(last updated: 31-Jan-2017)
The current configuration is in its original 'as acquired' condition, except that the internal HDD had been replaced. The system consists of the main chassis with internal HDD (4GB) and DSS tape driver, HP PS/2 keyboard and mouse via HIL connector box, 3-rd party 2x CD-ROM, two 3-rd party external HDD subsystems (5GB and 9GB), and 20" Sampo Sync-on-Green CRT monitor.
05-May-2010
This is one-half of a pair of HP PA-RISC machines that were offered locally on Craigslist. I couldn't resist, having become somewhat of a UNIX bigot, of late. The price was right...$50, which included the main CPU, keyboard, monitor, three-button mouse, external hard disks and CD-ROM drives, along with an HP PostScript inkjet printer, UPS and documentation and HP-UX 11.0 installation media.
The system needs a little cleanup, but is fully operational, as was demonstrated by the seller during the pickup session. I learned how to power up, boot the OS, navigate around the VUE desktop and so one. Very helpful to have had that quick overview.
07-May-2010
The systems is pretty simple, being really not much more than a desktop pizza box. The case splits in half and opens like a billfold. The drives and power supplies are all located in the 'upper' half, while the system PCB and backpanel connectors are in teh 'bottom' half. (photos)
Inside there was a good deal of accumulated dust, but otherwise clean of ramdom pet hair, sticky cola residue, etc. A quick vacuuming took care of the innards.
07-May-2010
Some random observations:
- The system can accept one (1) EISA add-on card, purposes unkown. - The system TOD battery is a BR2325. - All eight RAM sockets are filled, 72-pin DIMM, HP part A56101664871 / A2575-60001 with 18 Kyocera MSM416400BJ chips. - Mounts for two (2) 3-1/2" drives, plus one (1) 5-1/4" drive.
07-May-2010
This system was received with one 50-pin SCSI disk installed, an HP C3325A, which is 2.2GB. It is installed with HP-UX 11.0 and I want to get a complete copy before I do anymore with the system, inlcuding just booting it for the first time.
Using my SGI Indigo2 as a host (which conveniently has an external SCSI port and is running SGI's flavor of UNIX, IRIX, I connected the HP drive, which I had removed in it's mounting tray from the chassis, powering it with a spare PC power supply. The unit was already set as SCSI ID 6, which I left alone, and powered up the Indigo2. The drive was recognized and I was able to test its operation with some random 'dd' commands to confirm that it was working.
On the Indigo2, I mounted the second hard disk, which serves as kind of a scratch disk, in preparation for receiving the disk image:
# mount /dev/dsk/dks0d2s7 /mnt
Now to make the image.
Well! Fool! I messed up the HP-UX drive! I wanted to make sure I was talking to the correct device and used IRIX's 'fx' command as if I wanted to prepare a new drive, hoping for just some feedback about the drive's capacity etcetera. Well, it did that but when exited fx and it asked if I wanted to write changes, I said 'no'. Definitely. Still it wrote a new partition table!!!! Blowing away the drive's original contents. Egads! I said 'No!"
Well, for what it's worth, I'll image the drive anyway, and maybe someday a miracle will happen letting me make something useful out of it:
# dd if=/dev/rdsk/dks1d6vol of=/mnt/remains.dsk bs=512 Write error: invalid argument 4194304+0 records in 4194303+0 records out
09-May-2010
After the trouble yesterday I gave myself yesterday, trying to image the disk on my SGI Indigo2 system, and believing that I had blown away the disk configuration, I am GREATLY relieved to learn that the disk is still in great shape! Wahoo! I guess the SGI partitioning didn't really 'take' on the disk, or maybe only sectors were affected that have no impact on the way HP-UX uses the disk. In any case, I am one happy guy!
Using 'mel0dious' as the root password, I successfully logged into the system and its GUI. From there I played around a bit, and started poking around. The network configuration is set for static IP from an old configuration, so I'll have to change that after learning more.
Later, I shutdown and attached the Legasys single-drive external drive and powered up. I learned that the drive is set for SCSI ID 5 and one the system booted I see that it is completely empty. There is only the usual 'lost+found' and an empty 'tmp' directory. I tagged the drive with a sticker to help me remember what it has.
