ORBiT Software
Sponsor Message


Hidden Value details commands and procedures in MPE (and some in Vesoft’s MPEX) that can improve your productivity with HP 3000 systems. Get a free NewsWire HP 3000 Always Online cap — submit your MPE tip directly to us here at the NewsWire. Send your tips to editor@3000newswire.com, or fax them to 512.331.3807.

Edited by John Burke

I would like to know if there is a way to “erase” the REDO stack. For security reasons I would like to prevent anyone from seeing sensitive commands that had been issued.

Mike Hornsby, Bruce Toback, Joseph Dolliver and Tony Newton all reply:
Either :setvar hpredosize,0 or :setvar hpredosize,1followed by :setvar hpredosize,n where “n” is the size you normally use, will do the trick.
Obviously this could be set up as a UDC or command file, if it is something you do often.

We are using virtual hosts with Apache and would like to add additional security based on IP addresses for each of the virtual hosts. How can we do this?

Mark Bixby replies:
You want to put order/deny/allow in either “.htaccess” files, or in the main configuration files inside <Directory> definitions. You can find out more at www.apache.org/docs/mod/mod_access.html

While trying to do a setcatalog for a newudc, I came across the error “COMMAND.PUB.SYS is full”. How can I fix this?

Mike Hornsby replies:
Logon as manager.sys and type the command :listf command.pub.sys,2
Note the limit of the file so that you can make it bigger in the build command. For example:
:buldacct @%NODIRS
:rename command,commandx
:build command;disc=10000
:stream buldjob2
:purge buldjob1
:purge buldjob2

Donna Garverick notes that buldjob2 often contains in-line Pascal code and questions what one is supposed to do if you do not have the Pascal compiler.

John Backus and Doug Werth point out that all the Pascal code is doing is making sure that the various setcatalog jobs that are spawned get fired off singly, one after the other, waiting for the previous one to finish. If you set your system job limit to 1 and stream buldjob2 so that these jobs get run one at a time, you can achieve the Pascal code’s intended purpose. The Pascal code will barf out errors if you do not have Pascal, but since buldjob2 has SETVAR HPAUTOCONT TRUE in it, the job should finish.

Occasionally I have to abort a print job. Turning off the printer does not get rid of what is in the HP 3000 print buffer. Is there a way to empty the buffer so that when I turn the printer back on I don’t waste any paper?

Wirt Atmar replies (paraphrasing from the Edward Fitzgerald translation of the “Rubaiyat of Omar Khayyam”):
No. Once the moving finger writes, not for all your piety nor all your wit, can you erase half a line of it.

Gavin Scott adds:
Fortunately some recent HP LaserJet models (5Si, etc.) finally provide a “Cancel Job” function that you can invoke from the front panel which will drain all the remaining data in the job into the bit bucket for you.

I am trying to print a bit-mapped image created on a PC to a LaserJet 5si using the HP 3000 spooler. So far, I’ve printed out a forest of very interesting characters, but nothing that’s even close to what I intended. One approach I tried was printing the image with the output redirected to a file rather than the 5si, uploaded the file to the 3000 as a pure bytestream file and FCOPYed the print file to a spooled printer. How do I do this?

Doug Werth replies:
FCOPY won’t do it for you. FCOPY is sending the data to the printer with CR/LF every x bytes. What you need is to write a program that will call FWRITE using control option %320. This will write each record without CR/LF, essentially sending your bytestream file as a bytestream of data to the spooler. Furthermore, you could store the file as a fixed binary file rather than bytestream which would make it easier to read in with a “traditional” language like Cobol.

Lars Appel adds:
If the file to be printed is PCL and hefty graphics, you might want to try the RawLP program from Samba/iX package (can also be found as a separate download at the 3k Associates Web site: www.3kassociates.com/other/publicdomain.html as a program in the Samba/iX section.

I created that small beast when “lp -r” did not print as “raw” as expected. Don’t know if that has been fixed by now. Have been using RawLP only (eg /SAMBA/PUB/lib/rawlp myfile.pcl LP).

RawLP is somewhat similar to COPY /B myfile.pcl LPT1: in DOS.

How can I easily purge all the files in a group without destroying the group structure?

If GRPNAME is the name of the group then either:
1. chgroup GRPNAME
purgegroup GRPNAME
2. purge @.GRPNAME

Is it possible to set a password on the remote support modem (ldev 21)? If so, how?

