tomj@pnet16.cts.com (Tom Jenkins) (05/27/91)
HELP! Need hints (Xenix hangs with Digiboard)
I've got a problem you may be able to help me out with. Equipment
configuration follows. The problem is that our Xenix system hangs
every two-three weeks for no apparent reason. I have to manually
reboot to clean up the filesystems. Keyboard on the host locks up
and although our 8 dial-in lines answer the incoming call, no login
prompt. Everything just seems to freeze.
I can't take out the Digiboard since the system is used online and
can't be off.
This has been ongoing for more than a year, and I've changed out:
Motherboard, Digiboard, Power Supply, VGA Card, Monitor and Hard
drive (used to have a 100M Toshiba, same thing)
I've tried different memory (high and low) combinations on the Digiboard,
with no success (I/O port 0x320, Memory Address's of F00000 and D0000)
I'm running V4.6.0 on Digiboard software and standard Xenix drivers for
the Archive.
DTK 80386 at 25Mhz with 64K cache, 4M Ram
330M Imprimus EDSI, 1.2M Floppy
PC/Xe Digiboard 8 Port Serial I/O (7 ea PM2400SA Practical Perhipherals)
(1 ea T2500 Telebit Trailblazer)
Archive 60M Tape Drive
Relisys VGA with ATI WonderCard
NO printer
Tailing our /usr/adm/messages:
SCO XENIX SYSTEM V
Portions Copyright 1980-88 Microsoft Corporation.
Portions Copyright 1983-88 The Santa Cruz Operation, Inc.
All rights reserved.
Use, duplication, and disclosure are subject to the terms
stated in the customer license agreement.
SCO is a trademark of The Santa Cruz Operation, Inc.
XENIX is a registered trademark of Microsoft Corporation.
SysV release 2.3.2 kid 5.22 for i80386 Serial Number: xxxXXXXXX
A10 bits of I/O address decoding
Fri May 10 15:30:32
device address vector dma comment
----------------------------------------------------------------------------
%floppy 0x3F2-0x3F7 06 2 unit=0 type=96ds15
%tape 0x220-0x224 05 3 type=A
%pcxx 0x00000320 - - mem=0x000D0000 ports=8 PC/Xe V4.6.0
%console - - - unit=vga type=0
%disk 0x1F0-0x1F7 36 - type=W0 unit=0 cyls=1410 hds=13 secs=34
Autoboot from rootdev 1/40, pipedev 1/40, swapdev 1/41
Fri May 10 15:30:33
mem: total = 4096k, reserved = 4k, kernel = 1164k, user = 2928k
kernel: drivers = 0k, 12 screens = 68k, 507 i/o bufs = 507k, msg bufs = 8k
nswap = 15000, swplo = 0, Hz = 50, maximum user process size = 7428k
Wed May 15 10:30:31
Anyone have any other hints?
--Tom
(UUCP: ...!nosc!humu!nctams1!tomj)
UUCP: humu!nctams1!pnet16!tomj
ARPA: humu!nctams1!pnet16!tomj@nosc.mil
INET: tomj@pnet16.cts.comchip@chinacat.unicom.com (Chip Rosenthal) (05/27/91)
In article <1991May26.201728.26322@ncep1.uucp> tomj@pnet16.cts.com (Tom Jenkins) writes: >The problem is that our Xenix system hangs >every two-three weeks for no apparent reason. I'm not sure why you are blaming the Digiboard. The PC/8e is a fine card - the one I recommend for 8-port UNIX and XENIX systems. I'd suggest that you are running out of clists. If you've got eight ports activly banging away - ain't no way the distributed configuration is going to work without hanging. Increase NCLIST and rebuild your kernel. (I won't repeat the procedure - I summarized it earlier today in another article.) TFM (section 8.7.2) in the administrator's guide suggests 5 to 10 clists is the average required per terminal. (Don't count just serial lines - you've got a bunch of multiscreen's too, right?) The max is 16. So, if you've got 8 serial lines and 4 multi screens active, that means you need 60 to 120 clists average, 192 maximum. SCO distributes XENIX with NCLIST set to 100. I'd recommend 200 for your configuration. Systems do wierd things when you run out of clists - kinda like what you are describing. P.S. It sure would be nice to have a `crash' program for XENIX to debug these sorts of things. -- Chip Rosenthal <chip@chinacat.Unicom.COM> | Don't play so Unicom Systems Development 512-482-8260 | loud, Mr. Collins.
