0 INFO-VAX	Sat, 28 Jan 2006	Volume 2006 : Issue 55      Contents: Re: "modern" LaTeX etc on VMS  AlphaServer GS140 for auction  Burining dvd's & cd's B Re: circuit up/circuit down . . . messages filling up OPERATOR.LOGB Re: circuit up/circuit down . . . messages filling up OPERATOR.LOG Re: Cluster timeouts question  Re: Cluster timeouts question  Re: Cluster timeouts question  Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing Re: FOR070.DAT files appearing) Re: forgetting my DCL: what am I missing? 1 Re: Getting Advanced Server to work with VMS 8.2? : Re: HP, Intel becoming laughing stock of computer industry: Re: HP, Intel becoming laughing stock of computer industry: Re: HP, Intel becoming laughing stock of computer industry: Re: HP, Intel becoming laughing stock of computer industry! Re: HSJ50 hardware documentation? ! Re: HSJ50 hardware documentation? & Re: Intel/Hp spend $10 billion on IA64& Re: Intel/Hp spend $10 billion on IA64& Re: Intel/Hp spend $10 billion on IA64& Re: Intel/Hp spend $10 billion on IA64& Re: Intel/Hp spend $10 billion on IA64& Re: Intel/Hp spend $10 billion on IA64& Re: Intel/Hp spend $10 billion on IA64( Re: problems with a new vms installation Re: SimH V3.5-2 released9 Re: TCPIP Services: management of cluster alias interface 9 Re: TCPIP Services: management of cluster alias interface ? The processor wars Part V, Intel's Itanium: To be or not to be?   F ----------------------------------------------------------------------  % Date: Fri, 27 Jan 2006 18:33:18 -0500 , From: Carl Friedberg <frida.fried@gmail.com>& Subject: Re: "modern" LaTeX etc on VMSH Message-ID: <890539d90601271533yda8c03fg8bc1f8f5c34f230b@mail.gmail.com>   Hi Phillip,   C I tried to build TeX on I64, but got bogged down. It is a very time = consuming process, and I ran out of round tuits. I did manage 9 to crash Pascal a few times, but then HP produced a newer ; version of Pascal, and I think I got about halfway through. 7 Whether halfway through was 1% or 99% of the work, I'll ; never know. You are welcome to contact me off list for more F details, but since I didn't finish, I don't think it would help a lot.   Best,    Carl friedberg frida.fried@gmail.com   8 On 25 Jan 2006 14:29:42 -0800, stuie_norris@yahoo.com.au" <stuie_norris@yahoo.com.au> wrote:
 > Hi Phillip,  > 2 > I am in the same situation - circa 1997 version. > F > Actually I use Itaniums for number crunching, but have an Alpha with@ > LaTeX for document processing using the 1997 version of LaTeX. > I > I would also like to see a modern version - expecially with the ability  > to use non-LaTeX fonts.  > H > I do all my development on Alpha and copy the final version to Mac OSX7 > to produce non-LaTeX and PDF version for publication.  > H > About a year ago Ralf mentioned here he is producing a Itanium versionD > - using the latest distro, but I am not aware if this has occured. >  > Thanks >  > Stuart >  >    ------------------------------    Date: 27 Jan 2006 15:09:51 -0800$ From: "ataylor" <kusoneko@gmail.com>& Subject: AlphaServer GS140 for auctionC Message-ID: <1138403391.480969.153820@o13g2000cwo.googlegroups.com>    Hi,   G I just listed my GS140 for auction on eBay. Before, when I listed it as B a FS ad, I received a lot of interest from this group. I figured IG would let you guys know about the auction in case someone wants to bid.      Here's the link:= http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=5860036165    Hope it goes to a good home.   Thanks,      -Aaron Taylor    ------------------------------  % Date: Fri, 27 Jan 2006 19:47:22 -0800 # From: "Tom Linden" <tom@kednos.com>  Subject: Burining dvd's & cd's( Message-ID: <ops32ks8y0zgicya@hyrrokkin>  6 On my lan where my cluster lives I have a W2k box with> an Hitachi Gd-7500 CD/DVD player and a Plextor 716A with Roxio5 software.  Also on one of the nodes VMS7.3-2 there is A device type SONY  CD-RW  CRX185E1, hopefully to be augmented with ' DVD RW drive  (recommendations welcome)   ? I would like to use these devices for backups, and other things C The plextor with its 4.7GB capacity is clearly the most attractive.   G As an experiment, I tried simply burning a file onto TDK 4x DVD+R media H  from the Plextor  and reading it form the Hitachi drive on same system.K When burning I  was prompted for the format and used UDF 1.2 and ISO/Joliet G as this was  recommended for the ability to read on other systems.  But H could not read it on the Hitachi drive on same box, although plextor had? no problems.  Both drives are functional and otherwise working.   E I know W2K puts it a bit off topic but the objective is ultimately to K record VMS backups on DVD and for me this is part of the bootstrap as I was I planning to put a similar, VMS compatible drive into one of the cluster    nodes.  ( So, anybody fought this battle, and won?   Tom    ------------------------------  # Date: Fri, 27 Jan 2006 20:31:30 GMT & From: John Reagan <john.reagan@hp.com>K Subject: Re: circuit up/circuit down . . . messages filling up OPERATOR.LOG 2 Message-ID: <CqvCf.2228$hF6.2227@news.cpqcorp.net>   dpm_google@myths.com wrote: F > In the last twenty-four hours alone, the logfile for my IA64 box has > 8250 messages G > warning the circuit is down with 8251 celebrating that the circuit is  > up. 6 >           "Circuit EIA-0, Line synchronization lost"A > Precisely every ten seconds, followed *immediately* by the "up" 
 > message. >    Phase IV? Yes?  $ Which machine?  Some rx2600 variant?  E When you did the NETCONFIG, it probably found BOTH the EIA0 and EWA0  F devices (there are 2 Ethernet devices on the machine not counting the E management port).  When I install on my rx2600, I let NETCONFIG find  B both lines/circuits and execute the commands to put them into the I database.  I don't let it start DECnet however.  I then drop back to DCL  H and to NCP PURGE LINE EWA-0 ALL and PURGE LINE EWA-0 ALL.  I then start E DECnet.  (which one you pick, depends on the which Ethernet you use).   E You can also look at CLEAR LOGGING and look at the EVENTS subcommand.    --   John   ------------------------------    Date: 27 Jan 2006 13:38:01 -08003 From: "dpm_google@myths.com" <dpm_google@myths.com> K Subject: Re: circuit up/circuit down . . . messages filling up OPERATOR.LOG C Message-ID: <1138397881.915178.250600@g43g2000cwa.googlegroups.com>   F Well, shame on me.  Thanks for pointing out the obvious; I'm using EWAD and it's EIA that's flipping out.  The commands SET LINE EIA-0 STATE OFF @ and SET CIRCUIT EIA-0 STATE OFF stopped the messages, of course.   