Rick Clark replies:
Go to the console and hit ctrl-b. At the CM> prompt select ER (enable remote access). It will ask if you wish to make any changes. Just answer yes and it will then prompt you for the password. Follow the rest of the directions and you should be okay.

One disadvantage of this is you will be required to keep the console port enabled. If enabled, Predictive Support will not be able to dial out using this modem.

I’m attempting to get our 918 back up and running after months of neglect and a disk crash. I’ve rebooted the system several times and noticed that the time is incorrect. No big deal says I, since I have to do an INSTALL anyway, I’ll just correct the time before I do the restore of the system backup.
So, I INSTALL from an SLT and then when I boot from disk (getting ready to restore all the files and accounts), I change the time from 6:17 PM backwards to 3:15 PM. That was over half an hour ago and the START NORECOVERY is still sitting there, unmoved since I entered the new time. Any ideas on how long I can expect this to take, or how to avoid it in the future?

Tom Madigan replies:
You need to bring up your 918 normally without trying to enter the date/time during bootup. After you get the system back up, make sure that there are no other users on your system, then:

3. :SETCLOCK DATE=m/d/yyyy;time=hh:mm:ss;now
4. Shutdown and reboot.
You should be OK from that point forward. We found out the hard way after we finished up our Y2K testing and had to reset the date. Our panicked call to the Response Center resulted in the above instructions. It appears to happen to the 918 only!

I am having trouble getting OPTION LOGON UDCs to work. Can someone explain how they are supposed to work?

Eric Vistica replies:
There are 3 levels of UDCs: SYSTEM, ACCOUNT and USER. You can have 0, 1 or more files cataloged at each level. Each file can have multiple UDCs that have OPTION LOGON.

But, only the first UDC with OPTION LOGON in the first file at each level will be executed starting with SYSTEM then ACCOUNT then USER. Note that for non-OPTION LOGON (the default), the search sequence for a UDC name is the USER then ACCOUNT then SYSTEM level.

I cannot seem to find the correct printer type to set up a non-HP printer. Is it 22? 24? 26?

Doug Werth replies:
D. None of the above. Use type 18 for non-HP printers for Xon/Xoff flow control only, no status checking.

I have a HP 3000 979KS/100, with 256Mb RAM, and 120 to 150 concurrent sessions during normal operations. 99 percent of these users are running the same application, hitting the same database, via Speedware run-time. I know the average session count is a variable in the formula. Also, I see ODBC coming into the picture in the near future. What is the formula used to determine RAM requirements?

Bill Lancaster replies:
I love this question! The formula for how much memory you should put on the system is:
(How much can you afford?) = (How much you should buy.)

Seriously, from a practical perspective there isn’t a point at which additional memory won’t be more useful. There is a diminishing returns point but it’s not worth much effort to calculate.

You don’t say whether you have much batch processing or not. If you do a lot of serial activity, more memory will definitely make a difference. I would recommend a minimum of 2 GB, though. If HP memory is too expensive, there are great deals to be had in the third-party memory arena.

FTP and Telnet can run under inetd. Why can’t syslog run under inetd? Or can it?

Mark Bixby replies:
The 6.0 implementation of FTP was designed to run under inetd. Inetd listens on the ftp socket for new connections, and then spawns an ftp server process, which runs until the user finishes transferring files and then terminates.

Syslog is designed as a single always running process that listens to its own socket. If you tried to configure it into inetd, inetd would listen to that socket, then spawn a syslog process, which would then try to start listening to the same socket that inetd is listening to, thus producing a socket in use error.
On a busy system with lots of syslog messages, it’s more efficient to be using a single always-running process, rather than having inetd spawning a new process for every message.

I’m revamping the backup/recovery strategy for our site, and I’m wondering if someone knows what the maximum recommended usage or life span is for a DLT7000 tape?

Mark Klein replies:
The following is from our DLT8000 manual, but since it applies to the tape, it should also hold for DLT7000:
“DLTape IV - Shelf Life 30 years. Usage: 1,000,000 passes.”

Denys Beauchemin adds:
A good place to find out this type of information is at www.dlttape.com
I always go with the 15,000 uses for a DLT tape. If you slap it in a properly sized library and carve the library adequately, you should be able to never open the library. 

Copyright The 3000 NewsWire. All rights reserved.