jca@pnet01.cts.com (John C. Archambeau) (05/27/91)
tomj@pnet16.cts.com (Tom Jenkins) writes: > > [ Stuff deleted ] > >DTK 80386 at 25Mhz with 64K cache, 4M Ram >330M Imprimus EDSI, 1.2M Floppy >PC/Xe Digiboard 8 Port Serial I/O (7 ea PM2400SA Practical Perhipherals) > (1 ea T2500 Telebit Trailblazer) >Archive 60M Tape Drive >Relisys VGA with ATI WonderCard >NO printer > >Anyone have any other hints? I personally think it's the DTK motherboard. There's a documented problem with the DTK motherboard in running it under Novell Netware. Deals with the 386 going in and out of protected mode along with possibly V86 mode. I have no great love for DTK. Any motherboard that requires a Novell certified patch to run Netware makes me very suspicious and dubious about running Unix or Xenix under it. Have you tried a non-DTK motherboard? Such as an AMI or Micronics? It doesn't have to be a name brand, just non-DTK. A generic 386 board with an Intel or C&T chip set may do the job for you. Just avoid those that don't use the AMI or Phoenix BIOS. If somebody is running VP/ix, it would not surprise me one bit if your DTK motherboard is locking up on you. // JCA /* **--------------------------------------------------------------------------* ** Flames : /dev/null | What to buy? ** ARPANET : crash!pnet01!jca@nosc.mil | EISA or MCA? ** INTERNET: jca@pnet01.cts.com | When will the bus wars end? ** UUCP : {nosc ucsd hplabs!hp-sdd}!crash!pnet01!jca **--------------------------------------------------------------------------* */
tomj@pnet16.cts.com (Tom Jenkins) (05/30/91)
chip@chinacat.unicom.com (Chip Rosenthal) writes: >In article <1991May26.201728.26322@ncep1.uucp> > tomj@pnet16.cts.com (Tom Jenkins) writes: >>The problem is that our Xenix system hangs >>every two-three weeks for no apparent reason. > >I'm not sure why you are blaming the Digiboard. The PC/8e is a >fine card - the one I recommend for 8-port UNIX and XENIX systems. > >I'd suggest that you are running out of clists. If you've got eight >ports activly banging away - ain't no way the distributed configuration >is going to work without hanging. Increase NCLIST and rebuild your >kernel. (I won't repeat the procedure - I summarized it earlier today >in another article.) > >TFM (section 8.7.2) in the administrator's guide suggests 5 to 10 >clists is the average required per terminal. (Don't count just serial >lines - you've got a bunch of multiscreen's too, right?) The max is >16. So, if you've got 8 serial lines and 4 multi screens active, that >means you need 60 to 120 clists average, 192 maximum. SCO distributes >XENIX with NCLIST set to 100. I'd recommend 200 for your configuration. > >Systems do wierd things when you run out of clists - kinda like >what you are describing. > >P.S. It sure would be nice to have a `crash' program for XENIX to >debug these sorts of things. > >-- >Chip Rosenthal <chip@chinacat.Unicom.COM> | Don't play so >Unicom Systems Development 512-482-8260 | loud, Mr. Collins. Ok, I've changed it to 300. Will let you know if the same thing happens. Btw, thanks for all who responded, and especially Chip and John. --Tom Jenkins UUCP: humu!nctams1!pnet16!tomj ARPA: humu!nctams1!pnet16!tomj@nosc.mil INET: tomj@pnet16.cts.com
cliffb@cjbsys.bdb.com (cliff bedore) (05/31/91)
In article <1991May27.061616.5588@crash.cts.com> jca@pnet01.cts.com (John C. Archambeau) writes: >tomj@pnet16.cts.com (Tom Jenkins) writes: >> >> [ Stuff deleted ] >> >>DTK 80386 at 25Mhz with 64K cache, 4M Ram >>330M Imprimus EDSI, 1.2M Floppy >>PC/Xe Digiboard 8 Port Serial I/O (7 ea PM2400SA Practical Perhipherals) >> (1 ea T2500 Telebit Trailblazer) >>Archive 60M Tape Drive >>Relisys VGA with ATI WonderCard >>NO printer >> >>Anyone have any other hints? > >I personally think it's the DTK motherboard. There's a documented problem >with the DTK motherboard in running it under Novell Netware. Deals with the >386 going in and