Hidden Value May 2001

Hidden Value details commands and procedures in MPE that can improve your productivity with HP 3000 systems. Send your tips to john.burke@paccoast.com, or fax them to 512.331.3807

Edited by John Burke

We have a system to which we want to add a second 100Base-T card. Is a license required for the second card, or does one license cover both?

Chris Goodey replies:

Only one software license is required.

Where do I find a list of the HP 3000 models that support this 100Base-T network card?

Chris Gauthier replies:

I can tell you what models support the 100Base-T interface: All 9x7, 9x9 and 99x models; all 9x8 models (except that old 948 and 958, because they are CIO machines). The new A and N-Class systems come with 100Base-T standard.

I am trying to create a UDC that will prevent a user from having more than one open session. I am keeping track of who logs on by creating a file each time the user logs on. If the file exists, it is assumed that the user is logged on. When the user logs off, I want to delete the file. To do that, I redefined the BYE command in the UDC. If the user is already logged in, I want to log the second session off without deleting the file. When I issue the bye command it is calling the UDC BYE and not the system command BYE. The UDC file looks like:
.
Logcount
option logon
.
.
.
***
bye
.
.
.
bye
***
John Burke replies:

Change the order of the two UDCs LOGCNT and BYE in your UDC file. This should accomplish what you want since the UDC commands are processed sequentially; i.e. when you execute the LOGCNT UDC, the command interpreter will only look forward of the current position in the UDC file.

Can an HP 3000 NIC be configured to have multiple IP addresses?

Mark Bixby replies:

No, this cannot be done today on an e3000.

Doug Becker adds:
There must be a LANIC for each IP address on the HP e3000. Additionally, if you have more that one LANIC, they must be attached to different subnets.

During a move, our system fell off the back of the moving truck, damaging it beyond repair. We’re obtaining a replacement, but it is not the exact same model, though in the same class (9x7). I have an SLT and full backup. How should I proceed?

Gary Paveza replies:

When we do this for our disaster recovery, we just install from SLT and reconfigure the I/O. That way, all the software (subsys/patches/etc) is on the new system. Then it’s simply a matter of restoring your data.

Paul Courry adds:
Call your third-party software vendors, given them the new HPCPUNAME and HPSUSAN number and get them to generate new unlock codes.

I want to run a batch job that executes a Posix program. Actually, I want to update my aliases file in my SENDMAIL account. I got the other commands figured out but not the newaliases command. Part of my job is as follows:
:xeq sh.hpbin.sys “/SENDMAIL/PUB/bin/newaliases”
/SENDMAIL/PUB/bin/newaliases 149: : not found
/SENDMAIL/PUB/bin/newaliases 151: : not found

Mark Bixby replies:

:xeq sh.hpbin.sys “/SENDMAIL/PUB/bin/newaliases”
tells the shell that newaliases should be interpreted as a script file. But it’s a symlink to an NMPRG, so you see all of those weird error lines as the shell tries to parse a binary NMPRG as if it were a script.

You want to be doing it this way instead:

:xeq sh.hpbin.sys “-c /SENDMAIL/PUB/bin/newaliases”

I am trying to get Apache running. I ended up using telnet from my Linux box to the HP 3000 and then trying to use VI (yeck) to edit files. It looks like VI is sending HP terminal control stuff to my telnet client, which just gets a little boggled. Is there a better way to proceed?

Michael Anderson replies:

If you’re not using an HP Terminal Emulator than you probably need to set your termtype to ‘VT100’, for example,

shell/iX> export TERM=vt100

John Burke adds:
If you have a Windows PC, then I always say that before even thinking about setting up Apache, people should set up Samba. This way you can use your favorite PC editor to set up your Apache Web server. Setting up basic Samba access requires editing only one bytestream file. You could then FTP it to any system, edit it there and FTP it back to your 3000.

On an HP e3000, what is the relationship between the hardware and software clocks? Is the software clock a stored offset to the hardware clock? Or are the hardware clock and software clock reset at reboot, with the hardware clock running off a clock/calendar chip and the software clock being calculated based off CPU ticks, or is it something else?

John Clogg replies:

The hardware clock is supposed to be set to UCT (AKA GMT), and the “software” time is derived using a time zone offset that is manipulated by the SETCLOCK command. It is not unusual to see systems where this relationship isn’t understood, where the hardware clock has never been set correctly, so the offset is hundreds of hours! The SHOWCLOCK command will show you the current status of the hardware and software clocks. In general, the SETCLOCK command adjusts the offset, and the date and time setup during a system start adjusts the hardware clock.

Stan Sieler adds:
The free FIXCLOCK command file at www.allegro.com/software/hp3000/allegro.html can help you fix the hardware clock date/time.

How do I create a basic Samba share for a user to access his account?