ok dpm    ------------------------------  % Date: Fri, 27 Jan 2006 16:06:37 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> & Subject: Re: Cluster timeouts question, Message-ID: <43DA8B59.74760723@teksavvy.com>   Richard Brodie wrote: H > If a remaining part of the cluster has a quorum, then it can decide to > eject missing nodes.  F Ah, that makes sense. So in fact the error message of node voluntarilyF crashing is misleading. It should say "rest of cluster has requested I. reboot because they don't like me anymore :-)"    F And I understand the reasons for this (this was discussed before). AndF if the whole cluster had lost quorum, then since there was no activity? at all, nodes are still in sync and there is no need to reboot.    ------------------------------    Date: 27 Jan 2006 13:48:27 -0800 From: etmsreec@yahoo.co.uk& Subject: Re: Cluster timeouts questionC Message-ID: <1138398507.097984.315450@z14g2000cwz.googlegroups.com>   - Baldrick (a.k.a. Mr. Car Park Charges) wrote:  > G >Your last sentence suggests it. When DSSI is connected between systems I >then SCS can pass and by default does. Situations where DSSI is involved F >and SCS does not pass is generally where they are not connected, i.e.+ >one "private" DSSI bus per cluster member.  > H >The first part of the response isn't correct, if its DSSI it is part ofI >the SCA, and valid as a cluster medium. Shared SCSI never has passed SCS  >traffic of course.  >   F Unusually Nic, that's wrong.  check page 4-18 of VAXCluster PrinciplesF (""The KFQSA provides an interface between a Q-bus and a DSSI bus, andF supports only VMS-to-device communication.") or page 4-11 (as at v7.1)E of "Guidelines for OpenVMS Cluster Configurations" ("The KFQSA cannot = be used for node-to-node cluster comunication.  An additional B interconnect must be configured between systems that use KFQSA for access to shared storage.")   D Inherently, DSSI storage is presented (IIRC) using MSCP and/or TMSCPG protocols.  This does not mean that other inter-node SCS protocols will G also pass along the link, regardless of SYSGEN parameters.  However, JF > doesn't say what his hardware is, whether it's VAX or Alpha or
 Integrity.  F You're right JF in thinking that the "node voluntarily exited" messageF is actually a "they forced me to fall on my sword" message whereby theB other nodes force the one that comes back to crash.  The reason isB simple - the other cluster members may have carried on processing,G redistributed locks that the departed node was interested in, taken out F new locks etc etc etc.  To rebuild the data structures in memory wouldG potentially be a minefield.  It's easier for the rest of the cluster to = say "reboot you b*****d" and rebuild everything from scratch.    Steve    ------------------------------  + Date: Fri, 27 Jan 2006 23:10:00 +0000 (UTC) 7 From: moroney@world.std.spaamtrap.com (Michael Moroney) & Subject: Re: Cluster timeouts question( Message-ID: <dre988$4hg$1@pcls4.std.com>   etmsreec@yahoo.co.uk writes:  G >You're right JF in thinking that the "node voluntarily exited" message G >is actually a "they forced me to fall on my sword" message whereby the C >other nodes force the one that comes back to crash.  The reason is C >simple - the other cluster members may have carried on processing, H >redistributed locks that the departed node was interested in, taken outG >new locks etc etc etc.  To rebuild the data structures in memory would H >potentially be a minefield.  It's easier for the rest of the cluster to> >say "reboot you b*****d" and rebuild everything from scratch.  F If I remember correctly, one node (the one with the lowest scssystemidE or the one that notices the problem first or something) computes the  F "best" possible cluster given the connectivity issues, votes etc.  TheC best cluster may or may not include the node doing the computation. C So sometimes that node tells others to die and sometimes it commits E suicide and sometimes both.  Of course, in the case where one node is E disconnected and the rest of the cluster still has quorum and the one G node gets reconnected (after RECNXINTERVAL expires) the one node is the D one is the one that will die regardless of who does the calculation,G since obviously a functional cluster is better than one hung node.  But G if you look at the one node's crash dump, you'll see its picture of the E cluster is itself with every other node kicked out.  It's possible it J looked at the two possible clusters and saw "oops, your functional cluster3 is better than my one node hung cluster - bye bye!"    ------------------------------    Date: 27 Jan 2006 13:51:31 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ' Subject: Re: FOR070.DAT files appearing 3 Message-ID: <dsmJc7LhY1uV@eisner.encompasserve.org>   P In article <7IOdnQ_Gm8z63ETeRVnyrA@brightview.com>, Eccles <nospam@home> writes: > I > I've tried to see who the file creator is but the file ownership takes  I > the default ownership of the directory (is there any way around this?).  >   ?    You can create a security ACL which will tell you all access B    to the file.  You will need privileges to do this.  I once used>    this to track down a timing problem between two cooperating@    processes when the second one started failing to open a file.   ------------------------------    Date: 27 Jan 2006 13:47:26 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ' Subject: Re: FOR070.DAT files appearing 3 Message-ID: <XhYz1L8a7rKI@eisner.encompasserve.org>   W In article <43ugc5F1pibolU1@individual.net>, bill@cs.uofs.edu (Bill Gunshannon) writes:  > B > Temporary or not, relying on implicit behavior is just plain badG > programming practice.  Everything a program does should be explicitly  > stated by the programmer.   B    That sounds like a good way to create code with maintainability    problems.  @    I had a lot of Fortran code like that years ago.  The authors@    specified everything they could in OPEN statements, includingB    everything RMS would have picked up from the file header.  WhenA    we changed the program that wrote the file, we just took those H    keywords out of the OPEN statement in the program that read the file,A    knowing we would never again have to update that code for that 
