1 INFO-VAX	Sun, 12 Nov 2000	Volume 2000 : Issue 633       Contents:F Re: %EDT-F-INTERERR, Internal software error; contact a Compaq supportF Re: %EDT-F-INTERERR, Internal software error; contact a Compaq support0 Re: ? Tape Device owned by dead process on Alpha0 Re: ? Tape Device owned by dead process on AlphaE Re: AS 255/233, which graphics card do I have without opening the box ! Re: DECnet 5 -> 4 routing problem ! Re: DECnet 5 -> 4 routing problem  f$search improvement ? Re: Gartner and OpenVMS . Re: Severity of eBay's ecache problem detailed# Re: SPEC_int92? SPECrate_fp95? etc.  RE: Sun "uptime" belly laugher Re: Sun "uptime" belly laugher( Re: Sun UE10000 is *NOT* an HPC platform( Re: Sun UE10000 is *NOT* an HPC platform Re: Technology of US elections Re: Technology of US elections Re: Technology of US elections Re: Technology of US elections Re: Technology of US elections Re: Technology of US elections	 Test only # Re: Where's the Y2K compliance page # Re: Where's the Y2K compliance page ! [noise] A Maynard Mill Clock Tour   F ----------------------------------------------------------------------  % Date: Sun, 12 Nov 2000 10:18:12 +0100   From: Paul Sture <paul@sture.ch>O Subject: Re: %EDT-F-INTERERR, Internal software error; contact a Compaq support + Message-ID: <VA.0000016b.3839d395@sture.ch>   > In article <Xns6U5q4MdmT@eisner.decus.org>, Bob Koehler wrote:. > From: koehler@eisner.decus.org (Bob Koehler) > Newsgroups: comp.os.vms Q > Subject: Re: %EDT-F-INTERERR, Internal software error; contact a Compaq support ! > Date: 8 Nov 2000 08:16:50 -0500  > [ > In article <3A093DEC.562A4FDC@bbc.co.uk>, Tim Llewellyn <tim.llewellyn@bbc.co.uk> writes: . > >                                           < > > Edit /sum, well its there on AXP 7.1 and VAX 7.2 anyway. > > L > > I wouldn't have a clue how to use it without consulting the docs though. > >  > J > Mostly usefull for ingesting changes recorded with diff/sum.  Reminds meJ > of the way we used to insert change cards to an on disk card deck image.G > It appears that the internals of CMS and sccs are in part implemented  > using similar concepts.  > Y IIRC it was used for patches to macro sources in V2.n days. I believe that the Gnu folks  K use a similar mechanism for incorporating source changes from contributors.  ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sun, 12 Nov 2000 10:18:13 +0100   From: Paul Sture <paul@sture.ch>O Subject: Re: %EDT-F-INTERERR, Internal software error; contact a Compaq support + Message-ID: <VA.0000016c.3839d562@sture.ch>   C In article <200011111350_MC2-BA75-2C2E@compuserve.com>, Richard B.   Gilbert wrote:4 > From: "Richard B. Gilbert" <DRAGON@compuserve.com> > Newsgroups: comp.os.vms B > Subject: Re: %EDT-F-INTERERR, Internal software error; contact a > Compaq support' > Date: Sat, 11 Nov 2000 13:49:49 -0500  > E >         I believe that SOS was never native VMS code;  I believe it  > ran F > under the RSX emulator.  It was my first VMS editor.  (When I had an  > el-cheapo brand-X terminal).   > G I'm pretty sure EDI was a compatibility code editor too. I didn't miss   it when it disappeared :-)  $ > Message text written by Alan Greig= > >>Someone once told me that SOS originally stood for Son Of  > Standalone. Is > that > >correct?  > ? > Son Of Stopgap it said in the sources. There used to be a VMS 	 > version E > of SOS. Wonder if that still exists anywhere for possible inclusion  > onE > the Freeware CD. There are times on a dumb terminal or with poor VT C > emulation where I know I could do something quicker with SOS than > > trying to use EDT in line mode. Suppose there's always teco. > F > The biggest annoyance with SOS/EDIT was that the editors were almostF > identical (common source which then diverged) but that many commandsF > had changed letters For example on TOPS-10 to save a program withoutF > line numbers and then invoke the last compile class command you gaveC > the command *gsb (go strip nobackup). On TOPS-20 the same command  > was 3 > *gun (go unsequence no backup). Just plain daft.<  >    ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sun, 12 Nov 2000 10:18:14 +0100   From: Paul Sture <paul@sture.ch>9 Subject: Re: ? Tape Device owned by dead process on Alpha + Message-ID: <VA.0000016e.3839d8dd@sture.ch>   @ In article <3A0CEEC4.5083F990@bellsouth.net>, Robert Boyd wrote:* > From: Robert Boyd <rlboyd@bellsouth.net> > Newsgroups: comp.os.vms 7 > Subject: ? Tape Device owned by dead process on Alpha ' > Date: Sat, 11 Nov 2000 02:01:40 -0500  > P > There is a tool available to help clear things up when a tape drive is left inR > an allocated state when a process terminates without getting a proper rundown to# > clear the UCB/etc on OpenVMS VAX.  > R > I'm wondering if anyone knows of such a tool for doing the same thing on Alpha? R > We have a system running VMS V7.1 that occasionally winds up with the tape driveK > allocated to a non-existant process.  Any pointers would be appreciated.  N > Supposedly we shouldn't be having this problem any more, but .... it happens
 > anyway.  > S What tape controllers are you using? If using HSJs, then you _might_ get away with  , configuring the tape as another unit number. ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sun, 12 Nov 2000 10:18:14 +0100   From: Paul Sture <paul@sture.ch>9 Subject: Re: ? Tape Device owned by dead process on Alpha + Message-ID: <VA.0000016f.3839da64@sture.ch>   T In article <009F2F39.861D4DCF@SendSpamHere.ORG>, VAXman- Brian Schenkenberger wrote:? > From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)  > Newsgroups: comp.os.vms ; > Subject: Re: ? Tape Device owned by dead process on Alpha % > Date: Sat, 11 Nov 2000 12:29:00 GMT  > Z > In article <3A0CEEC4.5083F990@bellsouth.net>, Robert Boyd <rlboyd@bellsouth.net> writes:Q > >There is a tool available to help clear things up when a tape drive is left in S > >an allocated state when a process terminates without getting a proper rundown to $ > >clear the UCB/etc on OpenVMS VAX. > > S > >I'm wondering if anyone knows of such a tool for doing the same thing on Alpha?  S > >We have a system running VMS V7.1 that occasionally winds up with the tape drive L > >allocated to a non-existant process.  Any pointers would be appreciated. O > >Supposedly we shouldn't be having this problem any more, but .... it happens  > >anyway.   > K > Watch out!  This dead process might be elected posthumously to the senate  > in Missouri.  ;) >  <snip>  R The world press have had a field day on the election :-) My favourite is this one:  I The columnist Bepe Severgnini wrote in Milan's daily Corriere della Sera: Z "The other night when I went into a restaurant in Santa Monica, there was one president -  Clinton.  J "When I ordered a pizza there was another one - Gore. When I paid the billA there was a third president - Bush. When I walked out on to Ocean E Boulevard there was no president because Bill is now the husband of a  senator from New York."    :-)  ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sun, 12 Nov 2000 10:18:13 +0100   From: Paul Sture <paul@sture.ch>N Subject: Re: AS 255/233, which graphics card do I have without opening the box+ Message-ID: <VA.0000016d.3839d71a@sture.ch>   > In article <3A0C2FA3.41447A2B@bbc.co.uk>, Tim Llewellyn wrote:/ > From: Tim Llewellyn <tim.llewellyn@bbc.co.uk>  > Newsgroups: comp.os.vms L > Subject: AS 255/233, which graphics card do I have without opening the box' > Date: Fri, 10 Nov 2000 17:25:55 +0000  >  >  > Hi > E > I have a AS 255/233 that has recently been reunited with a graphics 