Lars Appel replies:

Something in the line of ...
[myshare]
comment = Access to user’s account
path = /ACCOUNT
guest ok = no
write ok = yes
user = USER.ACCOUNT
only user = yes

I want to restore the directory structure of an existing account but to a new account name. I tried restore;/OLDACCT/;account=/NEWACCT/;directory;create;keep but got /NEWACCT/GROUP/FILE NOT RESTORED: HPFOPEN RAW FAILED Intrinsic layer; the operation could not be completed because a nonexistent account was specified. How do I do this restore?

Jonathan Backus replies:

I usually just use the BULDACCT command and then edit the BULDJOB1 and BULDJOB2 stream files to change the account name to the new account and then stream them.
[Editor’s note: BULDACCT %HELP will show you how to use BULDACCT.]

Doug Werth explains the error message, and offers a word of caution when using the “directory” option:

/NEWACCT is an invalid account name. You don’t want to use HFS syntax to rename the account.
You are probably lucky that it failed. Don’t forget that the ;DIRECTORY option restores the entire directory for the system, not just the directory associated with the fileset to be restored. More than one system admin has accidentally overwritten current passwords with previous ones by using the ;DIRECTORY option to restore files from an old tape.

How do I reset the account CPU-second and CPU-minutes that appear in a REPORT @.@?

James Reynolds and Denys Beauchemin reply:
Reset account is RESETACCT. See HELP RESETACCT for usage details.

I was scanning a $stdlist for reply messages and came across the following message:
2:46/#J9115/2404/Are standard forms on OALASER (Y/N/R)? (MAX CHARS.=3)?
What does the “R” option mean when you reply to a message?

Kristian Thisted replies:

R = Repeat. It prints one more line for you to check the alignment.

How do you repair a damaged KSAM file?

Ronald Horner and Neil Harvey reply:
Run KSAMUTIL, then KEYINFO fileref ;
recover

I would like to erase all data from two disk drives (LDEVs 3 and 4) I have in a test volume set and add one to the MPEXL_SYSTEM_VOLUME_SET and the other to another user volume set. How would I go about this? Can it be done without disrupting the system?

John Burke replies:

Optionally (personally, I think it is a good idea), remove any groups and accounts on test_volume_set, then

1. vsclose test_volume_set
2. enter volutil
3. scratchvol 3
4. scratchvol 4
5. newvol mpexl_system_volume_set:member3 3 100 100
6. newvol appl_volume_set:member7 4 100 0
7. exit volutil

That’s all there is to it. And it absolutely can be done without disrupting the system.
Note: this is what you want to try to do with new disk drives — put them in a test volume set and test them for a while before moving them into production.

Is there a way in to purge all $STDLIST files without sending it to tape first?

Doug Werth and John Clogg reply:
To be safe use this command first to see if it is really going to delete the files you want.

:spoolf @;seleq=[filedes=$stdlist];show

Then add a ;delete to the command.

:spoolf @;seleq=[filedes=$stdlist];show;delete

Sometimes when I do “xeq sh.hpbin.sys -L”, when it is executing the file /etc/profile I get a message “resource busy, try again later”. I also get this message in error_log sometimes when the APACHE browser tries to fire up a CGI program (c-language, in MPE space) and sometimes when I get the shell up and it does get through the /etc/profile file I can still get the “resource busy...” message. What resource can this be?

Mark Bixby replies:

The most likely cause is insufficient contiguous free disk space which results in fork() failures. The fix is to defragment your free disk space, either by using :VOLUTIL CONTIGVOL, or a third-party product like the ones from Bradmark or Lund.

Is there a Web site that indicates what the bare minimums are for running on 6.5?

Denis St-Amand replies:

Go to mpeix.hp.com/operating/mpeix65/specs.html, where it says:

The 6.5 release will use slightly more memory on a system. HP believes this should not be noticeable on systems with at least 128 Mb of main memory. If your system has less than 128 Mb, you may want to consider adding memory before you update to the 6.5 release. This may not be necessary for development or test systems with just a few users. HP recommends that production systems should be configured with at least 128 Mb of main memory for optimal performance on the 6.5 release.

For the first time ever we have PCs using NS/VT connect. We are seeing some error messages that I haven’t seen before from NS/3000.

** NS/3000 INTERNAL ERROR IN VT; Job: 0; PIN: 2579 Info: 0
- Error: 12; Error Reported by VT
- VT error : 39; REMOTE NOT ACCEPTING SERVICE REQUESTS

We are on MPE/iX 6.5 PowerPatch 2. Why the errors?

John Burke and John Clogg reply:
Try nscontrol status=summary and look at the number of VTSERVERS versus number of active. This may be you problem. To change the maximum number to something higher, use

NSCONTROL SERVER=VTSERVER,0,new_value

The default is 300 and this value can be changed dynamically with the above command (many people include this in their system startup job that starts NS/3000).