    reason.  0    I'm sure there have been many other examples.   ------------------------------    Date: 27 Jan 2006 13:48:05 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ' Subject: Re: FOR070.DAT files appearing 3 Message-ID: <9XkDoRNfif$Q@eisner.encompasserve.org>   x In article <43da13de$1@merkur.rz.uni-konstanz.de>, vaxinf@chclu.chemie.uni-konstanz.de (Eberhard Heuser-Hofmann) writes: >  > $ define for070 nla0:a.   9    Or break the code well enough to pin down the problem.    ------------------------------    Date: 27 Jan 2006 13:58:19 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ' Subject: Re: FOR070.DAT files appearing 3 Message-ID: <T+DU+ZlcsZCf@eisner.encompasserve.org>   l In article <Pine.LNX.4.61.0601271020200.6452@localhost.localdomain>, Rob Brown <mylastname@gmcl.com> writes: > F > I think that file ownership following directory ownership is normal E > behaviour if you have permission to write to the directory without   > extra privileges.   G    This is one of the few intentional changes to VMS that I don't like. ?    Early versions would always create the file belonging to the H    creator user, who if privileged enough to create it in someone else'sI    directory then had to rememeber to change the ownership if the intent  ,    was that it be the same as the directory.  D    The original behavior caused problems because people would "give"D    someone else a file and forget to change the ownership.  But the B    current behavior makes it harder to track down and correct the      potential abuse of privilege.  "    I think the change came in 4.0.   ------------------------------  % Date: Sat, 28 Jan 2006 00:06:58 +0000  From: Eccles <nospam@home>' Subject: Re: FOR070.DAT files appearing 3 Message-ID: <GPydnawvt8Q4KEfeRVnyjA@brightview.com>   
 Eccles wrote:  > Hello and good evening,  > I > Where I work, we have a ES45, a DS25 and a DS10 all running VMS 7.3-2.  3 > The application we use is compiled using Fortran. E > Every so often in the application root the file FOR070.DAT will be  4 > created, sometimes up to twenty times in a second.F > We've contacted the application vendor and they don't know anything  > about it.  > I > I've tried to see who the file creator is but the file ownership takes  I > the default ownership of the directory (is there any way around this?).  > * > Any help would be very much appreciated. > 	 > Thanks,  > 	 > Martin.     I Thank you to everyone for thier help and advice.  As sods law would have  H it, when I got into work today I recieved an email from the application E manager saying the the vendor had worked out that there was indeed a  E problem with one of the modules and they'd be issuing a patch for it.   C On a different VMS topic, can anyone let me know where I can get a   *cheap* alpha box in the UK ?    Thanks,  Martin.    ------------------------------  % Date: Fri, 27 Jan 2006 18:20:15 -0800 ( From: Jeff Cameron <roktsci@comcast.net>' Subject: Re: FOR070.DAT files appearing 0 Message-ID: <C00014DF.1AECB%roktsci@comcast.net>  D On 1/27/06 5:34 AM, in article 43ulqgF1pri74U1@individual.net, "Bill% Gunshannon" <bill@cs.uofs.edu> wrote:   2 > In article <BFFF5E73.1AE43%roktsci@comcast.net>,, > Jeff Cameron <roktsci@comcast.net> writes:H >> On 1/26/06 1:38 PM, in article 7IOdnQ_Gm8z63ETeRVnyrA@brightview.com,  >> "Eccles" <nospam@home> wrote: >>  J >>> I've tried to see who the file creator is but the file ownership takes* >>> the default ownership of the directory >>  K >> This particular aspect of your problem indicates that the creator of the N >> file (or the image if it is installed as a global section) used a privilegeO >> (probably BYPASS) to create the file in the directory that it normally would 9 >> not be allowed to do so because of protections/rights.  >>   > G > So what we seem to have is a program running with elevated privileges H > the even the writer doesn't know what all it does.  Bet that gives you" > a real warm fuzzy feeling!!  :-) >  > billL If you need to find the process that is doing it, you could put an alarm acl5 on the directory that the files are being written to.    ------------------------------  % Date: Fri, 27 Jan 2006 18:27:31 -0800 ( From: Jeff Cameron <roktsci@comcast.net>' Subject: Re: FOR070.DAT files appearing 0 Message-ID: <C0001693.1AECD%roktsci@comcast.net>  D On 1/27/06 5:34 AM, in article 43ulqgF1pri74U1@individual.net, "Bill% Gunshannon" <bill@cs.uofs.edu> wrote:   2 > In article <BFFF5E73.1AE43%roktsci@comcast.net>,, > Jeff Cameron <roktsci@comcast.net> writes:H >> On 1/26/06 1:38 PM, in article 7IOdnQ_Gm8z63ETeRVnyrA@brightview.com,  >> "Eccles" <nospam@home> wrote: >>  J >>> I've tried to see who the file creator is but the file ownership takes* >>> the default ownership of the directory >>  K >> This particular aspect of your problem indicates that the creator of the N >> file (or the image if it is installed as a global section) used a privilegeO >> (probably BYPASS) to create the file in the directory that it normally would 9 >> not be allowed to do so because of protections/rights.  >>   > G > So what we seem to have is a program running with elevated privileges H > the even the writer doesn't know what all it does.  Bet that gives you" > a real warm fuzzy feeling!!  :-) >  > billE I just thought of a better idea if putting an alarm acl on the target G directory which would cause undue or wanted alarms when other files are 	 created.    L Instead, create a directory specially for this exercise, put an alarm ACL on= this special directory, then define this system logical name:   2 $DEFINE/SYSTEM FOR070 device:[directory]FOR070.DAT  K Of course specifying the full device and path of you special trap directory   F This would then force all FORTRAN attempts to use LUN 70 explicitly toJ create the file in this specific directory, setting of the ACL alarm whichC would then give you the process and the image doing the dirty deed.    Jeff   ------------------------------  % Date: Fri, 27 Jan 2006 18:39:39 -0800 ( From: Jeff Cameron <roktsci@comcast.net>' Subject: Re: FOR070.DAT files appearing 0 Message-ID: <C000196B.1AECF%roktsci@comcast.net>   On 1/27/06 9:28 AM, in articleC Pine.LNX.4.61.0601271020200.6452@localhost.localdomain, "Rob Brown"  <mylastname@gmcl.com> wrote:  * > On Fri, 27 Jan 2006, Jeff Cameron wrote: > ! >> On 1/26/06 1:38 PM, in article G >> 7IOdnQ_Gm8z63ETeRVnyrA@brightview.com, "Eccles" <nospam@home> wrote:  >>  D >>> I've tried to see who the file creator is but the file ownership0 >>> takes the default ownership of the directory >>  G >> This particular aspect of your problem indicates that the creator of F >> the file (or the image if it is installed as a global section) usedD >> a privilege (probably BYPASS) to create the file in the directory< >> that it normally would not be allowed to do so because of >> protections/rights. > E > I think that file ownership following directory ownership is normal D > behaviour if you have permission to write to the directory without > extra privileges.   
