Y2K info for IBM AS/400???greenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread
Just got off the phone with an old friend, working on an AS/400. I've played with lots of computers over the years, but never an AS/400. He says they're having few Y2K problems. This is a mid-ramge IBM system. I know there's a bunch of 'em out there. Anybody got any info? I think the older System/38 was similar???
-- Sysman (firstname.lastname@example.org), February 06, 1999
There's something like 300-400,000 AS/400s sold, but I don't know how many are still in use.
The current operating system is in pretty good shape if the client has kept pace with the upgrades. Many if not most apps on the AS/400 are hand-written, and lots of those were migrated up from the older System 36 and 38 platforms. This "ease of migration" was one of the many selling points for the IBM mini-computer series.
The issues are much the same as with mainframe app software. Old, patched-up stuff with atrocious database design and 2-digit years are commonly out there. Much of it was written in huge reams of poorly documented spaghetti COBOL and RPG code, and little of it was ever tested for roll-over. Since the mini platform is a bit more recent than the mainframe, the mini code, in general, is a bit less sclerotic. But on the other hand, the mini code is frequently much more ad hoc compared to equivalent software designed for a mainframe system.
Some of the AS/400 stuff was replaced over the last few years with app suites like J.D.Edwards financials, but much of it remains as it was written in the 70s and 80s. Frequently, it is your larger small businesses and the medium sized companies that use the AS/400, but some of the larger enterprises also use these at the deparmental level and may have dozens of AS/400s in use.
-- Nathan (email@example.com), February 06, 1999.
Nathan hit the major points about this system. The System/3X series are not Y2K compliant and basically aren't supported by IBM anymore. Its my understanding they can't be upgraded. I would re-emphasize Nathan's points about the mishmash application systems. The few systems of this ilk that I have direct knowledge of were virtually undocumented and had large portions of source code missing.
Sysman, are you knowledgeable about the old PDP-11 systems? These were very common 10 years ago. They have lots of Y2K issues regarding the operating system RSX. There is a "compliant" version around, but it uses a very awkward translational date post 2000. Some of the nuke plants still use the PDP-11. The one I know about is the New Hampshire plant. Another issue is the PDP-11 instruction set was emulated onto a number of embedded systems. A lot of engineering types grew up (in college) on the PDP-11 so when they designed systems, they may have had a tendency to use this set. I have no idea how wide spread this is, but it could be a hidden major headache.
Other issues include C language use of tm_year and the IBM OS SVC-11 problem. You might want to do a search on CSY2k ng for Cory Hamasaki's views on these items. I have visions of some companies application code being remediated and the mainframes crashing for systems internals reasons. Any company that doesn't fix their VSAM libs are toast no matter what the application people do.
Is your company using a full Time machine test?
-- RD. ->H (firstname.lastname@example.org), February 06, 1999.
The AS/400 is probably the machine of choice for most Japanese and Asian factory level accounting/production-inventory control systems. IBM Japan just released (Jan, 1999!!) a date-search software package. AS/400's are so prevalent that companies in the remediation business are setting a fixed price of approximately US$10,000 per machine(any language) and a 36 month payment plan.
-- PNG (email@example.com), February 06, 1999.
Thanks for the info! So I guess this platform is nothing exceptional. Re. PDP-11, believe it or not, we still have 2 running as "embedded controllers" in 2 of our ancient photo-typesetters at work! Luckily, no date problems, as far as I know, etc. etc. Some things just never seem go away... <:)= PS - I'll pass this link to my friend, Thanx again...
-- Sysman (firstname.lastname@example.org), February 06, 1999.
Sysman, Yikes! Those PDP-11 "controllers" are going to fail on rollover if you they use the current date. Unless you are using Mentec's latest version of the RSX operating system and the application software can handle a kludgy date translation, those PDP's are going to be useless. Perhaps for your application, the date isn't important. If so, just turn the clock back to 1980. The following is a link to Mentec.
-- RD. ->H (email@example.com), February 07, 1999.
Thanks for your concern RD. We're not sure what "operating system" the old boxes are running - something written by RCA I think. But we do have have a trusted engineer that assures us we'll have no Y2K problem here. Just need to figure out all this other stuff around! <:)=
-- Sysman (firstname.lastname@example.org), February 07, 1999.
Thanks for your concern RD. We're not sure what "operating system" the oldboxes are running - something written by RCA I think. But we do have have a trusted engineer that assures us we'll have no Y2K problem here. Just need to figure out all this other stuff around! <:)=
You do realize this is exactly the situation thats going to probably kill a fair number of people. Imagine this scenario. An ancient PDP-11 sits in an ignored corner of a chemical plant controlling the movements of various fluids into reaction chambers and distillation vessels. The computer was installed 15 years ago and has worked flawlessly for so long it is taken for granted. Once the initial reaction vessels are filled, and someone hits the Start button, the PDP does the rest. For the half-hearted Y2K assessment 6 months past, a "trusted engineer" told middle management that this old system didn't have Y2K problems because it didn't use dates in calculations.. At one tick past midnight on Rollover the PDP freezes all reaction valves in their current state as the operating system goes catatonic. A tremendous toxic explosion happens 10 minutes later. The noxious cloud covers a poor working class neighborhood a half mile away. Fifteen hundred people die and 5000 are made seriously ill.Can't happen? Over half a million PDP-11's were manufactured. The instruction set was ported to thousands (tens of thousands?) of other embedded systems. Many of these are still up and operational around the world.
To my knowledge, outside of one off experimental operating systems, there is exactly one Y2K compliant PDP-11 operating system in the world and its not made by RCA. All others fail on Rollover.
As an aside, why would you trust an engineer who is ignorant of exactly what operating system is in use? As I said before, if this system is keeping a current date as system date, it will fail.
-- RD. ->H (email@example.com), February 08, 1999.
The code jockeys on my projects usually get very annoyed with me during the test phase.
They'll say something like, "Looks like it works..." and I'll ask 'em, "Have you finished cranking the test plan?" They'll invariably say, "No..." and I say (with them saying it right along with me), "Then it doesn't work yet." They give me The Look and go back to testing.
Funny thing is, I flat out hate doing QA myself, but I know deep in my marrow how vital it is to a successful project. "If it hasn't been tested, you don't know if it works."
Software fails in weird and wonderous ways...
-- Mac (firstname.lastname@example.org), February 08, 1999.