12-May-2010
Peter found several additional SCSI drives, one enclosed, and three bare. He offered them to me, an offer I happily accepted. Here's what we have:
Bare drives:
- Seagate Hawk ST15230N (5.2GB)
===================
|___| : : : : : :
A A A R L R
0 1 2 S E S
V D V
- Quantum ProDrive GEM 80S (80MB) from Apple Macintosh
- Quantum ProDrive GEM 80S w/ LED (80MB) from Apple Macintosh
"nStor Parity 8300" (enclosed drive):
- Seagate Barracuda ST 19171N (9.1GB)
===================
|___| : : : : : :
L R R A A A
E S S 2 1 0
D V V
- red wire = A0
- yellow = A1
- blue = A2
- white = LED
Checked each drive for basic condition by connecting to a power supply and powering it up. In all cases, except of the nStor device, the drive would not spin up. Looks like each of the loose/bare drives are dead.
14-May-2010
Using the information learned while working on the 712/60 system, I want to try changing this system IP configuration. Currently it is:
Address: 137.190.42.50 Netmask: 255.255.255.0 Gateway: 137.190.42.1
I used /etc/set_parms to change the IP address and other network config elements. Then I added one of my SISNA DNS servers to the /etc/resolv.conf file. After a reboot, networking is alive!
27-May-2010
At last I got around to imaging the internal hard disk. Just in case I need to get back to where things are today, I wanted to take a snapshot of the drive and hold it in reserve.
Because I have no easy way to boot this system to from other media that has network connectivity I decided to try the imaging of this root disk in a 'live' situation. Of course the drive will be mounted and serving as the main disk store for the system, but I believe that a dump of the disk will still be valid. Upon restoration, if that is ever needed, I should probably expect a few errors to some temporary files, but otherwise it should be okay. I won't be doing any other work with the system during this imaging period, and hope it is pretty clean and is self-consistent when complete. Here's the command:
# mount 192.168.1.10:/home/jared /dtrium # time dd if=/dev/dsk/c0t6d0 bs=512 of=/dtrium/HP715.dsk I/O error 4194304+0 records in 4194303+1 records out 158.5u 1485.2s 12:28:23 3% #
wow, very slow! And, as I suspected, with this drive being a tad larger than 2GB, the transfer has reached the maximum file size limit (2GB) on my target machine. I'll pickup the transfer from block 419400:
# dd if=/dev/dsk/c0t6d0 bs=512 of=/dtrium/HP715_part2.dsk skip=4194000
This command took some time to work because it looks like this version of dd isn't smart enough to jump over the skipped blocks, rather it looks like it reads each block in sucsession until reaching the end of the skipped section before transmitting data to the output file. This process started at 23:06.
Because of the two chunks, the restoration of this disk will use the following two commands:
# dd if=.../HP715.dsk bs=512 of=/dev/dsk/c0t6d0 # dd if=.../HP715_part2.dsk bs=512 of=/dev/dsk/c0t6d0 seek=4194000 685+0 records in 685+0 records out #
31-May-2010
After a number of tests to see what's what, I've determined that the external drives are all more or less empty of anything useful. With further testing on this system, and on "fish" the 712/60, I'm now convinced that this is what I have:
- Legasys Double-decker (upper drive) (Seagate ST19171N, 9GB)
Initialized with HFS file-system, but completely empty.
- Legasys double-decker (lower drive) (Seagate ST15230N, 5GB)
Initialized with HFS, with a partial
bootable-only-in-single-user-mode installation of HP-UX 11.00.
- Legasys single-decker (Seagate ST15230N, 5GB)
Initialized with HFS file-system, but completely empty.
- nStore single-decker (Seagate ST19171N, 9BG)
Initialized with HFS file-system, but completely empty.
- generic (no label) enclosure is empty.
13-May-2011
Peter contacted me via email indicating that he had found a few extra bits of hardware related to this system. He was coming to Provo for some academic work and would bring them for me. Included were two SCSI hard drives, 9GB and 4.3GB, some cables and two mice.
