Net.digest summarizes helpful technical discussions on the comp.sys.hp.mpe Internet newsgroup and 3000-L mailing list. Advice here is offered on a best-effort, Good Samaritan basis. Test these concepts for yourself before applying them to your HP 3000s.
Edited by John Burke
Perhaps everyone is just worn out. Or, perhaps everything has been said that can to be said. But for the first time since last November neither the EOL for the HP 3000, nor the HP-Compaq merger dominated the off-topic conversations. In fact, they were barely even mentioned. However, that does not mean that 3000-L lacked those off-topic and wildly off-topic threads that make it so special.
There was a thread discussing why software is so bad and what is being done about it. Another on whether under God should be in the Pledge of Allegiance and, in fact, whether we even need a Pledge of Allegiance. There was also a long thread about H1B visas and some reminiscing about APL on the HP 3000. But my favorite was titled Darwin, Satan & Macintoshes you have to read it for yourself. Even with all of this noise, there were still hundreds of postings where people graciously shared technical information.
As always, I would like to hear from readers of net.digest and Hidden Value. Even negative comments are welcome. You can reach me at firstname.lastname@example.org.
More clarification about IBMs iSeries
Last month I wrote about how IBM was doing some throttling of the iSeries for green screen interactive applications and wondered how this would affect anyone thinking of migrating from the HP 3000 to the iSeries. It turns out that this only refers to the use of the 5250 green screen protocol and not to interactive applications that use telnet, for example.
Think of 5250 green screen as if it was a transaction monitor like CICS or Tuxedo it isnt really, but the analogy works to explain the difference. The bottom line is that unless you want to use the 5250 protocol/monitor, you can use the full power of your iSeries without paying extra. IBM agrees that the use of the term interactive in its literature is misleading, especially to anyone thinking about migrating to the iSeries from the HP 3000.
In Case You Missed It, Part 1
Wirt Atmar wrote that because enough people have not tried the Advanced Telnet patch yet, it still has not gone General Release. In fact, he quoted James Hofmeister: We were hoping to include this in MPE/iX 7.5, but we did not get sufficient customers to test, so I am expecting it will be included in the first beta 7.5 patch. I know, November 14 threw us all for a loop, but we need to help make sure that the last release of MPE has everything people have worked hard to get in it.
Wirt went on in his own inimitable way: If you are currently a user of QCTerm or you would like to get the Cntl+S/Cntl+Q Xon/Xoff feature into MPEs telnet, please request the patch, even if youre only mildly interested. These patches are probably the most thoroughly tested patches in the history of MPE, thus I guarantee that theyll work. In fact, Im so sure about their validity and safety that Ill buy anyone that finds an error a years worth of enchiladas.
The advanced telnet feature that appears in these patches makes an enormous amount of difference if youre telnetting over any distance at all (in fact, its like night and day). But it also makes a difference if your HP 3000 is only a few feet away. Your telnet connection becomes silky smooth.
But wait. Theres more! I believe that this is the first telnet patch that contains the Xon/Xoff flow-control behavior that youve always expected to be there with serial or NS/VT connections. With this patch, its finally in telnet as well.
Contact the RC and request the patches below for SR 4701-422436.
PTDGDC7 for C.60.00
is Beta Test
Serious Linux server on a shoestring budget
For several years now Ive had Linux running on an old P133 PC, mostly as just something to play with occasionally. After November 14, I decided to get serious about Linux. Which meant I needed a more serious machine to run Linux on.
One option was to get a copy of VMWare, something Gavin Scott first introduced to members of 3000-L. VMWare allows you to run multiple virtual machines, simultaneously if you like, on a single PC. In theory at least, you can have several versions of Windows and Linux all co-existing, and all active on one PC. This is particularly useful for anyone who is space challenged. The only trouble with VMWare is it requires a very beefy PC with lots of memory, disc space and CPU power. My primary PC at home is a 450 MHz system with 256 MB RAM and 10 GB disk plenty of power to run Windows 2000 and Office 2000, but hopelessly inadequate to run VMWare.
Of course I could buy a new PC, but then what to do with my current, fully adequate, PC? Besides, I wanted a real server with SCSI disks, hardware RAID, etc. I priced new machines on the Internet, from Dell and others, but as soon as you start talking SCSI and hardware RAID, the prices start escalating rapidly. Then I had an epiphany.
We all get the flyers from used equipment vendors. Until recently, I only looked at the HP 3000 component section. Then one day I started idly perusing the NetServer section. It seems that the confluence of two things has created great deals on machines that were state of the art just two years ago.