 Incorrect!  H If you are USERA (privileged or not), and you have normal permissions toH create a file in a directory owned by USERB, the file you create will be owned by you, USERA.  K However if you (USERA) do not have normal Rights or Access to create a file I in USERB's directory, but you have the system privilege to do so, such as J BYPASS, or any other privilege that allows you to do so, then the owner of/ the file is the parent of the directory, USERB.   I This is very easy to test, and documented in the guide to system security  at:   L http://h71000.www7.hp.com/doc/732FINAL/aa-q2hlg-te/00/00/78-con.html#filescr$ eatingdependencyondirectoryownership   Jeff > E > Assuming elevated privileges are not involved, it suggests a way to A > get more information.  Define logical name FOR070 to point to a G > directory that no one has write permission to.  Then whoever tries to B > write to FOR070 will crash and you will find out who it is.  SetD > process/dump for that guy and do it again.  Send the process crash > dump to the vendor.  ;-) >  > - Rob  >    ------------------------------    Date: 27 Jan 2006 11:37:18 -0800$ From: "AEF" <spamsink2001@yahoo.com>2 Subject: Re: forgetting my DCL: what am I missing?B Message-ID: <1138390638.606917.66420@f14g2000cwb.googlegroups.com>   Ken Fairfield wrote: > AEF wrote: > > AEF wrote: > >> Ken Fairfield wrote:  > >>> AEF wrote: > >>>> Ken Fairfield wrote:  > >>>>> AEF wrote: > >>>>>  > >>>>> [very BIG snip]  > >>>>> J > >>>>>> BTW: One can also ask why your application can't handle the extraA > >>>>>> space! Especially since the /11 starts with a slash. :-)  > >>>>>>N > >>>>>>>>      And of course, as others have noted, the reason PRIVREQUEST isL > >>>>>>>> giving you this problem is because it is *not* using DCL to parseL > >>>>>>>> the command line.  Otherwise, a space before a qualifier wouldn'tL > >>>>>>>> matter.  One of my pet peeves is "imitation" DCL command lines...5 > >>>>>> So why doesn't the program ignore the space? C > >>>>>      Because the person who wrote the program (not Phillip) D > >>>>> wrote their own command line parser which, for this program,A > >>>>> requires the "/11" to be "attached" to the 3rd parameter. K > >>>> So why does DCL put in that "extra space"? Because the person(s) who 0 > >>>> wrote that part of DCL wrote it that way. > >>> Of course.  Your point? I > >> That *is* my point. "Because it was written that way" answers *both*  > >> questions.  > > H > > IOW, why is the same answer satisfactory for the privrequest program$ > > but not for DCL command parsing? > . > [At the risk of beating dead horses, etc...] > @ > Because of 20+ years of DCL history, experience and associated > expectations.  > B > The context of this discussion was a deviation from expected DCLC > behavior (see the "BTW" at the top of the quoted text above).  In E > one case, a program was invoked from a foreign command symbol which C > specified only the image, while the command line supplied all the C > required parameters and a qualifier.  In the other case, the same A > program was invoked by a foreign command symbol which specified F > *both* the image and the required parameters, while the command lineE > supplied only the (remaining) qualifier.  It is/was unexpected that 5 > the results of two invocations should be different.  > B > Yes, we traced this down to the way DCL expands the command lineF > to include a space between the last parameter of the expanded symbolG > and the supplied qualifier.  And yes, that is just the way the people E > who wrote DCL wrote it.  But the *point* is that added space during E > symbol expansion is *not* an issue either for (a) a native DCL verb B > with its parameters and qualifiers, or (b) a program that uses aD > .CLD to define its syntax and LIB$DCL_PARSE, for example, to parseC > the command line.  It becomes an issue when the program defines a F > syntax that appears to be DCL, i.e., because it uses slash-delimitedH > "qualifiers", but it doesn't behave like DCL because it makes implicitA > assumptions about the command-line tokens it will receive...and @ > parse...that DCL doesn't do, i.e., requiring a qualifier to beA > "attached" to the 2nd parameter instead standing by itself as a  > 3rd blank-delimited token.  E I thought the point was why the extra space is added. Anyway, perhaps  that's nitpicking.  G Anyway, I checked the User's Manual and the instructions for defining a E foreign command do not say anything about including parameters in the E defitinion. The format clearly shows only the file-spec for the image G *without* any parameters. So, strictly speaking, we are in undocumented E territory here. Now this question of the extra space is actually more  general in scope. Consider  #     $ DIR:==DIRECTORY/SIZE/DATE/SEC 
     $ DIR.OBJ   = For some reason, this actually works. In the second line, DCL F interprets DIR as the command verb, performs symbol substitution, adds the "space", and runs   "     $ DIRECTORY/SIZE/DATE/SEC .OBJ  G So I think this is really the question that Phillip was asking. I don't ? know the answer. Perhaps there was some perceived need for this A behavior at one point. Perhaps it was the easiest way to code the A command interpreter when this part of it was written. But we must E remember, getting back to the original problem, that the documenation > does not say you can include parameters in the foreign-commandC definition, so we should not be too surprised about this unexpected F behavior. Also, I don't think it was ever intended for foreign-commandG definitions to include parameters. And if you need to do so you can use > the trailing-apostrophe trick, even though it is undocumented.    D > NOTE: I never meant to disparage nor denigrate David Jones's work,@ > or the OSU HTTP server.  I am _very grateful_ to David for his@ > service to the VMS community and, indeed, we used his OSU HTTP> > server at my previous employer!  There is no need to explain> > why the PRIVREQUEST command line parsing was done as it was.B > Indeed, I've certainly done similar things.  In fact, several ofB > my TPU functions use a "pseudo-DCL syntax" with qualifiers, etc.C > (And no, they don't parse exactly like DCL...which trips me up on E > occasion, especially if I haven't used the function in a while! :-)    Agreed.    >      Regards, Ken  > --8 > I don't speak for Intel, Intel doesn't speak for me... >  > Ken Fairfield # > D1C Automation VMS System Support $ > who:   kenneth dot h dot fairfield > where: intel dot com   AEF    ------------------------------  # Date: Fri, 27 Jan 2006 23:48:24 GMT E From: patrick@unknown (Patrick Young <patrick@hilux.ace.unsw.EDU.AU>) : Subject: Re: Getting Advanced Server to work with VMS 8.2?" Message-ID: <J5noDYGCD024@unknown>  Z In article <OnpCf.2201$Bi6.957@news.cpqcorp.net>, "PEN" <paul.nunez.nosp@m.hp.com> writes:   > You need ECO4.  4 Many thanks everyone, everything is all working now.   --    , -------------------------------------------- 4x4 Hilux Auto Service Centre, BP 106 Timbuktu, Mali (West Africa) Tel: 292 91 52, Specialising in turbo diesel and R290 aircon, --------------------------------------------   ------------------------------  % Date: Fri, 27 Jan 2006 15:50:43 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> C Subject: Re: HP, Intel becoming laughing stock of computer industry , Message-ID: <43DA87A2.C5019620@teksavvy.com>  ! VAXman-, @SendSpamHere.ORG wrote: I > If you mean a digital SLR, I don't believe so.  However, the rags often I > review the consumer "point and shoot" digital cams and HP's just do not  > fair well.  @ Here is how this was explained to me: HP is a baby in the cameraF business. Digital cameras are more than a few chips and a USB port andE PC software. While the film may have been replaced with an electronic ? sensor, the lens is still a very important part, as is exposure : metering. HP doesn't have the expertise in that area yet.   G Also, HP doesn't have the facilities to make its own lenses. One of the F big reasons for an SLR is to have interchangeable lenses, something HPF could not afford because it would be such low volume for it. (and very expensive to setup).  C And perhaps more importantly: HP's distribution is through comsumer H grade electronics stores. It doesn't have anything setup with real photo& shops to distribute high end cameras.   ? HP *might* evolve its cameras over time and eventually be taken G seriously enough that HP might venture into higher end models. But this F isn't something done overnight just because Carly passed an edict that( making cameras would increase ink sales.  J Nikon, Hasselblad and Carl Zeiss didn't build their reputations overnight.  G > I'm still fond of my d|i|g|i|t|a|l LNCO2.  The only problem is that I 3 > cannot find a source anymore for its consumables.   ? Is that a dot matrix with colour ribbons ? Or is it ink based ?    ------------------------------  % Date: Fri, 27 Jan 2006 15:48:15 -0700 " From: GreyCloud <mist@cumulus.com>C Subject: Re: HP, Intel becoming laughing stock of computer industry 0 Message-ID: <p8GdnT_ygM1fPkfeRVn-uA@bresnan.com>   JF Mezei wrote:   # > VAXman-, @SendSpamHere.ORG wrote:  > I >>If you mean a digital SLR, I don't believe so.  However, the rags often I >>review the consumer "point and shoot" digital cams and HP's just do not  >>fair well. >  > B > Here is how this was explained to me: HP is a baby in the cameraH > business. Digital cameras are more than a few chips and a USB port andG > PC software. While the film may have been replaced with an electronic A > sensor, the lens is still a very important part, as is exposure < > metering. HP doesn't have the expertise in that area yet.  > I > Also, HP doesn't have the facilities to make its own lenses. One of the H > big reasons for an SLR is to have interchangeable lenses, something HPH > could not afford because it would be such low volume for it. (and very > expensive to setup). > E > And perhaps more importantly: HP's distribution is through comsumer J > grade electronics stores. It doesn't have anything setup with real photo( > shops to distribute high end cameras.  > A > HP *might* evolve its cameras over time and eventually be taken I > seriously enough that HP might venture into higher end models. But this H > isn't something done overnight just because Carly passed an edict that* > making cameras would increase ink sales. > L > Nikon, Hasselblad and Carl Zeiss didn't build their reputations overnight. >  > G >>I'm still fond of my d|i|g|i|t|a|l LNCO2.  The only problem is that I 3 >>cannot find a source anymore for its consumables.  >  > A > Is that a dot matrix with colour ribbons ? Or is it ink based ?   B Wife and I were talking about getting a digital camera, but since H neither of us have any knowledge about these, what does one look for in 4 a good digital camera in regards to picture clarity?9 I used to have a nice Cannon 35mm, but it since has died.      --   Where are we going?   And why am I in this handbasket?   ------------------------------  % Date: Sat, 28 Jan 2006 03:07:54 +0100 ( From: Paul Sture <paul.sture@bluewin.ch>C Subject: Re: HP, Intel becoming laughing stock of computer industry , Message-ID: <4401vrF1pk0hhU1@individual.net>   GreyCloud wrote:   <snip>  D > Wife and I were talking about getting a digital camera, but since J > neither of us have any knowledge about these, what does one look for in 6 > a good digital camera in regards to picture clarity?; > I used to have a nice Cannon 35mm, but it since has died.  >   F It depends what you want to use it for. Is picture clarity your first E and foremost priority or do you want a handy slip-in-the-pocket item?   F FWIW in the digital camera field I started out with a relatively bulkyH one which took over 30 seconds to remove the lens cap and power up, but C the one I have now fits in my shirt pocket and fires up in about 3  E seconds, so is better for spontaneous shots, though doesn't have the  " same (subjective) picture quality.  G I'd recommend sticking with a brand with a long standing reputation in  G the photography field, because they understand lens and colour quality.   C www.dpreview.com is a useful source of digital camera reviews, and  / others here can no doubt recommend other sites.    ------------------------------  % Date: Fri, 27 Jan 2006 19:52:51 -0800 # From: "Tom Linden" <tom@kednos.com> C Subject: Re: HP, Intel becoming laughing stock of computer industry ( Message-ID: <ops32k2dy0zgicya@hyrrokkin>  H On Sat, 28 Jan 2006 03:07:54 +0100, Paul Sture <paul.sture@bluewin.ch>   wrote:   > GreyCloud wrote: >  > <snip> > F >> Wife and I were talking about getting a digital camera, but since  L >> neither of us have any knowledge about these, what does one look for in  7 >> a good digital camera in regards to picture clarity? < >> I used to have a nice Cannon 35mm, but it since has died. >> > I > It depends what you want to use it for. Is picture clarity your first   G > and foremost priority or do you want a handy slip-in-the-pocket item?  > H > FWIW in the digital camera field I started out with a relatively bulkyK > one which took over 30 seconds to remove the lens cap and power up, but   F > the one I have now fits in my shirt pocket and fires up in about 3  H > seconds, so is better for spontaneous shots, though doesn't have the  $ > same (subjective) picture quality. > J > I'd recommend sticking with a brand with a long standing reputation in  I > the photography field, because they understand lens and colour quality.  > F > www.dpreview.com is a useful source of digital camera reviews, and  1 > others here can no doubt recommend other sites. K I have been quite please with Canon Power Shot G6 and the software it comes  with.    ------------------------------  % Date: Fri, 27 Jan 2006 22:07:22 -0500 ' From: Dave Froble <davef@tsoft-inc.com> * Subject: Re: HSJ50 hardware documentation?