I imaged these drives (see notes in "SCSI disk imageing.txt" elsewhere).
27-Aug-2012
Sometime in the past month while using the system to test some SCSI drives, I inadvertently disconnected or connected an external SCSI drive with the system still powered. Since then, the internal drive has failed to boot the system, with a lot of head chatter. I guess my actions may have electrocuted some of the circuitry on the SCSI drive's controller.
The drive is an HP C3325A with jumpers in:
- AUTO SPIN-UP - SCSI PIN 29 - UNIT SELECT 3 (SCSI address 6) - UNIT SELECT 2 (SCSI address 6)
Tonight I removed the drive from the system and connected it to my SCSI imaging rig currently set up on my 386sx/16 ("CompuAdd") system, with an Adaptec AHA-1542B installed. There I was able to boot up into Debian 1.1 ("Buzz") and explore. At least the drive is visible and identifying itself. Sure enough, though, when I tried to 'dd' the drive, it gave many, many errors, all from even-numbered sectors, interestingly enough. That would mean perhaps that that single bit driver was damaged, or perhaps there was just a glitch on a single head inside the mechanicals.
At this point I had little choice but to hope it was the latter. I rebooted the CompuAdd system (currently hosting the drive) and rebooted to MS-DOS. From there I used Debug to execute the SCSI controllers BIOS utility with the command G=DC00:6. Again, I tested the drive with the 'verify' function which showed the errors, too. Last option is to reformat the drive and see if we can recover that way. Of course, that will wipe the data, but I do have a good image of this drives contents already (fortunately!) and can restore the data.
Well, the attempt was not successful. After the format was complete, I used the verify function again. No go. Same old errors. It looks like I permanently damaged the drive and it is no longer usable. Rats!
02-Sep-2012
I removed the original disk from its mounting bracket, leaving the system without a main hard disk. In order to learn about a new system installation I connected up a 2GB drive externally. The system powers up fine and if I break into the BIOS console (with ESC) I can 'search' for devices. Fortunately, the new disk is seen correctly. This assures me that my previous actions only affected the hard disk and not the system's SCSI controller.
With this set up, and after inserting the external CD-ROM drive into the SCSI chain, I mounted one o the HP UX Installation CDs that I have. I used the one with the bluish label. I learned that I can boot from this CD a couple of ways. I can break into the boot sequence (with ESC) and then manually boot to the CD with this command:
> boot scsi.2.0
or, the CD will be booted if I let the power on boot up sequence try to boot from the hard disk (which will fail), then wait and it will eventually discover that it can boot from the CD, which it does.
Once I was in the CD installation sequence, I played around checking all of the options, be eventually went ahead with the installation. At first I tried to just set up the entire drive as a single HFS filesystem on root (/), but the tool complained that the drive was too large(!) and that I would have to use LVM. I don't know much about LVM, but I chose LVM HFS. The last and third choice was LVM VP.xx+ or something, which I'll have to check out some other time.
With that taken care of I just chose to install ALL filesets and select 'Go!'. It looks like the basic installation CD-ROM has all that is needed for an operational system, but does not include much in the way of development tools such as a C compiler, etc. Perhaps I can snag that somewhere, or I'm sure that the GNU tools are available for this platform.
The installation process was lengthy, taking about 3 hours all together, but it worked...sort of. The system installed with about 60 warnings that certain filesets were not installed, and when the system finally came up and entered the GUI, it popped up a message indicating that the windows messaging system failed to start, and suggesting that I use a 'failsafe' console to fix the problem. The trouble is, I have no idea what the problem is! I did see, too, a message suggesting that I should install administrative tools from the Support Plus CD. I don't think I have that. One good thing...no code word was required to install this OS.
I have another HP UX 11.00 installation CD, that does include the Support Plus second disc. Let's try that one!
20-Aug-2013
After discovering a 4-disc set of HP-UX 11.11 CD images on the internet and downloading them it was time to return to this system. Still needed to identify and install a primary hard disk drive to replace the drive that I damaged in 2012.
