invoke multiple drivers' ISRs anyway. In MP1, SMP IRQ routing is figured out earlier, the kernel doesn't have to give out bogus IRQ numbers, and the display reflects this. -
WARNING istat=0A LSS_PAR on ha=0 sist0=41 sist1=00   1997-2004 
- It's really terrible how bad error messages can be, and how misleading. This one actually indicates a SCSI parity error. -
Intel P4 D865PERL motherboard   1997-2004 
- /boot reads a hard disk boot filesystem using BIOS hard disk read functions; it reads floppy and CD-ROM filesystems using BIOS floppy disk read functions (i.e. the same function, different "drive number"). -
can't mount full filesystem SCO osr5   1997-2004 
- can't mount full filesystem SCO osr5 -
SCO_OSR5 fsck   1997-2004  2010/01/04 TonyLawrence
- If you're getting those messages, the system is running fsck for youI think you would only get those messages about the root filesystem after a question like "Root filesystem is dirty, OK to check it?". One thing you can do is: say no. Then go to single-user mode and run: fsck -o full /dev/root -
OSR5 find -newer   1997-2004 
- This is just one of many examples where assumption can screw you up. You are familiar with Linux, BSD, Solaris, SCO or whatever and you find yourself on another Unix operating system. You use a command you are used to using and assume it will work as it always has. It might not - even if the man page says it will!. OSR5 `find` can only keep track of a single "-newer" flag. What actually happens is that the whole expression is evaluated, but it only keeps one timestamp around for the "-newer" flag(s). It stat()'s each file mentioned in a "-newer" clause, and ends up evaluating the whole expression relative to the last stat. -
Compaq Evo D550   1997-2004 
- OpenServer 5.0.4 shipped in 1997, of course it does not support the -
osr5 5.0.7 oss646 oss631 gcc gwxlibs   1997-2004 
- Most of what you need is in OpenServer 5.0.7 as it shipped; but at some date later than this writing, SCO intends to ship updated versions of oss646 and oss631 -- possibly named oss646b and oss631c -- which will then apply to OSR507, and correct a few problems. -
sudo and nis   1997-2004 
- On OpenServer, the user lookup routines (getpwent() etc.) in libc do not see NIS users. There is an override set of these routines in libsocket, which does see NIS users. -
p4 pentium iv hyperthreading osr5 acpi mps cputables   1997-2004 
- _Some_ BIOSes for Pentium 4 family CPUs that support Hyperthreading (HT) set the CPUs up in a manner that OSR5 supports. Others do not. The technical detail: OSR5 supports the Intel MultiProcessing Specification, versions 1.1 and 1.4 (MPS 1.1 / 1.4). An MPS-compliant BIOS sets up a particular table that shows what CPUs are available. Intel has chosen not to support MPS with the HT CPUs. They instead provide an ACPI (Advanced Configuration and Power Interface) table of CPUs. -
osr5 USB support uhci ohci ehci   1997-2004  2012/05/29 TonyLawrence
- The USB implementation in rs50a only supports one of three USB chipset standards. The implementation in OSR507 will support all three. (Before you ask, there are currently no plans to backport the 507 implementation to 506. It depends on significant changes in kernel infrastructure.) -
drive geometry 3.2v4.2   1997-2004 
- It doesn't matter what geometry you claim as long as it multiplies out to the right size (or a bit less) -
login at console ioctl.syscon   1997-2004 
- I rebooted the system from a telnet session on a pseudo terminal and now I've lost the login prompts on the console. -
xtod vs. lponlcr stairstepping printers printing   1997-2004 
- xtod vs. lponlcr : stairstepping printers printing -
3.2v4.2 boot disk images   1997-2004 
- I hope you mean "3.2v4.2". Back then we had a different tape for every possible combination; in particular, an ODT tape was very different from a Unix tape. I'm not sure whether they can be used with interchangable boot disks, or whether the boot disk is directly tied to the tape. -
/usr/lib/grafinfo/grafdev changing X graphicsresolution   1997-2004 
- /usr/lib/grafinfo/grafdev changing X graphicsresolution -
lock process to specific cpu; lockpid   1997-2004 
- lock process to specific cpu; lockpid -
NOFILES Limits -->Re file descriptors   1997-2004 
- The NOFILES maximum was increased to exactly 11000 in SCO Unix 3.2v4.2 -
setrlimit, ulimit -->Re setrlimit   1997-2004 
- Older shell's don't necessarily implement ulimit, causing confusion. -
vhand -->Re free-memory analysis, Re vhand runaway onQuad PIII   1997-2004 
- vhand -->Re: free-memory analysis, Re: vhand runaway onQuad PIII -
dmesg and panic -->Re redirect console error toemail?   1997-2004 
- 'dmesg' is an _old_ utility, by which I mean: it came on the tape from -
divvy filesystem larger than device -->Re Unable tomount 2nd hard drive, filesystem size is larger than device   1997-2004 
- divvy filesystem larger than device -->Re: Unable tomount 2nd hard drive, filesystem size is larger than device -
slow system out of memory swapping -->Re What is causingmy server to page?   1997-2004 
- slow system out of memory swapping -->Re: What is causingmy server to page? -
vmware bochs osr5 under linux -->Re INT13 function FAhduring boot?   1997-2004 
- That's probably the issue. Not a real problem anyway. (If they truly disagreed and the OS was _higher_ then it would eventually -- when the "disk" was >99% full -- attempt to write past the end of the image, get some sort of failure, and behave weirdly since it wouldn't expect that -- but since the OS is coming in lower, at worst it'll just miss out on one cyl's worth of storage.) -
mounting old hard drive divvy disk geometry -->Re Can'tmount drive with install disk.   1997-2004 
- SCSI disks on OpenServer obtain their geometry from a combination of the motherboard BIOS, host adapter, host adapter BIOS, and disk. -
boot images password recovery mount install -->Re SCOboot disk images   1997-2004 
- Before OpenServer 5.0.0, the boot disks were a set of N1 (boot and kernel) and N2 (root filesystem, including things like the mount command). In 5.0.x, there is a single boot disk (N00) which contains boot, kernel, and a minimal root filesystem. That root filesystem has enough code to ask you where to find the install media, mount that, and proceed. -
5.0.6 filesystem sizes -->Re SCO Unix 5.0.6a   1997-2004 
- 5.0.6 filesystem sizes -->Re: SCO Unix 5.0.6a -
tcp connections pseudo ttys NSTRPAGES -->Re No more than230 TCP connections.   1997-2004 
- You need to run `netconfig`, click on "SCO TCP/IP" below your NIC (not the one below "Loopback driver"). Then go to [Protocol -> Modify protocol configuration] and raise the number of TCP connections. You can also raise the number of pseudo ttys here. These thing scost a bit of memory, but not terribly much. Try setting TCP connections to 1024, pseudo ttys to 512. -
maximum number of nic cards -->Re Maximum number of NICsunder SCO OSR 5.0.x   1997-2004 
- NIC drivers which are shipped with SCO OpenServer undergo a complex testing regimen, one part of which is testing with four identical NICs in a single machine. From reading the AOF files I see that there are a few exceptions to this rule: some of the IBM Token Ring adapters are limited to two boards, the Racal InterLan NI6510 EtherBlaster ISA is limited to 1, and three board families support up to 8 NICs: DEC FDDIcontroller/EISA (DEFEA), DEC FDDIcontroller/PCI (DEFPA), and the Intel Pro100B family. -
umask, configuration and default settings   2010 
- A very detailed and quite geeky discussion of umask on SCO Unix 5.0.7 and how tcb was implemented by Secureware -
SCO Link Failure - Kernel Link Fails   1997-2004 
- I object to people not doing the simplest investigation, instead falling back on the MS standard of "reboot, then if that doesn't work, reinstall". -
 