/ Message-ID: <K7KdnVIes6AmQkfeRVn-uw@libcom.com>    Ian King wrote:  > Ian King wrote:  > E >> Good morning, all!  I'm taking a headlong plunge into a clustered  F >> world, with my recent purchase of a VAX 6660.  I also picked up an H >> HSJ52 and a RAID controller; the VAX has a CI controller and, once I B >> can find a star coupler, I want to hook it all up.  I'm having F >> difficulty, however, finding hardware documentation on the HSJ50s, J >> despite such wonderful resources as bitsavers.org (thanks, Al!).  Does D >> anyone have a pointer to an online resource?  Failing that, does H >> anyone have the documentation in hard copy that they would either be H >> willing to part with or have copied?  I'd gladly reimburse for costs. >> > G > Many, many thanks to everyone who responded!  I think I'm good to go  I > now.  I've powered up the unit with a terminal attached, it appears to  I > complete all power-up tests and I can talk to it with the CLI.  It was   > last shut down in 2000!  > I > Now I'm waiting for the 6660 to arrive (oh, and I'm looking for a star  G > coupler).  And I'm doing some wiring, so I have 220v available in my  F > computer room.  Fortunately, I've replaced both my electric clothes J > dryer and range with gas appliances, so I have additional capacity.  :-) >  > Cheers -- Ian  >   = Do you understand the difference between 220 and 3-phase 220?   5 I'm thinking all the 6000 class systems used 3-phase.    --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Fri, 27 Jan 2006 21:19:24 -0600 % From: Dan Foster <usenet@evilphb.org> * Subject: Re: HSJ50 hardware documentation?5 Message-ID: <slrndtlols.h0l.usenet@zappy.catbert.org>   X In article <K7KdnVIes6AmQkfeRVn-uw@libcom.com>, Dave Froble <davef@tsoft-inc.com> wrote: > ? > Do you understand the difference between 220 and 3-phase 220?  > 7 > I'm thinking all the 6000 class systems used 3-phase.   F A fair point, though there's been someone who's successfully done thatH special 3-phase to 2-phase conversion for one of the 6000s for home use.  H Did require some electricial rewiring on the VAX side that isn't for the faint of heart, though.    -Dan   ------------------------------  % Date: Fri, 27 Jan 2006 15:00:06 -0500 ( From: Bill Todd <billtodd@metrocast.net>/ Subject: Re: Intel/Hp spend $10 billion on IA64 G Message-ID: <BvGdnYH4v99a5kfenZ2dnUVZ_tydnZ2d@metrocastcablevision.com>    Bill Gunshannon wrote:2 > In article <11tke85n7un8e7d@corp.supernews.com>,, > 	Dave Froble <davef@tsoft-inc.com> writes: >> FredK wrote: 9 >>> "Bill Gunshannon" <bill@cs.uofs.edu> wrote in message * >>> news:43upnbF1pm66dU1@individual.net... >>>  >>> 5 >>>> One thing has to be kept in sight, while the way E >>>> things sit right now the failure of Itanium means the failure of . >>>> VMS, the inverse is not necessarily true. >>> N >>> With this, I agree.  So if you want to ensure the failure of VMS, root for >>> the failure of Itanium. K >> I don't think you'll find anyone in c.o.v who wants the itanic to fail.   >>   Well, maybe Bill.   >  > Don't shoot the messenger.  F Not clear that he is:  I thought he meant me.  For the record, I have E not so much the desire to see Itanic fail (though I do feel it would  G represent just desserts for those who so shamelessly over-hyped it and  C screwed others in the process) as the belief that it well may fail.   G My main interest is in making certain that any success it enjoys rests  A on whatever actual merit it may have rather than on mindless (or  E agenda-driven) boosterism:  while the eyes of the industry certainly  G seem to have opened over the past 4+ years, there's still a great deal  B of spin out there regarding things like success rate (taking over I existing markets from products that were specifically sacrificed to make  G way for Itanic does *not* constitute any proof of industry acceptance:  I it requires taking market share for actual *competitors* to do that) and  B performance (in commercial large-system benchmarks POWER5 utterly I demolishes Itanic and x86, SPARC64, and even in some areas PA-RISC equal  E or somewhat surpass it, especially on a per-core basis, and Itanic's  I performance in real-world use - as distinguished from benchmarks tweaked  4 to perfection - is allegedly even less competitive).  I In other words, just as some of us suggested 4+ years ago Itanic remains  D sandwiched between established (and in at least one case decisively H superior) high-end competition on the one hand and the ever-growing x86 E commodity market on the other:  we just had expected Alpha to be the  D decisively superior high-end product, but in Alpha's absence POWER5 B seems to be filling that role adequately if perhaps somewhat less 
 impressively.    - bill   ------------------------------  % Date: Fri, 27 Jan 2006 16:12:57 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> / Subject: Re: Intel/Hp spend $10 billion on IA64 , Message-ID: <43DA8CD5.DCD6FC61@teksavvy.com>   FredK wrote:L > With this, I agree.  So if you want to ensure the failure of VMS, root for > the failure of Itanium.   E We want the success of VMS. And we don't see this possible as long as ; VMS is restricted to a chip with very limited market niche.   C Is that 10 billion dollar going to go to re-opening the workstation  market and low end servers ?  D The Solutions Alliance is merely a marketing organisation that sends seed money to developpers.    F And it is my opinion that very little of that 10 billion will actuallyH be spent by the alliance. I think this is more a question of starting toG reserve budgets so that when the end of IA64 is announced, the costs to E migrate to 8086 will be less on paper because they will have reserves $ already which they can assign to it.  F Also, unless any of that 10 billion goes back to Intel for engineeringI of that IA64 thing, it won't help Intel continue development of that dog.    ------------------------------  % Date: Fri, 27 Jan 2006 16:27:45 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> / Subject: Re: Intel/Hp spend $10 billion on IA64 , Message-ID: <43DA904C.F3341116@teksavvy.com>   Bill Todd wrote:G > Not clear that he is:  I thought he meant me.  For the record, I have , > not so much the desire to see Itanic fail   F I want VMS to succeed. And as long as the future of that IA64 thing is cloudy, VMS cannot succeed.   C Consider the number of discussions on VAX and alpha emulators. That E should be a good hint that people aren't interested in moving to that N IA64 thing because 8086 based machines are what really counts in the industry.  A The 8086 may have had its origins as a toy controler. And its toy H architecture may not be as elegant as Alpha, but the business reality isR that the 8086 is king, it is the industry standard and cheaper due to competition.  D One need not justify buying an 8086. But you need to provide lots ofG reasons to choose a non standard platform such as IA64, especially when 7 the media is full of stories of IA64's lack of success.   