RAC Consulting Sponsor Message

July 2002

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

I just upgraded a 9x9 from 6.0pp2 to 7.0 express 1, and can’t get Apache to run. The error_log file has dozens of messages like this:

[Wed May 29 19:17:50 2002] [alert] (22)Invalid argument: setuid: unable to change to uid: -1
Am I missing something with the uid/gid of the Apache account?

Rick Gilligan and Mark Bixby reply:
As of (about) Apache 1.3.9, the User and Group directives in httpd.conf became important on MPE. Please see /APACHE/PUB/conf/httpd.conf.sample for the correct values to use in your real httpd.conf file.

As a general rule after Apache upgrades, you want to examine the /APACHE/PUB/conf/*.sample files and propagate any changes to your real config files. Or use the POSIX diff command to compare your real files to the sample files to see what is different and then decide on a case-by-case basis.

We have just recovered one of our systems from a disk failure. However the permissions etc for all accounts were lost on the restore. Is it possible to store accounts, groups and users and keep all capabilities or permissions? What is the best way to store the system and then restore this backup onto a blank system with the correct account, group and user capabilities?

Jonathan Backus replies:
It depends on how you do your backup. Check to see if you use the DIRECTORY option. Please note the special criteria if you have user volumes. If you use this option on your backup then you can use it on your restore. Another thing I often do is setup the backup job to use the BULDACCT utility to create the BULDJOB1 and BULDJOB2 job streams right before the “full volume” backup. I then make sure to purge the jobs off the system right after the backup is complete because they contain all of the account, group, and user passwords. Naturally that doesn’t help you at this point, just future reference.

[Editor’s note: It is absolutely critical you use either the DIRECTORY option or BULDACCT or both in your backup procedures — and use them correctly. The DIRECTORY option works great, but you have to take special care if you use user volumes — and you should be.]

When I build a file I specify the record length in bytes. But the display in LISTF is in WORDS (it says “128W FB” under size and TYP in the LISTF). Here’s the command I’m using:

:LISTF TESTFILE,2
Non-existent file. (CIERR 907)
:BUILD TESTFILE;REC=-256

I realize that 256 bytes equals 128 words; however, my programmers see a difference, and wish to avoid any differences in some test they are running. So how do I control either the BUILD or the LISTF commands to be consistent?

Ray Shahan and Patrick Thrapp reply:
You need to be more specific when using the rec= parameter. Try BUILD TESTFILE;REC=-256,,F,ASCII. Then the LISTF should display: TESTFILE 256B FA ...

On my MPE/iX 6.5 systems (non-A/non-N class), can I use a DLT8000 to create a SLT tape and boot from it?

Gilles Schipper replies:
I believe the answer is no. The reason is that the DLT8000 attaches only to a FWSCSI interface. On a non-A-Class and non-N-Class HP 3000, one can boot from a DLT drive as long as it is a SE SCSI connected drive. This can only be a DLT4000. Note that you can create a SLT, but you might rightly ask what’s the point? (There are also DLT4000’s that can connect to a FW SCSI interface, but you cannot boot from those on a non-a/non-n class machine either.)

You didn’t ask, but I will offer that you can boot from a DLT8000 on a A/N-Class HP 3000. (I suspect the same is true of a DLT7000 — but I cannot say that from personal experience.)

What is the maximum amount a DLT7000 will back up compressed?

Denys Beauchemin replies:
The DLT IV cartridge is supposed to hold up to 70GB @ 10MB/second, depending on the compression. The uncompressed amount is 35Gb @ 5Mb/second. Go from there.
[Editor’s note: On a DLT4000 the numbers are 40Gb /20 Gb and on a DLT8000 the numbers are 80Gb/40Gb.]

I want to find a ..gz file to FTP to a client. However, a listf doesn’t work. What syntax should I use to find a file with this suffix?

Michael Anderson replies:
:LISTFILE /,2;name=@.gz ought to do the trick. Or, you could just use the FINDFILE UDC that is contained in the UDC file HPPXUDC.PUB.SYS shipped with all systems. Or, in the shell, do the command find / -name *.gz -print

How do I deal with FTP and filenames containing spaces?

Elizabeth Thayer replies:
If you are on the current FTP patch, you can do a “Posix on” and be able to have Unix type filenames. It is documented in the ftpdocs.arpa.sys. When done you will need to turn Posix off.

We’ve just brought up a 979/400 with MPE/iX 7.0 and PowerPatch 1. The self-test light on the Core IO card is flashing (Xcvr/Net Fail) and we cannot communicate with our DTC’s (DTC48’s). When we recycle power on the DTC’s the LED shows “F2.” with the “.” flashing. Then the “C1.” displays with the “.” flashing. Does the core IO card have a switch that tells it whether to use the BNC or the AUI connection?

A number of people reply:
Yes, there is a jumper on the Core IO card.

How do I know when a disk in a Model 10 array, or on any mirrored type of arrangement, goes bad? Is there a message on the console? If so, does it repeat? How do you know when to call for service?

Andreas Schmidt replies:
Model 10/20/30 (H/W mirroring) has only the “traffic lights.” Especially having configured a “hot spare” you will only realize any defect in checking the lights. For S/W Mirroring (MirroredDisk/iX) you will get console messages.

What is the latest stable PowerPatch for 7.0?

Gary Jackson replies:
There is only 1 PP (Express 1) but there are many additional patches (like about 50) that should be downloaded. I have been told that there will probably not be a PP 2 because most resources have been assigned to get MPE/iX 7.5 ready.

How can I find all IMAGE databases on my system?

Tad Bochan replies:
The command listfile @.@.@,6;seleq=[code=-400] will list the root file names of all IMAGE databases on the machine.

It seems that when the cursor on the HP 3000 reaches the last column of the screen, the terminal itself automatically enters a Carriage Return. Is this the case?

Wirt Atmar and Eric Vistica reply:
Not exactly. The cursor does move down one line and to the far left (due to EOL-wrap) but no CR/LF is transmitted. You can set the behavior using Strap C (Inhibit end-of-line wrap), using the escape sequence: Esc&s<x>C, where <x> is 0 or 1, indicating whether you wish InhEolWrp to be active or not.