RAC Consulting Sponsor Message


Hidden Value details commands and procedures in MPE that can improve your productivity with HP 3000 systems. Send your tips to john@burke-consulting.com.

Edited by John Burke

We have two disk drives from an old HP 9000 we’ve stopped using. We would like to incorporate the two disk drives into the HP e3000 969/220 that we currently have. The two discs are the same model as the ones that we currently use on the 3000. We would like to add these discs to one of our user volume sets. The first SCSI slot on the back of the 3000 is full, but the second SCSI slot is available. How do we do this?

Stan Sieler replies:
Add them on the second SCSI chain. At some point you ought to look at how the disk drive I/Os are distributed, and consider balancing them by moving disk drives from one SCSI card to another. That assumes that both SCSI cards are of the same kind (e.g., both Single-Ended, or both Fast/Wide.

With the exception of LDEV 1, MPE generally doesn’t care which LDEV a given volume appears at, so you could swap the disk drive that is volume W of set X on SCSI card A at SCSI ID Y with the disk drive that is volume K of set L on SCSI card B at SCSI ID Z, where X/X/Y/K/L/Z are pretty much independent of each other, being sure to change the two SCSI ids to each other’s values.

I am getting an IO status FFDD0071 on a KSAM file. The file can’t be purged, renamed or restored over. It reports a sector number OX1A7740. How can I delete that sector/file in order to restore the file from a backup?

Larry Barnes, John Hornberger and Jeff Vance reply:
There is the ‘rm’ command in the POSIX shell and the PURGELINK command in the CI. Unlike the CI’s PURGE command, these two commands do not open the target file.

I’m trying to get an old 9x7 back up and running and it won’t boot. Booting from the primary path yields the following:
Booting ...
Entry_Test Status = 0
Failed to Initialize
Entry_Init Status = -4

Gavin Scott and Rich Trapp reply:
Check to see if you have a SCSI terminator in place. [Editor’s note: this was in fact the problem, no SCSI terminator.]

I’m currently showing a high level of locks in Glance. Several sessions are showing impeded. What’s the best way to drill down to see which session/process actually has what DB locked?

Ron Horner, Craig Lalley and Gavin Scott reply:
Use DBUTIL and the “SHOW <database> locks” command. DBUTIL has allowed fully qualified / non-creator database names for most commands like SHOW (if you have SM capability) since at least 6.0. So MANAGER.SYS can monitor any database on the system and there’s no longer a need to log on as the database creator for this sort of thing.

I know that there are some undocumented OS variables for Turbostore. How I can find them?

Jon Backus and Greg Stigers reply:
Create job streams that log on, issue a ‘showvar’, then do STORE, RESTORE or VSTORE and then issue another ‘showvar’. The additional variables in the second ‘showvar’ are the ones created by the TurboStore.

Logged on as MANAGER.SYS, I can get the User and Account passwords, but I do not know how to get the group password. What is the command?

Ron Horner and Tony Newton reply:
If you have AM capability, then for any group in the account or if you have SM capability, then for any group on the system, LISTGROUP groupname.accountname;PASS

I have a client that is having trouble getting a LaserJet 8150 to print. The device is several networks/firewalls away. We have:
• configured the device in IOCONFIG
• added the entry to NPCONFIG
• netcontrol update;net=lan1
• issued a startspool 663 to get it going
• installed the latest 7.0 patches
If I send a report to it, the file stays in the PRINT state. For what it’s worth, we can telnet into the printer.

John Burke and Chris Bartram reply:
The HP 3000 spooler uses SNMP to obtain some printer status information from Jetdirect-connected network printers. Ensure that your firewalls let SNMP/SNMPTRAP traffic through between the 3000 and the printer (TCP/UDP ports 161 & 162).

If you cannot do that, or even if you can for that matter, add the “snmp_enabled=false” directive to npconfig for this printer. Then do a stopspool/startspool on the printer.

By the way, the netcontrol update... step was not needed.

I have a CI variable that is a numeric string (like “2000”) and I would like to set another CI variable to be a numeric string that is one less (“1999”). How can I perform math on the first variable? Is there a simple way (that I’m just not remembering) to convert a string variable to a number, and then back again?

Jeff Vance, John Clogg and Paul Christidis reply:
setvar x decimal(!y-1) — or — setvar x “![!y-1]”

I’m trying to get a list of all the NMPROG files in an account, but “:listfile” does not seem to work as advertised:
listfile @.@,2;seleq=[code=nmprog]
Expected a SELEQ value of FTYPE, OBJECT, ACCESS, FCODE. (CIERR 104)

Gary Jackson, John Burke and Mark Boyd reply:
An incorrect error message is confusing the situation. There is nothing wrong with “code”; i.e., it is not supposed to be “fcode.” The problem is with “nmprog”, which should be “nmprg”.

I have an old Jamaica cabinet with some SCSI differential disk drives in it. Can I use this on my Intel-based server?

Denys Beauchemin replies:
You sure can. You will need to get a HVD controller. Look for an Adaptec AHA2944UW. They have it for $400. You can find them as pre-owned devices at other sites.

I recently upgraded our HP 3000 969KS/400 from MPE 6.0pp2 to 7.0 Express 1. I followed the instructions in the Software Maintenance Manual, and all the steps there went okay. However, when I issue the SHOWME command, I get
RELEASE: C.70.00 MPE/iX HP31900 C.39.06 USER VERSION: C.70.00
It appears not to know that I’ve put the Express 1 patches on. Our N-Class box, which came with 7.0 Express 1 pre-loaded reports
RELEASE: C.70.01 MPE/iX HP31900 C.39.06 USER VERSION: C.70.01
Furthermore, when I print the first few lines of HPSWINFO.PUB.SYS, I get the following:

***** MPE/iX MIT C.70.00 *****
**POWERPATCH C7001 P6 **
**THU, MAR 1, 2001, 5:02 PM**

While on our N-Class box with 7.0 Express 1 preinstalled:

***** MPE/iX MIT C.70.01 *****
**POWERPATCH C7001 P6 **
**TUE, MAY 22, 2001, 12:47 PM**

So, even though I followed all the steps, and they all looked like they ran properly, it doesn’t appear that I got the Express 1 installed properly. Is there any other way for me to tell? If the install didn’t go correctly, can I just reinstall the Express 1 as a PowerPatch, or do I need to do the complete reinstall process again.

John Burke, Michael Berkowitz and Gilles Schipper reply:
You do have Express 1 installed. The key line in each HPSWINFO is POWERPATCH C7001 P6. Also, if you look towards the middle of HPSWINFO you will see a header for Express 1 and all the patches added.
You probably used only AUTOINST to do it, instead of AUTOINST to install the FOS and Patch/iX to install the PowerPatch. Had you used Patch/iX during the PowerPatch installation phase, you would have seen the USER VERSION reflect PowerPatch 1.
Factory pre-loaded systems are different animals altogether and typically do reflect the latest PowerPatch level in USER VERSION.

John Wolff adds:
Should you like, you can set the USER VERSION yourself after the fact to 70.01 as follows:

Use SYSGEN and select: MISC
Enter the command: system userversion=C.70.01
Check it with: SHOW
Keep it with: HOLD
Save the change with: KEEP
Make a new CSLT with: TAPE

Now do an UPDATE from the new CSLT. The PowerPatch level will now be reflected in SHOWME.

Copyright The 3000 NewsWire. All rights reserved.