G As long as IA64 future remains uncertain, VMS cannot succeed. So either L HP fixed IA64's image ASAP, or it kills IA64 ASAP and moves VMS to the 8086.  C Instead of giving that alliance 10 billion bucks, HP could use that " money to do mass marketing of VMS.  G As long as HP doesn't market VMS and HP-UX openly and very visibly, the * hardware those systems runs on won't grow.  E If the goal is to grow IA64 market share, then HP needs to grow HP-UX B and VMS and make some key very visible slaes for NSK. And to do soE requires big marketing of the operating systems and applications, not  marketing of the platform.  D Alpha's lack or market penetration was easily explained by Digital's! total incompetance in marketing.    E You cannot explain IA64's lack of market penetration by either HP nor F INTEL's lack or marketing savvy. Both those organisations heavily relyE on marketing and have extensive experience and expertise that is used  for their core products.  G The current situation projects the image that HP will widdraw from IA64 H some time in the "near" future. And this leads to distrust of HP. EitherF HP uses its marketing savvy to fix the IA64 image problem once and forH all, or HP comes out with the truth and announces the end of IA64 sooner, rather than later and ports VMS to the 8086.   ------------------------------    Date: 27 Jan 2006 16:09:08 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) / Subject: Re: Intel/Hp spend $10 billion on IA64 3 Message-ID: <kur+CzdkSD7X@eisner.encompasserve.org>   \ In article <43DA8CD5.DCD6FC61@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes: > FredK wrote:M >> With this, I agree.  So if you want to ensure the failure of VMS, root for  >> the failure of Itanium. > G > We want the success of VMS. And we don't see this possible as long as = > VMS is restricted to a chip with very limited market niche.   H Where "we" is the subset of people on VMS who keep kvetching about this.   ------------------------------  % Date: Fri, 27 Jan 2006 22:33:02 -0500 ' From: Dave Froble <davef@tsoft-inc.com> / Subject: Re: Intel/Hp spend $10 billion on IA64 / Message-ID: <PO6dncKXf4IieEfeRVn-hg@libcom.com>    JF Mezei wrote:  > Bill Todd wrote: > G >>Not clear that he is:  I thought he meant me.  For the record, I have , >>not so much the desire to see Itanic fail    Well, yeah.   H > I want VMS to succeed. And as long as the future of that IA64 thing is > cloudy, VMS cannot succeed.   6 While succeed would be nice, I just hope for survival.  E > Consider the number of discussions on VAX and alpha emulators. That G > should be a good hint that people aren't interested in moving to that P > IA64 thing because 8086 based machines are what really counts in the industry.  C It means something else to me.  It means that even in the last 6-7  E years, VMS is/was a successful OS, and it's platforms, VAX and Alpha  E were also successful.  Unless those developing emulators are smoking  G some of the good stuff, they have reason to believe that there remains  I significant demand for those platforms, regardless of the companies that  # have truncated the future of Alpha.   C > The 8086 may have had its origins as a toy controler. And its toy J > architecture may not be as elegant as Alpha, but the business reality isT > that the 8086 is king, it is the industry standard and cheaper due to competition.  H Looking a bit deeper into that concept, I'm thinking of the disasterous G thinking of the non-technical people we have suffered.  They looked at  G Alpha and didn't see the leader in volume sales, and therefore decided  E that the platform had no value.  Thus, kill it off, and go with what  , Intel will make the 'new' industry standard.  3 No need to again go into what AMD did to that plan.   E The flaw in their thinking would be similar to auto companies seeing  F that the high end models were not the volume leaders, and deciding to D kill them off.  Cadillac, Lincoln, and the more exotic sports cars. E However those are high profit autos, and they also add prestige to a  & manufacturer.  They are good business.  E Alpha and VMS are/were good business.  Never the volume desktop, but  J high profit and fill specific demand.  This is what the idiots didn't see.  F > One need not justify buying an 8086. But you need to provide lots ofI > reasons to choose a non standard platform such as IA64, especially when 9 > the media is full of stories of IA64's lack of success.   H But in support of the above, IBM does rather well with Power5, and will B continue to do well with it's successors.  There is good money in 7 services, and there isn't much service business in PCs.    --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Fri, 27 Jan 2006 22:37:21 -0500 ' From: Dave Froble <davef@tsoft-inc.com> / Subject: Re: Intel/Hp spend $10 billion on IA64 / Message-ID: <PO6dnf2Xf4Jce0feRVn-hg@libcom.com>    Bill Gunshannon wrote:, > In article <43DA60AF.5040306@comcast.net>,* > 	BRAD <bradhamilton@comcast.net> writes: >  >>Bill Gunshannon wrote: >>3 >>>In article <11tke85n7un8e7d@corp.supernews.com>, - >>>	Dave Froble <davef@tsoft-inc.com> writes:  >>>  >>>  >>>>FredK wrote: >>>> >>>>O >>>>>With this, I agree.  So if you want to ensure the failure of VMS, root for  >>>>>the failure of Itanium. >>>>L >>>>I don't think you'll find anyone in c.o.v who wants the itanic to fail.  >>>> Well, maybe Bill.   >>>  >>> H >>>Don't shoot the messenger.  I really don't care one way or the other. >>+ >>Wrong Bill - I think he meant Mr. Todd...  >><snip> >  > E > Could be, but it was a comment from me that he was replying to so I  > just assumed......    ;-)   B Nope, it was to Fred I was replying, and yes, Bill Todd was in my H thoughts.  Not saying that he does or doesn't, just allowing that there ( may be some who want to see itanic sink.   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Fri, 27 Jan 2006 22:56:08 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> / Subject: Re: Intel/Hp spend $10 billion on IA64 , Message-ID: <43DAEB4D.F545D075@teksavvy.com>   Dave Froble wrote:C > Nope, it was to Fred I was replying, and yes, Bill Todd was in my I > thoughts.  Not saying that he does or doesn't, just allowing that there * > may be some who want to see itanic sink.    A If HP/Intel,s plans are to sink it, then they might as well do it 8 quickly instead of prolong the speculation indefinitely.   ------------------------------  % Date: Fri, 27 Jan 2006 16:03:25 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 1 Subject: Re: problems with a new vms installation , Message-ID: <43DA8A9A.644B2566@teksavvy.com>  ! michael.grunditz@telia.com wrote: " > UCX                          DEC# >  UCX-IP-CLIENT                DEC $ >   UCX-IP-NFS                   DEC% >    UCX-IP-RT                    DEC   J You may wish to LICENSE DISABLE every one except UCX. UCX is all you need.  5 > Compaq TCP/IP Services for OpenVMS VAX Version V5.1 0 >   on a VAXstation 4000-90 running OpenVMS V7.3  D When was XDM introduced ? I thought it was with TCPIP Services 5.3 ?  : >   The output from TCPIP SET CONF ENABLE NOSERVICE XDM is@ >   %TCPIP-E-CONFIGERROR, error processing configuration request> >   -TCPIP-E-INVPARM, invalid parameter value for Service name  >   -RMS-E-RNF, record not found     If you:   ( SEARCH SYS$SYSTEM:TCPIP$SERVICE.DAT XDM    does it find a record in it ?   G Not knowing more about your setup, one  "Windows like" suggestion would  be to:   $DELETE SYS$SYSTEM:TCPIP*.DAT;*   3 And then to @SYS$MANAGER:TCPIP$CONFIG from scratch.    Another thing would be to  $SET VERIFY  $@SYS$MANAGER:TCPIP$CONFIG  H it will be ugly, but eventually you get to type in the option number forC XDM and you'll then see why the error message is issued. With TCPIP D Services 5.3, the "Invalid configuration option selected, please tryH again" is issued when a symbol has not been set to 1. This symbol is set@ by the various subroutines, each corresponding to a menu option.  < Looking at the verify output, you'd see what really happens.   ------------------------------  # Date: Sat, 28 Jan 2006 05:38:08 GMT . From: Larry Finger <Larry.Finger@lwfinger.net>! Subject: Re: SimH V3.5-2 released H Message-ID: <4rDCf.303805$qk4.264336@bgtnsc05-news.ops.worldnet.att.net>   Timothy Stark wrote:@ > "vaxorcist" <hoelscher-kirchbrak@freenet.de> wrote in message ? > news:1138098235.825842.155470@o13g2000cwo.googlegroups.com...  > < >>Anyone who wants to try one of the following VMS versions:* >>- 3.0, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7 >>- 4.4, 4.5, 4.6, 4.7, >>I'll be glad to supply those listed above. >  > I > Well, I tried to e-mail him about that but he did not seem response my  M > e-mail message. I think that his address might be spam-proof.   Vaxorcist,  * > please check my e-mail message.  Thanks. > 	 > Thanks,  > Tim   ' I also e-mailed him about getting V4.7.    Larry    ------------------------------   Date: 27 Jan 2006 20:56:28 GMT' From: ruben@CCNMR.MIT.EDU (David Ruben) B Subject: Re: TCPIP Services: management of cluster alias interface? Message-ID: <43da88fc$0$556$b45e6eb0@senator-bedfellow.mit.edu>   \ In article <43DA0292.9238CB60@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes: >An update on the issue... >  >I read the RFC on ARP.  > A >When Node A goes down and node B takes over the cluster alias IP G >address, node B cannot send a simple broadcast to tell other nodes its  >new ethernet address. > E >for node C to know the new ethernet address of the cluster alias IP:  > I >Node B must send a ARP broadcast asking C to identify itself. C receives F >the packet, sees its an ARP that it must answer, and in doing so willH >automatically update its own ARP table with the IP-ethernet combination >sent by B.  > G >Node D will ignore this activity because it is not directed to itself. G >Thefore D will not update its ARP table with the new etherenet address ' >corresponding to the cluster alias IP.  > L >The taking over of an IP by a new node is explicetely mentioned in the RFC. > G >Thefore, for the cluster alias changeover to work, node B must send an E >ARP request to each correspondant knows about on the LAN, which will 3 >cause each correspondant to update its ARP table.   >   D There is an easy solution to this that many computers already use. IH ran across this while using Ethereal on Linux to monitor the ARP traffic on my network.  @ Occasionally Ethereal would report a "gratuitous ARP request". AH gratuitous ARP request is something like "who has address 192.168.0.1? -F tell 192.168.0.1". This didn't make any sense to me and I assumed thatC it was just another bug in a Windows machine. Later I ran across an C article that said that all nodes should, not must, send such an ARP G request when they configure or reconfigure their interfaces. The reason B is obvious if you look at the ARP packet in detail. In addition toG fields for the target ip address and hardware address, there are fields / for the source ip address and hardware address.   H An ARP request is a broadcast packet and is therefore seen by all nodes.H The target node is required to respond with the target hardware address,D but all nodes should, but not must, update their ARP caches with the& information in the source node fields.  G Thus, if a node which reconfigures its interface sends a gratuitous ARP F packet and all other nodes do in fact update their ARP caches with theF information in the source node fields, then all nodes will immediately7 be able to communicate with the reconfigured interface.    D.R.   ------------------------------  % Date: Fri, 27 Jan 2006 20:07:51 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> B Subject: Re: TCPIP Services: management of cluster alias interface, Message-ID: <43DAC3E6.93AEF209@teksavvy.com>   David Ruben wrote:B > Occasionally Ethereal would report a "gratuitous ARP request". AJ > gratuitous ARP request is something like "who has address 192.168.0.1? - > tell 192.168.0.1".    G Ok., I reread the RFC. And there is a less "textual" portion which does $ deal with such "gratuitous" updates:  / ftp://ftp.rfc-editor.org/in-notes/std/std37.txt   ' ?Do I have the hardware type in ar$hrd?  Yes: (almost definitely)/   [optionally check the hardware length ar$hln] %   ?Do I speak the protocol in ar$pro?    Yes:1     [optionally check the protocol length ar$pln]      Merge_flag := false ;     If the pair <protocol type, sender protocol address> is :         already in my translation table, update the sender8         hardware address field of the entry with the new>         information in the packet and set Merge_flag to true. &     ?Am I the target protocol address?     Yes:=       If Merge_flag is false, add the triplet <protocol type, >           sender protocol address, sender hardware address> to            the translation table.    H So it does seem that all nodes should update their own ARP table when anD IP sends out a broadcast request. So now, I'll have to check out theD tracing to see if VMS really does send an ARP when the cluster alias switches nodes   ------------------------------  % Date: Fri, 27 Jan 2006 17:23:43 -0500 ) From: "Neil Rieck" <n.rieck@sympatico.ca> H Subject: The processor wars Part V, Intel's Itanium: To be or not to be?: Message-ID: <K3xCf.16996$ft2.293690@news20.bellglobal.com>  I For all you chip heads out there, part 5/5 was posted today (2006-01-27). 7 http://www.digitimes.com/bits_chips/a20060127PR201.html   G Please note that this same site contains another article titled: "Intel J Itanium gets US$10 billion new lease on life". Itanium Solutions Alliance?F This sounds a lot like Alpha Alliance between Digital and Samsung etc.2 http://www.digitimes.com/mobos/a20060127PR204.html    
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada.8 http://www3.sympatico.ca/n.rieck/links/cool_openvms.html   ------------------------------   End of INFO-VAX 2006.055 ************************