 > console.H > I am sure the monitor worked with DECWindows when I configured the boxF > with VMS 6.2-1 in 1997. Now it is running VMS 7.1 and Motif 1.2-5. I > have a monitorE > connected to the SVGA port. I have a PS/2 mouse and keyboard (Dell)  > connected. > G > The monitor works as a dumb console. However, DECWindows keeps saying E > "No graphics devices are present". Also, and the console mode (>>>)  > there are / > only disks and tapes reported by SHOW DEVICE.  > H > How can I find what graphics card is installed (and in use by VMS as a
 > console)8 > without opening the box? Is this just an Open3D issue? > J I'm unsure precisely where Open3D fits in here. From my PWS 600au console 9 prompt, a "show dev" doesn't list a graphics card either.   D What does "show console" say (at the console prompt)? Mine's set to  "graphics" (no quotes).  ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sun, 12 Nov 2000 18:11:48 +0000 + From: "antonio.carlini" <arcarlini@iee.org> * Subject: Re: DECnet 5 -> 4 routing problem' Message-ID: <3A0EDD64.1FA99531@iee.org>    John Santos wrote:  6 > Watching on a terminal, you see it init, then try an6 > ISO8802 load.  If it can't find a load host, it then9 > tries an Ethernet load, followed by a Serial line load. 8 > It then waits a short while and tries again.  (This is9 > what you see happen if you don't have a load host or if 8 > you unplug the ethernet port before it has time to try7 > a load.)  DECnet seems to respond to the ISO8802 load ' > request, but must do something wrong.   : Or the DR250 does something wrong. Actually I am surprised1 that the DR250 attempts a MOPv4 load at all given / its age (maybe they are younger than I thought)   < > So, there is obviously some difference between ISO8802 and9 > Ethernet downloads, but I couldn't find this documented  > anywhere.   : MOPv3 used ethernet downloads, MOPv4 (I think) uses 802.3.= So it looks like the DR250 tries a MOPv4 download and if that < times out it then falls back to MOPv3. If the DR250 came out7 at about the time Phase V came out then that would have ; been a sensible strategy: try MOPv4 (Phase V) and then fall % back to the older MOPv3 just in case.   8 DEMSAs, for example, have MOPv3 in ROM and will download< the initial system file using MOPv3. If the DEMSA is running; Phase V X.25 Gateway or WANrouter code, that file will then = request the rest of the stuff it needs (such as profile files  or CMIP commands) using MOPv4.  < Since you say those work, I would initially be suspicious of: the DR250. Especially since both MOP (in DECnet) and LANCP9 (in openVMS) fail - those programs are (AFAIK) maintained , independently and without using common code.  : You seem to have a support contract so I would try and get9 a call logged against either DECnet or OpenVMS. If it can < be easily reproduced in the lab I would expect that tracking7 this down would be quite simple. Of course, if it turns ? out to be a DR250 firmware issue then you will almost certainly  never see a fix.   Antonio    --     --------------- - Antonio Carlini             arcarlini@iee.org    ------------------------------  % Date: Sun, 12 Nov 2000 18:25:17 +0000 + From: "antonio.carlini" <arcarlini@iee.org> * Subject: Re: DECnet 5 -> 4 routing problem' Message-ID: <3A0EE08D.23269E52@iee.org>    John Santos wrote: [snip]  F > I think this conforms with what you are thinking about, but I posted% > the whole thing to avoid confusion.   : All good information - but I still cannot see the problem.  B Your DR250 can see the VAX, the VAX (from an earlier post) can see> the DR250. The DR250 can also see both Alphas, as can the VAX.  8 The Alpha in segragated mode will send a packet directly5 to the DR250 and it then gets passed on correctly and  communcitcation ensues.   7 The Alpha in integrated mode will initially head off to ; the VAX. The VAX clearly knows how to get to Area 33 (since 7 a SET HOST from the VAX works OK) so I don't see why it ! cannot route the packet properly.   = The only difference I can see is that the Alpha in integarted < mode will send a packet to the VAX in Phase V format and the= VAX should convert it to Phase IV format to pass it on to the : DR250. The Alpha in segregated mode will talk to the DR250 in Phase IV format directly.  ; If the VAX, when it converts the packet to Phase IV format, 7 forgets to clear the on-NI bit, then the return traffic 9 might go directly from the DR250 to the Alpha in Phase IV 9 format (rather than through the VAX to get converted) and : the Alpha might not be prepared to handle that. It's a bit3 of a stretch but it's all I can think of right now.   8 I might get my DR250 back next week (but don't hold your@ breath) so I might be able to hook it up and see. I suspect thatA to get to the bottom of this you will need to run some NSP traces B on the Alpha, and the VAX to see exactly what path the data takes.6 Always assuming your network is sufficiently quiet for/ the signal not to be lost in a sea of noise ...    Antonio    --     --------------- - Antonio Carlini             arcarlini@iee.org    ------------------------------  % Date: Sun, 12 Nov 2000 13:12:27 +0100 > From: "Jean-Franois Marchal" <jean-francois.marchal@x9000.fr> Subject: f$search improvement ? . Message-ID: <8um195$7vi$1@reader1.imaginet.fr>   Bonjour  tous.   H What would you think of allowing qualifiers in the argument of f$search.   ie :(     FILE = f$search("*.FOR/since=TODAY")  ? All DIRECTORY qualifiers could be usefull to use with f$search. ; There would be no impact on existing usage of the function.   H This would allow to avoid using temporary output files from DIR commandsD or add more or less complicated tests with f$file_attribute calls to check additionnal conditions.   