 
Kerio Control Firewall
 
 
Install SCO Unix on non-default division/partition   1997-2004 
- To install anywhere other than the default division-2-of-active-partition requires hackery that few people know -
vmware and old sco image   1997-2004 
- Old SCO 3.2v4.2 in vmware? Obtaining memory map with 'defbootstr prompt' -
old sco version memory mem=   1997-2004 
- old sco version memory mem= -
prngd killed all processes   1997-2004 
- In the first event, the process was being killed by signal 9 (SIGKILL). Many other processes had pending SIGKILL signals. -
Moving OSR5 Swap device   1997-2004 
- Probably the most proper way to change this is to actually change the kernel's default idea of the swap device. This is done in the file /etc/conf/cf.d/sassign; see sassign(F). -
Understanding serial 'hold-open' scripts   1997-2004 
- 2 billion seconds is about 63 years. If the system stays up that long and someone has to figure out why the ancient serial printer goes wrong, I'll owe you a nice dinner. -
slow sort SCO OSR6   1997-2004 
- Comparing various versions of SCO Unix sort, including xenix-386 and Unixware. There's a chance that the GNU sort source tree includes alternate collation libraries that would be faster. -
Why OSR5 sort is so slow   1997-2004 
- It's caused by the implementation of internationalized collation sequences. This is code in libc which was imported into OSR5 from UnixWare 2. -
Proliant 1600 Trap E   1997-2004 
- It looks like 5.48a was the last version claiming to support your Proliant 1600. -


More Bela articles

pavatar.jpg
book graphic










Have you tried our Tests?