net.digest May 2002

May 2002

Net.digest summarizes helpful technical discussions on the HP 3000 Internet newsgroup and 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

It may have been an aberration, but last month the number of off-topic postings to 3000-L, particularly rants, was way down, and the technical content was way up. On the off-topic front, Gavin Scott posted this about the “Unix Haters Handbook:” “This classic work (which has been out of print for some time) is now available for free as a single 3.5Mb PDF file from research.microsoft.com/~daniel/unix-haters.html. It is a humorous and surprisingly useful guide to the wonders of Unix operating systems.” Those of you considering a migration to HP-UX might want to check it out.

April’s postings also noted the 50th anniversary of the publication of Watson and Crick’s seminal paper on the identification of DNA as the molecule that carries the information of life, the source of essentially all heredity. On a more sobering note, the list also noted in April the death of both Adam Osborne, pioneer of the portable PC, and Edgar Codd, database pioneer and recipient of the A.M. Turning Award in 1981.

I always like to hear from readers of net.digest and Hidden Value. Even negative comments are welcome. If you think I’m full of it or goofed, or a horse’s behind, let me know. If something from these columns helped you, let me know. If you’ve got an idea for something you think I missed, let me know. If you spot something on 3000-L and would like someone to elaborate on what was discussed, let me know. Are you seeing a pattern here? You can reach me at john@burke-consulting.com.

News of its death was premature

This from Denys Beauchemin, of HICOMP and a member of the Interex Board of Directors: “HP has just announced the release of the latest generation DDS drive, the DAT72, contradicting its announcement that DDS was dead just a couple of years ago. The last version of DDS was DDS-4 initially released four years ago. The DDS-5 drive seems to have the same throughput but somewhat higher data density compared to the DDS-4. The capacity is 36Gb native, 72Gb with 2:1 compression.”

“The transfer rate remains the same as DDS-4. DAT72 uses a longer tape, 170 meters versus 150 meters for DDS-4, in order to accommodate some of the extra storage. They are also presenting a roadmap depicting the sixth and seventh generation of DDS drive in 2005 and 2007 respectively with rapidly increasing capacities if not throughput.”

Christian Lheureux, networks manager of Appic R.H. and a member of the OpenMPE board of directors, put together a DDS compatibility matrix that shows what each type of DDS drive can read/write. If you email him he will send you a copy, or you can get it at.burke-consulting.com/DDS_Comp_Matrix_2003_04.xls

Ah, what might have been?

“Has anyone done any work at possibly porting PHP to MPE?” This question was asked recently on 3000-L. The answer is yes, but unfortunately most of the significant work was done around the time HP decided it did not want MPE in its future and news of the development effort got lost in the resulting noise and turmoil. Campbell Feathers of the Australian company CANNEX (www.cannex.com.au) brought everyone up to date:

“It’s not that difficult. See invent3k.external.hp.com/~MGR.FETHERS/. This is now a fairly old version of PHP and Apache, but the same instructions can be used for porting a more recent version. It does need some attention from a porting guru, but it works. Due to (1) lack of time and (2) lack of demand, I haven’t kept the invent3k version up to date. If requested, I could probably make a more recent version available for download.

“Compiled into Apache, performance has been more than adequate on our N4000 for our external customers. Obviously, the CGI (i.e. standalone) version is totally unsuitable for serving Web pages due to MPE program load time. The TurboIMAGE function library that we created has turned out to be very useful. PHP and IMAGE really complement each other well. We now have about 74,000 lines of PHP code, used for mission-critical, highly interactive data retrieval, and also used for some complex iterative annuity calculations.”

I only have limited experience with PHP and Apache but came away very impressed. Coupled with Feathers’ TurboIMAGE extensions, we could have had a real killer tool for the HP 3000. If only…

I learn something new on 3000-L every month

This is kind of geeky, but neat. It is something I could have used a year or two ago when I had a situation just like this poster: “I have been asked to set up a News and Weather capability for our intranet. Access to the intranet is through a gateway but is restricted for the users so I am looking at using RSS feeds with PHP to populate a MySQL database with items that the users can view. I am having difficulty finding much information on how to do this.”

Mark Wonsil supplied the following response: “Here’s a PHP module that consumes RSS feeds, magpierss.sourceforge.net. For those who are asking, “What in the hell is an RSS feed?” check out backend.userland.com/rss.