F As a second point, is there any reason why f$search does not return ""A at the second call when the argument does not contain wildcards ?    Cordialement Jean-Franois Marchal  X9000 - LYON (FR)    ------------------------------  % Date: Fri, 10 Nov 2000 17:13:59 +0000 0 From: andrew harrison <andrew.nospam@uk.sun.com>  Subject: Re: Gartner and OpenVMS( Message-ID: <3A0C2CD7.1ED977@uk.sun.com>   bawilhelm@my-deja.com wrote: > E > We just got off the phone last week with Gartner and they are still E > down on VMS.  This is really making it difficult for me to convince A > management that we need to stay with VMS for our new PeopleSoft E > install.  It seems to me that Compaq needs to find out exactly what I > Gartner is looking for and do their best to accommodate them.  It seems I > that would go a long way towards improving the impression of VMS in the D > marketplace.  Whether you think Gartner is right or not is not theG > issue.  The issue is that their recommendations have an impact on the ) > business decisions made.  Any thoughts?  >   > Gartner place a great deal of emphasis on something they call * ISV enthusiasm and available applications.  < What they mean by ISV enthusiasm is the commitment of ISV's ! to put new apps onto a platform.    ? They prefer platforms that have both and are much less keen on  4 platforms who have one and not the other or neither.  7 Compaq are almost certainly aware of this but it is not 9 something that Compaq can influence Gartner on directly,  7 they have to get the ISV's to port to OpenVMS and also  5 to be committed to the platform which are not always  5 the same things. If they do this then it will filter  6 through to Gartner but it is a vicious circle because $ the ISV's also read Gartner reports.  5 Getting the ISV's to modify their behaviour requires   a number of things:    Cash from Compaq% An available market for their product  Joint marketing  etc    > Brent Wilhelm  > System Admin/DBA > ( > Sent via Deja.com http://www.deja.com/ > Before you buy.    --   Andrew Harrison  Enterprise IT Architect    ------------------------------  % Date: Fri, 10 Nov 2000 14:38:29 +0000 0 From: andrew harrison <andrew.nospam@uk.sun.com>7 Subject: Re: Severity of eBay's ecache problem detailed * Message-ID: <3A0C0865.497FBF17@uk.sun.com>  & "Brian Schenkenberger, VAXman-" wrote: > Z > In article <vXePVDXmGW4S@eisner.decus.org>, young_r@eisner.decus.org (Rob Young) writes: > >{...snip...} & > >       Come spin with us Andrew!!!! > > % > >                               Rob  > >  > L > Rob, really!  Andrew's contribution here is to tout Sun and FUD Compaq andK > VMS technologies.  You continue to invite him.  I know that you and Kerry J > and several others are trying to dispell his FUD and show what a FUDsterJ > he is himself but the truth is that I believe we all know what a FUDster) > he is.  Can we try to stop baiting him?  >   E How ironic, you see it was Rob's origional FUDDING of Sun over eBays eC outage that really drew me into this group. What I found was reallys interestingnB a whole group of people where someone could publish either totally	 unfoundedeF and as it turned out totally untrue allegations about another vendors C systems and no one would even question the veracity of the posting.0  < A group where incorrect technical comparisons of one vendors
 capabiliites CH vs generally OpenVMS would elicit nothing more than supporting comments " from a choir of backing vocalists.  @ A group where someone from Compaq can say the world is flat and > someone else will cheer and provide a URL to the "prove" that  it is.  D Just occasionally someone pops out of the woodwork and says no thats+ B****H**T but only very very occasionally. o   No wonder I stayed.n  E You have mostly Rob to blame for this so you are posting to the rights person.n  B Incedentally despite labeling me as a FUDSTER and a spin merchant = the fact is that my objections to mainly Robs hype on Galaxy,tI Spiralog, ISV interest in OpenVMS, the performance of the WildFire boxes e etc have been correct.   Regardss Andrew Harrisoni Enterprise IT Architecta   ------------------------------  # Date: Sun, 12 Nov 2000 16:10:13 GMTe+ From: Craig A. Berry <calepine@my-deja.com> , Subject: Re: SPEC_int92? SPECrate_fp95? etc.( Message-ID: <8umfd4$mv$1@nnrp1.deja.com>  6 The specs you mention are all rather out of date.  See  ; <http://www5.compaq.com/alphaserver/performance/spec2000.hta ml>i  # for more current benchmarking info.h    ) In article <3A0D7239.94D5C55@Easynet.fr>, 3   Didier Morandi <Didier.Morandi@Easynet.fr> wrote:o > In: http://www5.compaq.com/alphaserver/archive/comp/nov95.html > : > I see SPEC_int92, SPECrate_int92 and in the VMS 20 years presentation, SPECint_rate95.  >n, > What are the differences, and the meaning? >s: > I remember that SPECmarks superceded the VUPS, then spec int (integer?)? > and fp (floating point?) came. Where are these units defined,p please?i9 > And are they still evaluating nr of CPU operations/sec?n >i	 > Thanks,  > D. >d    & Sent via Deja.com http://www.deja.com/ Before you buy.g   ------------------------------  % Date: Sun, 12 Nov 2000 11:58:31 -0600e+ From: "Main, Kerry" <Kerry.Main@compaq.com>i' Subject: RE: Sun "uptime" belly laugheroN Message-ID: <910612C07BCAD1119AF40000F86AF0D8052849BD@kaoexc3.kao.cpqcorp.net>  
 G'day Andrew,   L Apologies for not responding your FUD and misrepresentation (or is it reallyE a lack of understanding??) of details what I originally posted... VMS0L business is doing great these days and I am swamped right now, so I have notA been able to get back to these postings as often as I would like.g  K First, I stated that there were some situations where this solution did nothJ work. Users that stay logged in for many days, long running batch jobs etcF require some added work such as "hitman" for inactive processes in offL hours, use of checkpoints in long running batch jobs (which likely should beJ done anyway as system might crash) etc...If "power users" like programmers? don't like this approach, then a compromise needs to be found..r  F > > Of course this is how you should do it, but sadly Kerrys marketingC > > homilly talked about the fact that people expect systems to runAB > > with very low amounts of "scheduled" downtime. You are talkingE > > about using a process that Kerry somehow thinks OpenVMS magicallye7 > > does not need to invoke, namely scheduling outages.e  H >>> I can only think that Kerry was confused and didn't actually mean toF suggest that logging people out every night improves security. LoggingD people out every night and then changing their password....