1 INFO-VAX	Sun, 10 Jun 2001	Volume 2001 : Issue 320       Contents:A Re: "Talk Baq" to Compaq - but boycott www.compaqworkinggroup.com  Re: $QIO and TCP/IP server ACMS & DECtrace  Re: ACMS & DECtrace  Re: Bloated software3 Re: DECamds vs Availability Manager - Some Findings ? Re: DNS questions 1:M host_name:host_address (address ordering) 3 Re: esa12000fc and ema12000 differences/experiences 6 Re: Low level format of SCSI disk, so VMS can read it." RE: NCP to NCL command equivilants Re: The future of VMS  Re: The future of VMS  Re: The future of VMS  RE: The future of VMS  Re: The future of VMS  Re: The future of VMS  Re: The future of VMS - Re: Urgent: Oracle 8 parallel server on VMS 7 < Re: Use of SYSMAN to keep password file in sync on 8 systems% VDE on FreeWare 4. Manuals unreadable / Re: VMS Installation and set up (from a newbie) ) Re: [OT] Current Microsoft v DOJ status ? ) Re: [OT] Current Microsoft v DOJ status ?   F ----------------------------------------------------------------------    Date: 10 Jun 2001 09:52:24 -05009 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) J Subject: Re: "Talk Baq" to Compaq - but boycott www.compaqworkinggroup.com3 Message-ID: <DfagQ7LtqiHx@eisner.encompasserve.org>   q In article <CSqU6.3272$Tc.648974@typhoon.ne.mediaone.net>, "Terry C. Shannon" <terryshannon@mediaone.net> writes:   H > And if you're of the opinion that Compaq's marketing and direction are$ > wrong, by all means tell 'em so at > # > http://www.compaqworkinggroup.org  > J > Capellas and Elias and Marcello, et al don't read newsgroups and I doubtL > they pay much attention to fan mail over the Internet. They will, however,7 > see the results of the survey being conducted online.   E No, send them a paper letter (proving you are not restricted to using D only electronic media) and tell them to stop trying to undermine theG customer's browser security with the current www.compaqworkinggroup.org  implementation.    ------------------------------  % Date: Sun, 10 Jun 2001 17:05:30 +0100 , From: "Richard Maher" <maher_rj@hotmail.c*m># Subject: Re: $QIO and TCP/IP server 2 Message-ID: <9g07j1$q1$1@neptunium.btinternet.com>   Hi,   - From: JF Mezei (jfmezei.spamnot@videotron.ca)  Date: 2001-06-07 07:30:07 PST   D >If you use a coordinating process, you have to either find a way to >"transfer" E >incoming calls to one of the processes in the pool so that the TCPIP L >connection is established directly between the remote client and one of theH >server processes, OR, your coordinator has to act as a tunnel, with theJ >ability to funnel data back and forth in with multiple connections at the >same time.   H With UCX it's easy to "transfer" or make the socket available to anotherC process by setting the ucx$c_share socket option. The other process B can simply $assign a channel to the _BGnnnn: device. NB: This does= *not* work with TCPware. IIRC Hunter thought it would be nice D to have for some future release but it would be a back-burner thing.  @ Your second or "tunnel" option is how TIER3 is implemented and IA obviously believe that there are many definite advantages to this C approach and little or no down-side (certainly _not_ performance!), H but each to his/her own. I'll send a snapshot of the Tier3 Configuration? Maintenance Screen under separate cover so you can get a better E idea of the performance tuning options available to you but briefly:- < For each and every application TIER3 creates a Communication> Server process and a variable number (min/max/idle timeout) of$ reusable Execution Server processes.  A You tell TIER3 what's the biggest(1) message that you're going to C exchange and this is used to size the private mailboxes for each of B the Execution Servers. The "funnel", as you refer to it, manifestsC itself in the form of a further mailbox. The size of which can also + be configured between 100 and 64Kbytes (2).   ; TIER3 Communication Servers execute at the VMS priority and : quotas set in the UAF record for the TIER3$SERVER username9 (3). Execution Servers execute at the priority and quotas 9 appropriate for the username that you have chosen for the $ particular application to run under.  : Because your application may enjoy thousands of concurrentB users it's a pretty safe bet that they won't all be working at the@ same time so TIER3 let's you configure the number of i/o buffers8 that your application will need. IE: Although your TIER3: Communication Server may be listening on and managing 3000< sockets there may only be a few hundred concurrent i/o going* on so only allocate a few hundred buffers.  7 Finally, unlike some of the ODBC implementations on VMS 9 TIER3 creates a separate Communication Server / Execution < Server instance for each application. Therefore you can tune< your Cash Tills application differently to your Lotto number9 generator and run them on the same node. (TIER3 also lets ; you tune the same app differently on different nodes in the 