It is already starting

While Dave Wilde, head of the HP 3000 business group, reiterated again at the recent HP e3000 Solutions Symposium that HP would continue to supply high quality support for the HP 3000 until HP’s end-of-support in December, 2006, the complaints of poor support are already starting to roll in. There are still many good HP 3000 support people at HP, but we all live in fear that their jobs are at risk every day. And, we live in fear of the cost of “high quality” HP support in 2006. Consider these recent support complaint examples posted to the newsgroup:

“We placed a call to HPQ (24x7 4-hour response), got a call ID and then waited for 2.5 hours for a call back only to be told we will have to wait for another engineer call us back ‘shortly.’ So ‘shortly’ turned out to be an additional 2.5 hours. Fortunately, Rene Woc of Adager was able to tell us in the meantime that we needed a specific patch to fix our repack problem. By the time we got the call back we were already up and running and repacking data. Oh, and the engineer who finally called back was not the most knowledgeable apple in the cart but did finally agree with Rene’s fix.”

“In my opinion, staffing at the Response Center for MPE related questions is at frighteningly low levels.”

“Since the middle of 2002 I have tried to use HP support less and less. You can tell the difference between the old ‘award winning service days’ and the present.”

“A customer of mine had problems getting sendmail going on a brand-new N4000 on MPE/iX 7.5. I placed a call to the HPRC and was asked what type of system we had. I told her it was an HP 3000, and was connected to printer support.”

“Are the days of ‘What’s an HP 3000/MPE system?’ back? Unfortunately they are.”

There is something wrong when users have to be trained on how to work with support personnel, instead of support personnel being trained on how to work with users.

I believe that there are now more HP 3000-knowledgeable support people/entities in the third-party arena than there are available from the manufacturer. If I am looking for answers to some HP 3000 issues, I do not contact HP. I come to 3000-L or a third-party support organization. It’s just a fact that you’re now more likely to get good support from third-party vendors.

This has been the case on the hardware side for many years. In the San Francisco Bay Area you can usually find your ex-HP CE working at your favorite third-party hardware support vendor.

By the way, it was “third parties” who found the problem originally referred to above, determined how to detect if you had it, and reported it to HP.

A little gotcha with the CI function “finfo”

You may not realize it, but your MPE system already has a number of symbolic links. One that almost everyone has is /etc/resolv.conf, which points to RESLVCNF.NET.SYS. Many people have taken to using symbolic links in sometimes very creative ways. Unfortunately (my opinion), it turns out that the CI function “finfo” not only does not follow symbolic links as you might expect, but, as someone rather graphically pointed out, “blows chunks” when handed a symbolic link. According to HP’s Jeff Vance, it would be a one-word patch to tell “finfo” to follow symbolic links. Putting my co-Chair of SIGMPE hat on, please let me know if this is something SIGMPE should pursue.

While I’m on the subject of “finfo,” let me point out another gotcha. This one derives from MPE’s sophisticated file management system. All you can ever say with certainty about the location of a file is the volume set it resides on. The file label and extents can physically all reside on different disk drives. Finfo(“filename”, “ldev”) only gives you the location of the file label. “LISTFILE filename,3” will also only show the location of the file label. MPE requires that certain system files be on LDEV 1.

Unfortunately, there is no simple way to determine if a file, including its file label and file extents, resides wholly on a particular disk drive. The DISPLAYEXTENTS command of FSCHECK will show the location of all extents, but it is difficult to interpret. One way to ensure a file including its file label and all extents resides on a particular disk drive is to store it to disk with STORE and then RESTORE it with the “;DEV=#” parameter.

Ever wonder how the PAUSE command works?

I did. Recently HP’s Jeff Vance provided a description: The PAUSE command is implemented by polling the JMAT (table that the SHOWJOB command scans), not based on receiving a signal from the terminating job, or some other technique. The JMAT polling algorithm starts out with short sleeps before re-scanning the JMAT, but each subsequent JMAT poll occurs after a longer and longer sleep interval. (Note: the sleep interval is reset when the job ID or job state changes. Assuming you are pausing for a single job then only the state could change.) The longer the PAUSE command is “sleeping” for a particular job (or session), the longer each real sleep interval will be.

John Burke is the editor of the NewsWire’s HiddenValue and net.digest columns and has more than 20 years’ experience managing HP 3000s.