[snip] <<  K Wow, are you saying leaving active connections to applications on open WS'seL or PC's is not a security risk? While not always popular with power users orI developers, surely you do not contest that leaving open WS's or PC's is a  security issue ? h  L On the topic of changing passwords, thats right out off left field. Who said$ anything about changing passwords ? & Not even worth a response on that one.  G We have been over the details on how OpenVMS clusters can TRANSPARENTLYyI shutdown individual systems for planned maintenance with NO middleware ortL application changes with no user warnings or Business Group approvals and NO> user impact. Just re-read previous postings on how to do this.  H So, perhaps you can explain how you do this in a SUN environment wherebyC multiple users need concurrent read-write access to the same files?2  D I am sure the eBay folks will be interested in this procedure and/or process.   Regards,  
 Kerry Main Senior Consultant9 Compaq Canada Inc. Professional Serviceso Voice: 613-592-4660u Fax  :  819-772-7036 Email: Kerry.Main@Compaq.com     -----Original Message-----7 From: andrew harrison [mailto:andrew.nospam@uk.sun.com]d Sent: November 7, 2000 1:03 PM To: Info-VAX@Mvb.Saic.Comt' Subject: Re: Sun "uptime" belly laugherh     jlsue wrote: > 5 > On Thu, 02 Nov 2000 12:18:20 +0000, andrew harrisond# > <andrew.nospam@uk.sun.com> wrote:f >  > >>L > >> For power users like yourself he'd prewarn you of the downtime schedule andeI > >> ask you at some convenient time to log out of the node you are usingd and into another	 > >> one.  > >dE > >Of course this is how you should do it, but sadly Kerrys marketing-B > >homilly talked about the fact that people expect systems to runA > >with very low amounts of "scheduled" downtime. You are talkingeD > >about using a process that Kerry somehow thinks OpenVMS magically@ > >does not need to invoke, namely scheduling outages. Of courseB > >the let out for Kerry is that with 2 nodes or more in a cluster@ > >you only need to schedule downtime for the users or processesD > >attached to one of the nodes but given Curtis's example of havingB > >to kick 1500 users off a node this won't cut much ice with many > >IT professionals. > >  > F > Andrew, you're distorting the message... on purpose I'd wager (since0 > you don't have nearly the same story to tell). > C > The fact is that programming/development systems are a completelyiH > different model, and they are *not* the business critical systems thatG > require 24x7(x365) uptime.  No business I've ever visited thinks thisnG > way.  In fact, some of them are abandoning their internal developmenti< > groups... *that's* how much the businesses care about this > environment. > B The users caught up in Curtis's Cull may well have been developersE but I wasn't talking about systems that are being used by programmers E or developers though I would not agree that developers should expect dE lower availability, lost development time is a cost just in the same o5 way that a call center going off-line is also a cost.h  G > And show me the example where 1500 users are logged-on 24x7.  Kerry'stC > example is valid because those 1500 users do log-off and go home.. >   D I have just been part of a project to roll out a Seibel application H worldwide, users in Tokyo, Sydney, Singapore, Frankfurt, London and New ? York. The total user population is around 3000, now you tell metC where the window is that I can use to log users out. You could log wG them out at the weekend but reporting and roll-up is run at the weekend A so you can kill the users off but you will also be unpopular for e other reasons. o  H This forces the customer to do some remedial work, particularly where it is 6 urgent during the week.   iH Now in this case these are actual connected users but the same customer I also has systems with no connected users but lots of connected processes  G which also run 24 hours a day. Some of the processes are extremely long D running making the cost of killing the processes higher further into the process run.  B Same for their treasury applications same for the upcoming general" ledger implimentation etc etc etc.  B These are not eCommerce apps which are even worse because they do = have to be available 24x7 with connected users 24 hours a dayF 7 days a week.      D > Andrew, I can honestly tell you that I've administered VMSclustersD > where *nobody* saw any application availability problems in over 2H > years.  And these were systems where "telnet" was the main entry pointE > (i.e., stateful connections, unlike remote database/transaction ande > http connections). > @ > I even upgraded from OpenVMS 5.3 to OpenVMS 5.5-2 without userG > interruption on those clusters.  I've applied patches (few that theredF > were, in comparison to other systems I've seen), added memory, cpus,B > disk drives, etc.  All without users being inconvenienced by the, > outages.  THAT is what VMSclusters can do. > D > Will it do this well for *every* environment?  Of course not.  AndE > nobody in here is advocating that it will - nobody, that is, exceptoD > for you, who keeps pushing that idea out here and then throwing upE > individual cases where it hasn't worked to prove it wrong.  This iseH > typical "Andrew-style FUD," and I've seen you use it for years.  It is: > getting very old by now,  can't you be more imaginative? > 3 > Not speaking for anyone, certainly not DEC/Compaq:/ > (get rid of the xxxx in my address to e-mail)t   -- a Andrew Harrisong Enterprise IT Architect9   ------------------------------  % Date: Fri, 10 Nov 2000 15:53:53 +0000t0 From: andrew harrison <andrew.nospam@uk.sun.com>' Subject: Re: Sun "uptime" belly laugherR* Message-ID: <3A0C1A11.E33D0A20@uk.sun.com>   jlsue wrote: > 5 > On Tue, 07 Nov 2000 18:02:38 +0000, andrew harrisont# > <andrew.nospam@uk.sun.com> wrote:i >  > >jlsue wrote:e >  > >cJ > >> And show me the example where 1500 users are logged-on 24x7.  Kerry'sF > >> example is valid because those 1500 users do log-off and go home. > >> > >hF > >I have just been part of a project to roll out a Seibel applicationJ > >worldwide, users in Tokyo, Sydney, Singapore, Frankfurt, London and NewB > >York. The total user population is around 3000, now you tell meE > >where the window is that I can use to log users out. You could log J > >them out at the weekend but reporting and roll-up is run at the weekendC > >so you can kill the users off but you will also be unpopular for  > >other reasons.o > F > Typical Andrew, you misrepresent what I said, AGAIN.  What I said isH > that the individual users don't stay logged-on 24x7.  And, as has beenF > explained to you many, many, many times, the users will "migrate" toF > the other systems over time via disabling the logons; thus they will, > "move" transparently to the other systems. > @ Its not the users directly that cause the problem, but the users	 worldwide<H update the master Seibel database, all the databases worldwide are kept ? syncronised using a replication mechnism provided by the Seibel 