out of protected mode along with possibly V86 mode. I have >no great love for DTK. Any motherboard that requires a Novell certified patch >to run Netware makes me very suspicious and dubious about running Unix or >Xenix under it. > >Have you tried a non-DTK motherboard? Such as an AMI or Micronics? It >doesn't have to be a name brand, just non-DTK. A generic 386 board with an >Intel or C&T chip set may do the job for you. Just avoid those that don't use >the AMI or Phoenix BIOS. > >If somebody is running VP/ix, it would not surprise me one bit if your DTK >motherboard is locking up on you. > > // JCA > > /* > **--------------------------------------------------------------------------* > ** Flames : /dev/null | What to buy? > ** ARPANET : crash!pnet01!jca@nosc.mil | EISA or MCA? > ** INTERNET: jca@pnet01.cts.com | When will the bus wars end? > ** UUCP : {nosc ucsd hplabs!hp-sdd}!crash!pnet01!jca > **--------------------------------------------------------------------------* > */ I have 3 machines running on DTK boards (2 20MHZ 1 25MHZ) and they run XENIX just fine. You do need to get the slow boot mod but everything else works fine (including VPIX what little we use it) Cliff
jca@pnet01.cts.com (John C. Archambeau) (06/01/91)
cliffb@cjbsys.bdb.com (cliff bedore) writes: >In article <1991May27.061616.5588@crash.cts.com> jca@pnet01.cts.com (John C. Archambeau) writes: >>tomj@pnet16.cts.com (Tom Jenkins) writes: >>> >>> [ Stuff deleted ] >>> >>>DTK 80386 at 25Mhz with 64K cache, 4M Ram >>>330M Imprimus EDSI, 1.2M Floppy >>>PC/Xe Digiboard 8 Port Serial I/O (7 ea PM2400SA Practical Perhipherals) >>> (1 ea T2500 Telebit Trailblazer) >>>Archive 60M Tape Drive >>>Relisys VGA with ATI WonderCard >>>NO printer >>> >>>Anyone have any other hints? >> >>I personally think it's the DTK motherboard. There's a documented problem >>with the DTK motherboard in running it under Novell Netware. Deals with the >>386 going in and out of protected mode along with possibly V86 mode. I have >>no great love for DTK. Any motherboard that requires a Novell certified patch >>to run Netware makes me very suspicious and dubious about running Unix or >>Xenix under it. >> >>Have you tried a non-DTK motherboard? Such as an AMI or Micronics? It >>doesn't have to be a name brand, just non-DTK. A generic 386 board with an >>Intel or C&T chip set may do the job for you. Just avoid those that don't use >>the AMI or Phoenix BIOS. >> >>If somebody is running VP/ix, it would not surprise me one bit if your DTK >>motherboard is locking up on you. >> */ > >I have 3 machines running on DTK boards (2 20MHZ 1 25MHZ) and they run XENIX >just fine. You do need to get the slow boot mod but everything else works fine >(including VPIX what little we use it) Doesn't matter what your experience is. Novell has acknowledged the problem and you had to apply a modification to get Xenix to run. Supposedly Novell Netware runs just as well with the patch, but the problem is that it has to be patched just for a DTK motherboard. What other problems will one encounter? Will certain cards not work with it because of it being a bit off from the ISA spec? Since I'm the head of tech support where I work, I can make certain decisions regarding what we will and will not support, and DTK products are HIGH on the list. Other problems I've had with DTK. I put a guinine DTK floppy drive controller in a 386SX machine and the video card wouldn't work. I know it's not the video card because of the fact that the same video card was used with the floppy drive controller in my ancient 286. Pull out the DTK floppy and the display comes back. To be fair, I tried the DTK floppy drive controller in another machine, sure enough, it worked. The 386SX motherboard had an Intel/Zymos chip set, nothing unusual. DTK products have given me nothing but problems and I certainly will not buy a DTK product nor will I support them in the field. // JCA /* **--------------------------------------------------------------------------* ** Flames : /dev/null | What to buy? ** ARPANET : crash!pnet01!jca@nosc.mil | EISA or MCA? ** INTERNET: jca@pnet01.cts.com | When will the bus wars end? ** UUCP : {nosc ucsd hplabs!hp-sdd}!crash!pnet01!jca **--------------------------------------------------------------------------* */