 same cluster)   > Sorry to have made a short story long, but in a nutshell there7 is nothing wrong and everything right with the "tunnel" $ approach when implemented correctly.   Regards Richard Maher   D (1) If the client sends a message that is bigger than "biggest" thenB the message is not lost but merely broken up and delivered to your> RECEIVE User Action Routine in successive calls. TIER3 informs4 your UAR of the message fragmentation by setting the> T3$V_MORE bit in the "flags" argument. If you are using DECnetA and would like to _send_ a message bigger than "biggest" then you = simply set the T3$V_MULTIPLE bit in the "flags" argument when @ you call the T3$SEND System Service. (This flag has no effect on? stream i/o and is therefore ignored for TCP/IP) If you normally = send messages of a few hundred bytes but occasionally need to > transceive 10Kbyte messages then it may be to your performance= advantage to set your "biggest" message at 500 bytes and cope " with the occasional fragmentation.  > (2) A large command mailbox quota is useful when you have many; Execution Servers or if your RECEIVE UAR will be generating = multiple outbound messages for each inbound message delivered : and you would like to take advantage of the T3$V_NOW flag.A TIER3 also lets you configure the size of the deletion (Execution ; Server rundown) mailbox and the Network message mailbox (if  you are using DECnet).  > (3) The TIER3$SERVER account is created when you first install5 TIER3 on your system and has a default priority of 6.    ------------------------------  # Date: Fri, 08 Jun 2001 15:25:04 GMT 2 From: robert_jm_barron@hotmail.com (Robert Barron) Subject: ACMS & DECtrace5 Message-ID: <3b20edcd.15137876@news.netvision.net.il>    Hi,   @ Does anyone have any experience with using DECTrace to tune ACMS
 applications?   F I use ACMS with servers which we wrote & I'm trying to get DECtrace to( give me information on specific servers.   Thanks in advance,   Robert.    ------------------------------  % Date: Sun, 10 Jun 2001 13:42:09 -0400 - From: Michael Austin <miaustin@bellsouth.net>  Subject: Re: ACMS & DECtrace- Message-ID: <3B23B171.8D309CDF@bellsouth.net>   I Once you have collected the data, what did you do with it?  You should be A able to load the information into a database and look at specific  information from there.   & Michael Austin -- Currently available.= DBA Consultant (and former ACMS Support/Rdb Support Engineer)    Robert Barron wrote:   > Hi,  > B > Does anyone have any experience with using DECTrace to tune ACMS > applications?  > H > I use ACMS with servers which we wrote & I'm trying to get DECtrace to* > give me information on specific servers. >  > Thanks in advance, > 	 > Robert.    ------------------------------    Date: 11 Jun 2001 00:25:17 +0800, From: Paul Repacholi <prep@prep.synonet.com> Subject: Re: Bloated software - Message-ID: <87k82kqp3m.fsf@prep.synonet.com>   + fabio_compaq@ep-bc.petrobras.com.br writes:   A > You are ABSOLUTELY RIGHT in your comments about development and  > resources!  	 > But ...   ? > About ALLIN1 : I really didnt like ! It was heavy and machine C > consumer....  DEC was not really sucessfull with this product.  I C > think OpenVMS servers cannot be "Office Servers" , because office C > stuff dont need high availabily in most of cases, so a WNT box is  > enough ! ! !  8 Oh please... There where an enormouse number of allinoneA installations, and many of the very large ones. Many have gone to F getting billyware to try to do the job. I know of no case that has not( been a finacial and functional disaster.  A If a system like thay is US, then a large part of your busness is 5 dead, and customers etc are walking away cursing you.    --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov     ------------------------------    Date: 11 Jun 2001 01:23:16 +0800, From: Paul Repacholi <prep@prep.synonet.com>< Subject: Re: DECamds vs Availability Manager - Some Findings- Message-ID: <877kykqmez.fsf@prep.synonet.com>   + rob.buxton@wcc.govt.nz (Rob Buxton) writes:   	 > Hi All,   F > Thought this might be of interest if anyone is looking at DECamds or > Availability Manager.   F > Got hold of Availability Manager 2.0 (VMS & NT variants) and DECamds > 7.3A.   E > Installed the Availability Manager NT variant on my NT Workstation. D > Worked fine, picked up all of the DECamds Data Collectors, running0 > various flavours of DECamds 7.2, 7.2-1 & 7.3A.  H > Installed the VMS version of the Availability Manager on a small AlphaB > (A DEC3000 with 64MB of Memory) This is well below the suggestedE > config but I was only running the data Collector on 8-9 nodes. This B > machine just could not handle the load, CPU usage up to 95-100%,E > performance was abysmal. Must mention that I did this from an REXEC  > session from eXcursion.   F It won't handle 2 node... It loses the path, sso poof go you displays,6 and you then wait for 10 min for them to be re-inited.  B > Using DECamds on the same VMS machine and again using REXEC from > excursion is fine.    A With AM, constant soft and hard paging, 30-70% CPU used. AMDS you  can't find in CPU use!  B > I installed the Availability Manager Data Collector modules on aE > couple of VAXes and I can't see any performance differences between 6 > the DECamds & Availability Manager Data Collectors.   / > The Analyser, well there's a huge difference.   > There are other issues as well as it's performance 'problems'.  F It is now all one window, so takes up muck more space, and you can notF monitor lots of combinations as they are on different pages of the oneD display. Also, many data items seem to have now gone, so forget them