 application. o  @ So a seibel user in Tokyo updating a database is also updating a databaseG in London, Frankfurt etc while the users that access these systems are V all logged out and asleep.  E Turning off a node  even if it has no users attached to it also stopsl the H replication from running. While this isn't an unrecoverable situation it is iE downtime for the replication server and can cause quite a big impact.f  H This kind of architecture which is complex is quite common particularly F in Investment Banking. Customers of the Investment bank want one worldI wide database to avoid inconsistency but some local financial regulators aH require all data relating to their market to be held physically in theirC country. Singapore financial regulations for example require this. s     > >lK > >This forces the customer to do some remedial work, particularly where itE > >is  > >urgent during the week. > >dJ > >Now in this case these are actual connected users but the same customerK > >also has systems with no connected users but lots of connected processessJ > >which also run 24 hours a day. Some of the processes are extremely longG > >running making the cost of killing the processes higher further intos > >the process run.u > @ > Any processing that runs over, say 10 hours, that doesn't haveH > built-in restart capabilities, deserves all the problems that E*bay isD > seeing today. (I know, that's not a good example, I'm just jerkin' > your chain.) >  > >VE > >Same for their treasury applications same for the upcoming generalt% > >ledger implimentation etc etc etc.e > # > These jobs run for over 10 hours?u >   C I thought you said you were familiar with this kind of environment.   B 10 hours is quite short for some jobs, a general re-pricing can inB the customer I work for run for up to 20 hours. Now of course you @ can restart the re-pricing, you don't lose data but you do waste, processing time and you may run out of time.   > >'D > >These are not eCommerce apps which are even worse because they do@ > >have to be available 24x7 with connected users 24 hours a day > >7 days a week.e > B > Sure, but they are single, persistent connections for those 24x7F > windows.  Also, most "good" applications will detect connection lossF > and re-connect and re-transmit (if necessary).  This is a real world. > scenario, and networks go down all the time. >     E Ahh you used the word "good". A lot of the apps that my customer runs H at least at the back end were written long before anyone had ever heard H of the WEB or at least before the first edition of Bill Gate's book :):)     Regardso Andrew Harrison- Enterprise IT Architect@   ------------------------------  % Date: Fri, 10 Nov 2000 11:58:28 +0000n0 From: andrew harrison <andrew.nospam@uk.sun.com>1 Subject: Re: Sun UE10000 is *NOT* an HPC platforms* Message-ID: <3A0BE2E4.DA483F2D@uk.sun.com>   Arne Vajh=F8j wrote: > =h   > andrew harrison wrote:@ > > I don't think that is the point, the real question is should? > > Kerry be taking some sort of holier than thou attitude witha? > > other vendors when Compaq and Digital before them have such 8 > > a poor track record or should he refrain from making > > pronouncements.t > > G > > Compaq/Digital may have "ponied up to the bar" over the RA-XX diskss< > > they didn't over the VAX9000 the MIPS fiasco, the Armada4 > > laptops the broken NT on Alpha promises etc etc. > =h  B > MIPS and NT/Alpha are business marketing decisions not technical > problems.p > =t    7 Ohhhh no they were not. The business/marketing decisonso6 killed off the MIPS product range but a system going =  9 slower when you add more CPU's is not caused by marketingt its a technical issue.  ? Sure sales and marketing at Digital over hyped the capability =:  9 fo the MIPS MP server but technically it didn't deliver =   0 what it was supposed to for a number of reasons.  ( 1	Ultrix wasn't optimised for MP systems. 2	The architecture of the MP systems ment that 	they were unlikely to scale.h" 3	Digital were not helped by MIPS.  : Now you could forgive some of this, MIPS did shaft Digital; if Digital had stopped trying to sell the systems when theyB6 knew that they didn't work, but that was not the case.   For example in the UK   @ Sun and HP lost out in one EDU procurement because another EDU =  H establishment provided possitive references for the MIPS boxes despite =  C knowing from their own experience that they didn't work. They did =   B so because they were promised a working solution in return for a =   reference. =    @ The was a huge row when the first EDU organisation found out and@ neither system ever worked, HP picked up one of the accounts and< Sun the other ending Digitals involvement in either account.  B Please don't get me wrong, this is not a Sun vs Compaq argument itC is an argument about Kerrys right to assume some sort of Moral high A ground in this discussion which he quite obviously does not have.i    , > VAX 9000 and RA-disks are 10-15 years ago. > =     < Right so why is Kerry using the RA-XX story as an example of? Compaq/Digital "Ponying up to the bar". There are more recent =d  A examples of Digital/Compaq doing exactly the opposite e.g dumpingsB customers in the S**T. I didn't notice you correcting his posting.  B > So basicly you are comparing SUN's UE10000 problems with Compaqs > Armada laptop problems !!!!i > =a    ? No you missed the point, Kerry is trying to assume some sort ofs7 holier than thou persona which Digital and Compaqs pasty& actions give him no right to assume. =     Regardsn Andrew Harrisonl Enterprise IT Architecto   ------------------------------  % Date: Fri, 10 Nov 2000 12:20:38 +0000y0 From: andrew harrison <andrew.nospam@uk.sun.com>1 Subject: Re: Sun UE10000 is *NOT* an HPC platforme* Message-ID: <3A0BE816.86BBF6FC@uk.sun.com>   Jordan Henderson wrote:o > , > In article <3A082785.BAC246E0@uk.sun.com>,5 >   andrew harrison <andrew.nospam@uk.sun.com> wrote:r > >a