bblue@crash.cts.com (Bill Blue) (06/02/91)
In <1991Jun1.015603.6996@crash.cts.com> jca@pnet01.cts.com (John C. Archambeau) writes: }>I have 3 machines running on DTK boards (2 20MHZ 1 25MHZ) and they run XENIX }>just fine. You do need to get the slow boot mod but everything else works fine }>(including VPIX what little we use it) }Doesn't matter what your experience is. Novell has acknowledged the problem }and you had to apply a modification to get Xenix to run. Supposedly Novell }Netware runs just as well with the patch, but the problem is that it has to be }patched just for a DTK motherboard. What other problems will one encounter? }Will certain cards not work with it because of it being a bit off from the ISA }spec? Yes, it does matter what his experience is. It might be different from yours, but it is every bit as significant. I ran crash on a DTK 25mhz board for over a year with nary a problem like you describe, and no patches or mods necessary for it to work. There may be certain DTK products that have incompatibilities, but you certainly can't make a blanket assertion that they all do. Well, you can, but not everybody is going to take it seriously. --Bill and p.s. Xenix != Novell
jca@pnet01.cts.com (John C. Archambeau) (06/03/91)
bblue@crash.cts.com (Bill Blue) writes: >Yes, it does matter what his experience is. It might be different >from yours, but it is every bit as significant. I ran crash on a DTK >25mhz board for over a year with nary a problem like you describe, and >no patches or mods necessary for it to work. There may be certain DTK >products that have incompatibilities, but you certainly can't make a >blanket assertion that they all do. Well, you can, but not everybody >is going to take it seriously. I personally like to eliminate a potential problem before it happens. If there's a potential problem, the product gets chucked out the window. It's much easier to use a product that has been proven to work than one that has questionable track record. If it works for you, great, but it isn't going to change my opinion on the product or manufacturer. >and p.s. Xenix != Novell I'm aware of that fact. One of the guidelines I use is if the product is Novell certified and will run without any problems. If it does and there's driver for the product under the flavor of Unix that is to be used, then chances are there will be little or no problems with the piece of hardware. This is just a rule of thumb that I use which generally does work. // JCA /* **--------------------------------------------------------------------------* ** Flames : /dev/null | What to buy? ** ARPANET : crash!pnet01!jca@nosc.mil | EISA or MCA? ** INTERNET: jca@pnet01.cts.com | When will the bus wars end? ** UUCP : {nosc ucsd hplabs!hp-sdd}!crash!pnet01!jca **--------------------------------------------------------------------------* */
tomj@pnet16.cts.com (Tom Jenkins) (06/06/91)
cliffb@cjbsys.bdb.com (cliff bedore) writes: >In article <1991May27.061616.5588@crash.cts.com> jca@pnet01.cts.com (John C. Archambeau) writes: >I have 3 machines running on DTK boards (2 20MHZ 1 25MHZ) and they run XENIX >just fine. You do need to get the slow boot mod but everything else works fine >(including VPIX what little we use it) > > >Cliff Yeah, since I've changed CLIST's, and relinked the kernel, all seems to be fine. That's the only change. (Btw, factory default on CLIST was 100, I've upped them to 300. Thanks Chip!) --Tom UUCP: humu!nctams1!pnet16!tomj ARPA: humu!nctams1!pnet16!tomj@nosc.mil INET: tomj@pnet16.cts.com