First, technology is turning over at an ever-increasing rate and a certain percentage of users just must have the latest and greatest systems. Second, the dot bombs have created a glut of barely used servers. This particular vendor had gotten in a number of E60 NetServers with 550 MHz processors, 128 Mb RAM and a 9 Gb disk. They were asking $200 each! The E60s are dual processor-capable, so I bought two and, for an extra $135, bought a used SCSI RAID controller. I took the processor, RAM and disk from one and created a dual-500 MHz server with 256 MB RAM and mirrored 9 GB disk. As a bonus, Ive now also got a spare power supply, motherboard and video controller all for less than $600 including shipping.
In Case You Missed It, Part 2
Do you make your own CAT 5 patch cables? Do you know how to do it correctly? If not, you may find that your 100 base-t network is only running like 10 base-t. Denis St-Amand provided the following helpful technical tidbit: Real CAT5 cable must have a twisted pair for RJ45 pins 1 and 2, another pair for 3 and 6, another for 4 and 5 and a final pair for 7 and 8. Only pins 1-2-3 and 6 are in fact used.
PURGELINK and The Cheshire File
I could not resist this one. This is exactly how the e-mail to 3000-L was titled. But first some background. Several years ago, Hidden Value and net.digest first extolled the value of the PURGELINK CI command and the Posix shell rm and mv commands to purge or replace open files. The basic idea with PURGELINK and rm is that you could purge an open file (it would immediately appear gone to anyone not already having it open) without affecting current users of the file. When the last user closed it, the file would be gone.
Well, maybe. Here is what the e-mail said: For the last three mornings I have successfully used PURGELINK to remove an opened MPE file only to have it mysteriously reappear the next morning. It seems to be the exact same file since all characteristics are the same even to having the same create date.
Obviously, the question is what happened? Gavin Scott explained it all: No, youre not dreaming, its probably actually happening this way. When you FOPEN (from the MPE point of view) your old permanent file, the system knows that you have an OLD, PERMENENT file open. When you :PURGELINK the file it then becomes a NEW file from the point of view of the MPE program that has it open. Eventually the program will FCLOSE the file, at which point it has a chance to override the default disposition for the file.
In your case it would appear that the process that has the file open is explicitly specifying the SAVE PERMANENT disposition (1) in the FCLOSE call (or the ;SAVE option on a file equation). This causes the file to re-materialize as a permanent file at the point that it is closed. Because its still the same file, it will reappear with all of the original attributes, including the creation timestamp.
How many are too many in a group?
You can get into trouble if you put too many files in a group. This can happen if you have too many files in OUT.HPSPOOL (output spoolfiles) or even in a regular group (you need to be careful how you purge groups of files if there are thousands in a group do it in reverse alphabetical order). In either case, you can have transaction manager (XM) issues that can bring your system to its knees.
How many are too many? It depends. But before you can determine whether you are pushing the envelope, you need to know how many files exist in a group and its directory tree. Unfortunately MPE/iX does not give you any means of counting the number of files in a group. Perhaps you missed it, but Jeff Vance posted the following:
In conjunction with Donna Garverick and Bill Cadier there is a new script on Jazz that counts files (and directories, etc.) that live at all levels under a target node (acct, group, dir, root). This script just does a simple directory tree traversal and counts files (and other objects). See jazz.external.hp.com/src/scripts
By the way, there are really two versions of the script. The file FILECNT is completely stand-alone and is the faster of the two. The file FILECNT2 runs a very simple NM program to avoid CI if nesting limitation.
After several false starts, that staple tool of network administrators, trace route (TRACERT on MPE) finally made it to MPE/iX. If you already have it on your system, TRACERT is located at the group shown below:
It is included in the General Release NST NS-Transport patch for all supported releases 6.0, 6.5 and 7.0.
The NST patch must be installed on both the local and the remote HP 3000 to support Tracert from HP 3000 to HP 3000.
The current General Release NST patches are:
NSTGDB2 for C.70.00
NSTGDB1 for C.65.00
NSTGDB0 for C.60.00
TRACERT is not available for MPE/iX 5.5 and previous.
Tools for moving
Mark Bixby introduced two new tools for those planning or considering a move to HP-UX. One tool maps MPE commands to the closest equivalent HP-UX commands, and the other tool maps MPE intrinsics to the closest equivalent HP-UX function(s). The direct tool links are:
invent9k.external.hp.com/~csy/cmds.cgi is a commands cross-reference. Browsing to invent9k.external.hp.com/~csy/intr.cgi yields an intrinsics cross-reference.
John Burke is the editor of the NewsWires HiddenValue and net.digest columns and has more than 20 years experience managing HP 3000s.
Copyright The 3000 NewsWire. All rights reserved.