 > > [snip] > > : > > Sure so you are happy to beat Sun over the head with a: > > Gartner report but whine when someone uses one to beat3 > > Compaq over the head. What was it you called ito > > an anecdote. > >s > < > I'm not whining.  I think the Gartner report is a cause ofB > great concern for Compaq.  I was just pointing out the hypocrisyA > of a Sun employee using a Gartner Report as backup material fore@ > an accusation of quality problems.  You know, glass houses and > all that?b >   ? Why, Sun takes the Gartner report seriously because the ecache oA problem is a serious issue. Your initial dismissal of my statmentHD as an interesting anecdote included by definition a dismisal of the B Gartner report after all the Gartner report backs up the anecdote.  ? From this it is obvious that you didn't take the Gartner reports	 seriouslymF you do now because you have relaised belately that your argument isn't very water tight. More spin.  E > Oh dear, it took me months of explanations to get you to understandn@ > what the term FUD means, am I going to have to spend much timeA > getting you to comprehend what an anecdote is?  Let's hope yourw< > comprehension difficulties don't get in the way this time. >   @ Sorry Jordan you have provided months of explanations which you = comprehensively blew in the last paragraph of the discussion   we were having.t  : "You see Andrew, I'm not in the business of FUD.  I don't < challenge _everything_ said here, pro or con Compaq/OpenVMS.9 No, mostly I only get involved in these debates with one r7 Sun employee who is obviously on a mission to discredit.9 Compaq, OpenVMS and it's adherents on this newsgroup set -% aside for discussions about OpenVMS."0  ; And if this is really the case why are you objecting to my i8 posting, all I am doing is showing that Compaq/Digitals 9 past misdemeanors disqualify Kerry from passing judgment .6 on Sun's handling of the eCache issues. Do you really : think that this is FUD. If you hadn't worked it out Kerrys5 point is that based on the RA-XX episode Compaq couldM5 be trusted to handle something like the eCache issue  4 better than Sun, this is by your own definition FUD.  5 I didn't notice you correcting Kerry, where were you.   4 Nor does your "I only get involved in these debates"4 statement have any ring of truth about it, you were 3 after all the person who insisted that the eBay 22 u0 hour outage was Sun's fault long after the head 4 choirmaster had addmitted that it was all a cut and  paste problem on his part.    F > My dictionary says that an anecdote is "a usually short narrative ofD > an interesting, amusing, or biographical incident".  A story aboutG > your friend's Compaq laptop is an anecdote, while a Gartner Report is  > not. > I Jordan in case you hadn't worked it out your story about the reliability "B of your Armada vs the Dell was also an anecdote, unfortunately my D anecdote at least has a report from Gartner to corroberate it which  yours does not.g  D Also in case you hadn't worked it out most of you Java postings are B also based on a series of anecdotes hence the unconcious irony on 
 your part.     regardse Andrew Harrisona Enterprise IT Architects   ------------------------------   Date: 12 Nov 2000 09:06:48 GMT- From: djweath@attglobal.net (Dave Weatherall)a' Subject: Re: Technology of US electionsw5 Message-ID: <DTiotGxQ0bj6-pn2-1Tjud48Q7c7F@localhost>s  E On Sun, 12 Nov 2000 02:24:32, "Glenn C. Everhart" <Everhart@GCE.com> c wrote:  < > As it is with the present election, we are being shown howC > one particular county chosen after the fact has issues. It should.@ > be remembered that misvoting, mistabulation, voter confusion, A > and mistake occur all over. (I found the voting booth this timeh9 > slightly confusing, but figured it out quickly enough.)g  F Speaking as a euorpean. I first saw the Palm Beach voting form on the E TV when they frist reported the complaints and I couldn't see how it iE was meant to work. It was printed on the front page of our newspaper tF on friday and yes it is obvious _when_ you understand it but it is notB the traditional name and mark which, I suspect, most of the older A people grew up with. It is,  however,  a perfect example of what :E happens when people design things in ways that are meant to make it   F easy for the computer and not the human being.  Not to mention keeping
 the TV happy.e  D The other thing that struck me was that I thought that after one of E the previous elections (Reagan/Mondale?) when it was widely accepted  D that the announcement of results in the east had a marked effert on A voting in the west, the U.S. had decided that no result would be tE published until all the polls had closed. Even the Russians do that,   iIRC.t  n> > That has nothing much to do with the merits of the electoral> > system. When the vote totals are equal to the precision with> > which the counts of legal, single-voting, qualified citizens@ > are known, some anomalies must be expected. We will never knowB > who actually won the popular vote. Neither did we in 1960. There= > is too much error, and too much probable fraud somewhere orO> > other to know. I do suspect though that the electoral system9 > removes the temptation from anyone able to produce huge @ > numbers of fake votes from doing so. Enough to swing the state; > is all that would be needed. We don't need the floodgates 2 > opened to people who decide to swing the nation.  D That's true in normal cases. vis a vis Reagans 'landslide'  victory ? over Carter (?) where he had approx 48% of the popular vote to  B Carter's 47%. Hardly a landslide in those terms but the electoral B college made it seem so. The problem you've got is when those low E margins get so close that you can't trust the result at all. A total sA tally of the votes can make it harder to 'fiddle' and maybe even  D averages out the error. However, if it is close, then its close and ? every care needs to be taken that the electorates true wish is y' implemented. Government of, by and for.t  F It's a federal election so shouldn't there be a federal voting system?B Should each constituency choose is own method. Is that fair or an C invitation to skewing the ballot. Isn't that partly what the Civil aF Rights movement was trying to achieve in the 60's and 70's? (Still?)  F Oh well that's enough OT. It definitely ain't COV unless VMS gets the  job of doing the count :-)   Cheers - Dave.   ------------------------------  % Date: Sun, 12 Nov 2000 04:22:44 -0400 - From: JF Mezei <jfmezei.spamnot@videotron.ca> ' Subject: Re: Technology of US electionsa, Message-ID: <3A0E5352.954E560D@videotron.ca>   Rick Cadruvi wrote:i > H > Glenn's points are well taken.  Consider for a moment that without theA > electoral college, we would be doing recounts all over the US.    