After discovering that I can boot into a minimal system from the two installation CDs that came with this system, I temporarily installed the two "Unmark Extra" drives. First, starting with the Seagate ST39173N (Unmark Extra #2), I connected the drive and the external CD-ROM drive and powered up. I interrupted the boot sequence by pressing
BOOT_ADMIN> boot scsi.2.0
After choosing Keyboard Type 61 I was presented with an Installation/Recovery menu. I chose
The system has very limited facilities, but I was able to fsck the hard disk and discovered that it was last mounted on '/dtrium':
# fsck /dev/dsk/c0t6d0
and after mounting it:
# mkdir /dtrium # mount /dev/dsk/c0t6d0 /dtrium
by creating /dtrium preserved the state of last mount point on the drive, even after I messed with it.
Looking through the drive's contents, it appears to be mainly a few data files of various sorts, a bunch of downloaded Eudora email installers, and some webserver docs. In the subdir ./opts I found a number of interesting applications, including Apache, Perl5 and other useful things. I don't know how hard/easy it may be to download and install these files in the future, so I chose to leave this drive as is. Even though I have an archived image of its contents, it's just easier to find a different scratch drive.
Next, I repeated the process with the other extra drive, a Seagate ST34371N, "Unmack Extra #1".
This drive was also last mounted on /dtrium. Has some of the same contents, particularly 'www', but it has a whole bunch of software distribution 'depot' files that might prove interesting, so again I don't want to disturb this drive. Have to find another...
I then scrounged up a 1GB disk that was definitely scratchable, cabled it up, and booted. I started the installation process directly, this time, and eventually found after selecting all of the defaults except having one root volume (NO LVM), found that the drive was too small! At least 2GB or more is required. Time to find another drive...
25-Aug-2013
Once installed, when I logged using the graphical CDE login screen, I was unable to get past a popup with the message, "The desktop messaging system could not be started," which only had an 'OK' button which returned to the CDE login screen. The problem seems to be related to the worldwide DNS system NOT having this system listed, i.e., "perch.rogerwilco.org", and because I have no local DNS server that also has that name, the error resulted. A search of that error term produced an answer.
Turns out that in this case the creating of /etc/nsswitch.conf with the following contents will fix this problem:
# echo "hosts: files [NOTFOUND=continue UNAVAIL=continue TRYAGAIN=continue]
dns [NOTFOUND=return UNAVAIL=continue TRYAGAIN=return]" > /etc/nsswitch.conf
# echo "aliases: files" >> /etc/nsswitch.conf
I'm now successfully logging in and exploring HP-UX!
15-Aug-2014
At last I have selected a drive to replace the original internal hard disk that failed. I have a 4GB IBM IBM DCAS-34330 drive that is uses the 60-pin-D connector, but I have a converter that allows me to cable it up with the SCSI-2 50-pin IDC cable inside the box.
For grins, I'm installing HP/UX 10.20 to this new scratch drive, just for the experience. Then I'll restore the original drive's contents.
15-Aug-2014
Using the 10.20 installation CD, I fired up and started the process of installation with this command at the PROM monitor prompt:
BOOT_ADMIN> boot scsi.2.0 hpux
This started the standalone installation kernal and initiated the process of selections to install. I more or less took the defaults through the process, other than editing the disk layout to have only '/', '/usr', '/opt', and swap. Then let it chug away.
At the very end of the installation after the reboot into X, a dialog box gave a hint for future mods to the network configuration I needed to run:
/sbin/set_params addl_network [return]
Sadly, upon reboot and the presentation of the graphical CDE login screen, I can't log in. It appears that my credentials are being accepted, I see a page of copyrights, then the X window starts with an hourglass, then back to the CDE login screen. Need more investigation.
15-Aug-2014
With some good success installing HP/UX 10.20 to the 4GB drive, I felt it was time to restore the original disk image to this new drive. Even though the drive is larger than the image, since SCSI devices are just a sequential series of data blocks and are addressed in such a way it shouldn't fail to work. Of course the filesystem will not "see" the empty second half of the drive but that's fine for now.