M The electoral college has nothing to do with recounts. Recounts should happenh: whenever they are not certain of the results in one area.   K What the electoral college does is blow problems out of proportions because L problems of one area affect a large block of votes for the whole state. If ,N in the case of florida, you have 25 independant electoral votes, then problemsL in palm beach would only affect 1 vote instead of a whole block of 25 votes.  F But electoral college or not, what is important is to have systems andL procedures in which the citizens can trust to properly count the ballots and prevent frauds.w  K When you deposit money at the bank, you expect the bank to have the systemsuK and procedures in place to prevent fraud, theft and to ensure that the bank B won't lose your money. The same should apply when you cast a vote.  M And I beleive that the USA is making a fool of itself by allowing votes to benN tabulated after the closing of the polls. Those who vote remotely should do soH BEFORE the election to ensure their ballot has arrived on the dat of theJ elction, not 10 days later. Once they close the polls, it should be final.   ------------------------------  % Date: Sun, 12 Nov 2000 04:30:27 -0400s- From: JF Mezei <jfmezei.spamnot@videotron.ca>,' Subject: Re: Technology of US electionsn, Message-ID: <3A0E5520.BECF84AA@videotron.ca>   Dave Weatherall wrote:G > Speaking as a euorpean. I first saw the Palm Beach voting form on theeF > TV when they frist reported the complaints and I couldn't see how it > was meant to work.  E In all fairness to the Yanks, their elections include a whole slew ofeH referendums and elections, not just voting for the president. They electD everything from your sheriff to the president, and have all sorts ofI resolutiosn you vote on. So the ballot is longer and more complex than inaL other nations. But it should still be clearly laid out. The use of colour or* even gray would have helped in the design.  L I.e. the rectangle that sourrounded Alo Gore should have narrowed to form anL arrow right at the hole, while the rectangle that surrounded Buchanan shouldL have done the same. This way, it would have been clearer that the hole aboveL the all gore hole did not belong to al gore. The way I saw the ballot on TV,> it make both holes appear to be part of the al gore rectangle.  E > that the announcement of results in the east had a marked effert onsB > voting in the west, the U.S. had decided that no result would beF > published until all the polls had closed. Even the Russians do that, > iIRC.o  M In Canada, where we have 5 time zones, the media agree not to release resultsnM to a province until that province's polls have closed. There are also no exituI polls. All the media can report is heavy or light turnout. Of course, the-K advent of the internet is making it much harder to prevent election resultsbM from being released to regions where the polls are still opened. But at leastaI the mass media are not sending out information to the public before theirg polls have closed.   ------------------------------  % Date: Sun, 12 Nov 2000 10:36:23 -0500a2 From: rdeininger@mindspring.com (Robert Deininger)' Subject: Re: Technology of US electionssL Message-ID: <rdeininger-1211001036240001@user-2ive631.dialup.mindspring.com>  [ In article <3A0E5352.954E560D@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> wrote:i   > Rick Cadruvi wrote:  > > J > > Glenn's points are well taken.  Consider for a moment that without theC > > electoral college, we would be doing recounts all over the US. L > O > The electoral college has nothing to do with recounts. Recounts should happena< > whenever they are not certain of the results in one area.   AThere are currently uncertain results in MANY areas in the U.S.  Florida is the only one that appears likely to matter, so the others are mostly being ignored.  If the popular vote mattered, we would be seeing widespread recounts, since the overall difference in votes is well within the expected uncertainty in counting.r   M > What the electoral college does is blow problems out of proportions becauseuN > problems of one area affect a large block of votes for the whole state. If ,P > in the case of florida, you have 25 independant electoral votes, then problemsN > in palm beach would only affect 1 vote instead of a whole block of 25 votes.  { Yes, that would help.  Maine and Nebraska use that system, by the way.  Each state sets the rules for picking its electors.h  H > But electoral college or not, what is important is to have systems andN > procedures in which the citizens can trust to properly count the ballots and > prevent frauds.c   The biggest frauds don't tend to come from counting "errors".  The most common problem comes from false voter registrations and the opportunity this gives for non-eligible folks to vote, or for eligible folks to vote more than once.  It has gotten VERY difficult to remove people from the registration lists in a lot of places -- counties have been known to spend many millions of dollars to purge their rolls.  Registration clean-ups don't happen nearly often enough.  There have historically been cities with w  :There are always problems with irregular ballots.  Relatively few of these are fraud.  If the election isn't close, these ballots are simply ignored.  Careful recounting usually lets some fraction of the irregular ballots be included in the result.  Fraud comes in when you start picking which precints to recount.  z The type of ballot used in Florida has been widely used in many states for decades.  There's really nothing wrong with it.  O > And I beleive that the USA is making a fool of itself by allowing votes to betP > tabulated after the closing of the polls. Those who vote remotely should do soJ > BEFORE the election to ensure their ballot has arrived on the dat of theL > elction, not 10 days later. Once they close the polls, it should be final.   This varys according to state law.  Some states require mail-in ballots to arrive by the time the polls close.  That is much safer than the Florida system, IMHO.w   The other thing we need to do is refuse to release ANY results until the vote is over nationwide.  Again, this is a matter of state law.  Maybe the current fiasco will convince some states to make this change.s  AThere's absolutely nothing we can do about the exit polls and the early announcements they lead to.  But the Voter News Service is (or ought to be) completely discredited.  Maybe noone will pay attention to them, and they will just fade away.  But we can't outlaw the exit polls.  Our first amendment wouldn't allow that.    -- l Robert Deininger rdeininger@mindspring.comt   ------------------------------  % Date: Sun, 12 Nov 2000 10:45:32 -0500i2 From: rdeininger@mindspring.com (Robert Deininger)' Subject: Re: Technology of US elections L Message-ID: <rdeininger-1211001045320001@user-2ive631.dialup.mindspring.com>  [ In article <3A0E5520.BECF84AA@videotron.ca>, JF Mezei <jfmezei.spamnot@videotron.ca> wrote:g   > Dave Weatherall wrote:I > > Speaking as a euorpean. I first saw the Palm Beach voting form on the H > > TV when they frist reported the complaints and I couldn't see how it > > was meant to work. > G > In all fairness to the Yanks, their elections include a whole slew of J > referendums and elections, not just voting for the president. They electF > everything from your sheriff to the president, and have all sorts ofK > resolutiosn you vote on. So the ballot is longer and more complex than in N > other nations. But it should still be clearly laid out. The use of colour or, > even gray would have helped in the design.  j The pictures I've seen are somewhat misleading.  They give a 2-dimensional view of a 3-dimensional device.  The ballot was designed by a Democrat election judge, BTW.  It was presented to both parties ahead of time, and both approved.  Then a sample was mailed to every registered voter.  There weren't any objections before, during, or after the election --- until the result was known to be close.  If you want to challenge a very close election, there's ALWAYS a semi-plausible reason available.   Speaking of color, in some parts of Indiana, they managed to print the ballots with all (and only) the Democrats hi-lighted in bright yellow.  I guess nobody thought of outlawing that yet.  O > In Canada, where we have 5 time zones, the media agree not to release results 9 > to a province until that province's polls have closed. i  The U.S. media makes the same promise.  They often ignore it, and there's no way to punish them.  This time, they cheated in both Florida and Texas, but Texas was so lop-sided nobody cared.  (Also, the initial guess of the Texas result turned out to be correct.)   -- s Robert Deininger rdeininger@mindspring.com    ------------------------------    Date: 12 Nov 2000 13:52:10 -0500/ From: jordan@lisa.gemair.com (Jordan Henderson)s' Subject: Re: Technology of US electionsa* Message-ID: <8umosq$o4v$1@lisa.gemair.com>  L In article <rdeininger-1211001036240001@user-2ive631.dialup.mindspring.com>,3 Robert Deininger <rdeininger@mindspring.com> wrote:: >3 > [snip] >mB>There's absolutely nothing we can do about the exit polls and the early announcements they lead to.  But the Voter News Service is (or ought to be) completely discredited.  Maybe noone will pay attention to them, and they will just fade away.  But we can't outlaw the exit polls.  Our first amendment wouldn't allow that. >o  D I agree that exit polling and announcing poll results is a matter ofI protected First Amendment speech.  I also think that responsible citizens D should refuse to answer any pollster's questions or, if you choose, H this is a situation where you are ethically allowed to lie to pollsters.  H Certainly lying to pollsters is also protected speech and it's justified@ on ethical grounds in that it might help to end this destructive	 activity.h   >--  >Robert Deiningert >rdeininger@mindspring.com   -Jordan Henderson  jordan@greenapple.coms   ------------------------------  % Date: Sun, 12 Nov 2000 19:08:48 +0010u% From: paddy.o'brien@zzz.tg.nsw.gov.aua Subject: Test only5 Message-ID: <01JWGP3IHTTU006H6O@tgmail.tg.nsw.gov.au>i   ------------------------------   Date: 12 Nov 2000 07:23:30 GMT- From: djweath@attglobal.net (Dave Weatherall) , Subject: Re: Where's the Y2K compliance page5 Message-ID: <DTiotGxQ0bj6-pn2-EBLkDnOdfdSK@localhost>s  D While on the subject of Y2K, is there anybody out there still using D old DOS-11 date formats? If so, don't forget that they could easily F break in 2003. To explain, the DOS-11 format is a 16 bit integer with 
 the format :-   8        (years since 1970)*1000 + no of days in the year.  E If the integer in question is treated as signed you're alright until wC 1970+32. After that you get negative years :-) If it is treated as oF unsigned you've got until 2036 (1970+66). Our major task for next yearC is to get rid of  all these old formats, which have been kept, for wD reasons of backward compatibility, from DOS-11 over RSX and now VMS  and ELN.   Cheers - Dave.   ------------------------------  % Date: Sun, 12 Nov 2000 18:59:59 +0010a% From: paddy.o'brien@zzz.tg.nsw.gov.aun, Subject: Re: Where's the Y2K compliance page5 Message-ID: <01JWGOSKLRRM006H6O@tgmail.tg.nsw.gov.au>u   Dave,   E >While on the subject of Y2K, is there anybody out there still using sE >old DOS-11 date formats? If so, don't forget that they could easily -G >break in 2003. To explain, the DOS-11 format is a 16 bit integer with i >the format :- > 9 >       (years since 1970)*1000 + no of days in the year.   L This is interesting, since my understanding of some OS's I never wish to be @ involved with (anything Gatesian) still have foundations in DOS.  F >If the integer in question is treated as signed you're alright until D >1970+32. After that you get negative years :-) If it is treated as G >unsigned you've got until 2036 (1970+66). Our major task for next yearaD >is to get rid of  all these old formats, which have been kept, for E >reasons of backward compatibility, from DOS-11 over RSX and now VMS l	 >and ELN.   O I thought our VMS engineering guys have always maintained no problems.  I also i thought the U**x date was 2038.i  L You leave me in more hope than I had before.  2038/6 I shall be over 90.  I O wanted to live to see that collapse.  Now I've only got two years to go to see gL the demise of DOS and its younger brethren.  I won't even be 60 then, close 
 but... :-)   Regards, Paddy   Paddy O'Brien, Transmission Development, 
 TransGrid, PO Box A1000, Sydney South,  NSW 2000, Australia-   Tel:   +61 2 9284-3063 Fax:   +61 2 9284-3050& Email: paddy.o'brien@zzz.tg.nsw.gov.au  M Either "\'" or "\s" (to escape the apostrophe) seems to work for most people,x; but that little whizz-bang apostrophe gives me little spam.2   ------------------------------  % Date: Sun, 12 Nov 2000 08:46:20 +0100h0 From: Didier Morandi <Didier.Morandi@Easynet.fr>* Subject: [noise] A Maynard Mill Clock Tour* Message-ID: <3A0E4ACC.3FD57341@Easynet.fr>  , For Mill Enthousiasts, do you know that one?  / http://web.maynard.ma.us/history/millclocktour/h   D.   ------------------------------   End of INFO-VAX 2000.633 ************************