Using my Indy #2 as a platform for restoration, I cabled up the new drive to the Indy's external SCSI port addressed as #4 on the SCSI chain. This address must be changed to #6 before installing on this 715/100 system. The Indy is running IRIX 5.3. I copied the disk image files (2) to an exported NFS drive on 'gusoyn' and then used these commands to get the image back onto the new drive:
% su - (enter password) # mount -t nfs gusoyn:/home/jared/Temporary_Files/HP715/ /mnt # dd if=/mnt/Temporary_Files/HP715/Unmack_HP715_internal_HP_C3325A_chunk1.dsk bs=512 of=/dev/rdsk/dks0d4vol # dd if=/mnt/Temporary_Files/HP715/Unmack_HP715_internal_HP_C3325A_chunk2.dsk bs=512 of=/dev/rdsk/dks0d4vol seek=4194000 # umount /mnt exit %
The restoration took about 15 hours in all! A long time! I'm not sure if it was the Indy, the drive, the 10Mbps ethernet or what, but it did finish, eventually.
17-Aug-2014
I fiddled around with the extra external drives, knowing that they were once part of the overall configuration. With the restored contents on the new internal drive in place, I took a look at /etc/fstab which shows:
6 / 5 /ctrium 4 /dtrium
Then, I recalled that the default boot device in ISL shows as scsi.4.0 and knowing from my imaging notes that that device has an installation on it, I tried booting from that device. It all looks great until it tries to start the CDE GUI. All I get is a raw X window, then nothing. After about 10 minutes the system beeps and tries to restart the X system. It just continues like that over and over. So, booting from the #4 drive isn't going to work.
Then I figured it out! It turns out that the other two external drives are the ones currently configured for this installation. The system will boot without errors when I have the nStor Parity+ and the Legasys single external drives attached as SCSI units #4 and #5. Turns out that both drives are completely empty, but at least the system is now happily booting with no errors.
17-Aug-2014
I've been searching for freeware/open-source software binary repositories for this system. HP appears to offer a freeware download site, but it focuses on much later versions of HP-UX than 11.00 so is not helpful at present. The UK-based CONNECT group also offers freeware builds, but is phasing out all of the older builds in favor of the modern HP-UX versions.
I did find some useful sites though:
http://www.cmve.net/~merijn/downloads.html http://mirrors.develooper.com/hpux/downloads.html ftp://ftp.thewrittenword.com
26-Aug-2014
Well, I sure did it!!! Doing some external SCSI HDD testing, I messed up the newly installed replacement drive's contents. Time to restore the original image once again!!!
I tried to use the HP-UX 11.00 installation media to boot to a recovery kernel and shell, which works fine, but there is no means of mounting an NFS drive. So...
I removed the HDD from the system, connected it to a Pentium PC booted to the NetBSD/i386 install kernel, configured the network, and dropped out to a shell. No problem NFS mounting and performed the restoration using dd.
02-Sep-2014
The replacement hard drive is twice as large as the original that it replaces. The restored disk image of course has a filesytem that uses the original drive's capacity, leaving the second half of the new drive unused. Though I could live without it, it would be nice to have the filesystem use the whole device. While poking around, learning more and more about HP-UX, I discovered the 'extendfs' command, that appears to do exactly what I need. It will 'grow' the filesystem to consume the entire capacity of the drive.
Reading the manpage for extendfs indicates that the filesystem must not be mounted when running extendfs. So, how to extend the root filesystem? Hmm...
The way is to boot with the installation/recovery kernel from CDROM, then extend the filesystem. I booted from the CDROM, selected Recovery Shell, from the menu load necessary/useful tools, then exited to the shell and ran extendfs:
l. load files fsck mount umount extendfs bdf (type 'y' and load the files)x. exit to shell
# extendfs /dev/rdsk/c0t6d0 # sync # reboot
It went to work, showing a good number of new super-block locations and that was that! I issued the sync command for grins, probably isn't necessary, and rebooted.
Well, not so fast! All looked to be copacetic, but in the end, the system refused to boot. I could try automatically, or manually, and all I would get is the original spash screen then blank, and all activity stops.
I tried 'fsck' and it had to make a correction to the cylinder groups but other than that, all was well. I could mount the drive and examine directories and files, but when I tried to boot again, no good, it just blanks the screen and hangs. Have to press the RESET button.
In the meantime, while trying to sort this out, I discovered that I have the ability to mount nfs drives once the 'mount' files were loaded:
# mkdir /mnt # cd /sbin/fs/nfs # ./mount [nfspath] /mnt
So, I restored the original disk image all over again. All went well, and after about 6 hours, I was ready to reboot the restored image.
After the restoration, there are a few changes to make the system more convenient to use:
- Using SAM, create a new daily user with password.
- rename the following files:
-- mv /etc/copyright /etc/copyright.orig -- mv /etc/issue /etc/issue.orig -- mv /etc/motd /etc/motd.orig -- mv /etc/hosts /etc/hosts.orig -- mv /etc/resolv.conf /etc/resolv.conf.orig
- then recreate them:
-- echo "Welcome to perch!" > /etc/copyright -- echo "" > /etc/issue -- echo "" > /etc/motd -- echo "nameserver 192.168.1.1" > /etc/resolv.conf
And:
-- Use ftp to copy hosts to /etc/hosts
05-Sep-2014
With this system running again, I want to add it to my 'rack' of systems that are operated remotely from time to time, with no need for keyboard, mouse or monitor.
Well, I've discovered that the system requires a keyboard at the very least, or it fails to start. I can find no way in the PROM configuration to disable this requirement. Some web searching turns up nothing either, except one reference that indicated that it wasn't possible, confirming the requirement for the keyboard. So be it.
Also, a mouse is required for the system to start the X Windows system. Otherwise, it just tries over and over the fire up X without success. I can still login remotely, or on the local terminal, again as long as the keyboard is attached.
31-Jan-2016
I removed the CR2025 that I had in the system, replacing it with the correct size of battery. The CR2025 still was at 2.7v even though the system has been powered-off for some time.
30-Jul-2016
For this effort I used a scratch SCSI disk connected to the sytsem via the external SCSI port. This allowed experimentation without disturbing the HP-UX installation on the internal drive.
This was a very difficult installation. The NetBSD distribution, even though available in a CD image (.ISO), is only installable by first netbooting the installation kernel. The installation documentation from the distribution has explicit instructions for setting up the BootP and TFTP (and potentially NFS) servers for netbooting. The example /etc/bootptab is very clear and echoes the configuration for other platform distributions. However....it doesn't work.
I spent hours in all, confirming that each server was installed and running on my usual netbooting host server (Debian GNU/Linux 3.0 "Woody") and finally (at last!) hit upon what really worked.
The NetBSD/hppa installation instructions suggests using a DHCP deamon in BOOTP mode. In the past, I've used BOOTP itself, so I went with it. In another generic netbooting How-To the /etc/bootptab configuration example is:
#/etc/bootptab
client.test.net:\
:ht=ether:\
:ha=CCCCCCCCCCCC:\
:sm=255.255.255.0:\
:lg=192.168.1.5:\
:ip=192.168.1.10:\
:hn:\
:bf=[/tftpboot/]boot.netbsd:\ # see note below
:bs=auto:\
:rp=/export/client/root/:\
:vm=auto: # see note below for next hardware
:vm=rfc1048: # see note below for alpha hardware
Using this kind of configuration DOES NOT work with this HP system. It seems that the boot routine in the HP firmeware has a particular quirk.
The key is to NOT specify the
rp=element. Instead, included the full absolute path in the
bf=element. Here is the final configuration that worked for me on this machine:
#/etc/bootptab HP715-100:\ :ht=ether:\ :ha=0800097b8ed6:\ :ip=192.168.1.58:\ :bf=/tftpboot/netbsd-hp700/7.0.1/netinstall.lif\
Another key item to bear in mind was that I set the download file to 'netinstall.lif', not 'SYSNBSD' as the documentation suggested. It appears that the .LIF file includes the bootloader, the installation kernel, AND the installation tools (sysinst, etc.). When using the SYSNBSD file, it loads but then wants to use NFS to load additional installation code, so it is much easier to just use BOOTP and TFTP with the '.LIF' file.
Copyright © 2006-2025 Jared Blaser. All rights reserved.