 all together.   D God knows how you are expected to have it be any use to you when theD systems are in trouble. I real fair weather friend if ever there was one.   Not even zero out of ten.    Avoid. Use AMDS instead.  A Oh, did we mention that you get yet another copy of Java? Looking C more and more like the world of gates-shit every day. In every way.    --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov     ------------------------------  % Date: Sun, 10 Jun 2001 12:09:54 +0100 , From: "Richard Maher" <maher_rj@hotmail.c*m>H Subject: Re: DNS questions 1:M host_name:host_address (address ordering)2 Message-ID: <9fvkf7$n0$1@neptunium.btinternet.com>   Hi Paul,   Thanks very much for the reply.   L I have to admit to being disappointed (nay gutted!) at the lack of a uniformJ method of accomplishing the completely resonable task of returning a givenI host's addresses in a predictable order. So much for "open" systems. (And J surely this has to have an effect on the TCP/IP DNS cluster load balancing? claims of TCP/IP 5.1? Cacheing servers? Resolver sorting?. . .)   K What I was looking for was a poor man's failover. Not just because it would G be easy to implement but also *fast*. Now I'll have to go down the more D robust and transparent (and slower) road of every node accepting allG incoming connections and routing subsequent messages around the cluster J internally. I hope VMS Intra-Cluster Communication Services are as fast as they are purported to be.   E Are they really so fast that they can still quite happily be deployed  between ; client and server even when they are both on the same node?    Regards Richard Maher   > PS. In *June* London is still colder and wetter than Perth :-(   Message 4 in thread , From: Paul Repacholi (prep@prep.synonet.com)H Subject: Re: DNS questions 1:M host_name:host_address (address ordering) Newsgroups: comp.os.vms  Date: 2001-05-26 09:48:33 PST   . "Richard Maher" <maher_rj@hotmail.c*m> writes:  D > "When I have a DNS hostname that maps to more than one IP address,B > *what do I have to do* to *guarantee* that DNS will return those+ > address in the order that I have chosen?"   A None, it is implementaion and phase of moon dependant. IE, the RR D could come from some other DNS that sends it out in its idear of theD right way. Then you have what ever the systems resolver code does to it.   # > Is there an RFC that covers this?   G Yes, several. But better to get the BIND stuff, since that is what most  people use for DNS now.   C > Like Telnet, RSH et al. people who will connect to my application @ > will do a gethostbyname() and then proceed down the list of IPD > addresses until they find one that works. I *need* to control (andF > insist upon) the IP address/node that will be the *first* connection > candidate?  E Then you will have to do it. BTW, you can't depend on getting a list. ( You may only get one member of the list.   --< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   Message 3 in thread * From: Richard Maher (maher_rj@hotmail.c*m)H Subject: Re: DNS questions 1:M host_name:host_address (address ordering) Newsgroups: comp.os.vms  Date: 2001-05-25 14:53:18 PST   
 Hi Didier,  E Thanks for the reply and I hope you can help. You're right my earlier I question was a bit of a Dog's Breakfast so I'll try and be more succinct.   
 Question:-  B "When I have a DNS hostname that maps to more than one IP address,@ *what do I have to do* to *guarantee* that DNS will return those) address in the order that I have chosen?"   ! Is there an RFC that covers this?   ? Does it differ from TCP/IP Services to TCPWARE? VMS to Solaris?   F Like Telnet, RSH et al. people who will connect to my application willC do a gethostbyname() and then proceed down the list of IP addresses I until they find one that works. I *need* to control (and insist upon) the > IP address/node that will be the *first* connection candidate?  H Alles Klar? I will happily provide any other information you require but7 I won't muddy the waters by going over old ground here.    Regards Richard Maher.  L PS. I would have replied sooner but I can only see your reply via Google and# not via my ISP and outlook express.    Message 2 in thread , From: Didier Morandi (Didier.Morandi@gmx.ch)H Subject: Re: DNS questions 1:M host_name:host_address (address ordering) Newsgroups: comp.os.vms  Date: 2001-05-24 13:15:32 PST    Richard Maher wrote: >  > Hi,  > G > Sorry if you've already seen this, but I can see my original post via L > outlook but not google. It might be the attachment or it might be replying  L > to my own note (with an example) but if there's anyone here that knows DNS! > I'd really appreciate the help!    Richard,  @ If you wish a reply, maybe you should ask your pal to write easyH questions instead of a summary of the differences between his experience and the O'Reilly book.  H I attended the Digital TCP/IP Services for OpenVMS training recently andD still have my notes and other resources under the hand, but I didn't, succeed to understant what was the question.  + Why is this gentleman bothering with sorts?e   D. -- Compaq Computer (Schweiz) GmbH! berlandstrasse 1, 8600 Dbendorfe GSM: +41 79 7054670c   ------------------------------    Date: 11 Jun 2001 00:09:30 +0800, From: Paul Repacholi <prep@prep.synonet.com>< Subject: Re: esa12000fc and ema12000 differences/experiences- Message-ID: <87snh8qptx.fsf@prep.synonet.com>   - young_r@encompasserve.org (Rob Young) writes:   ^ > In article <874rtqv1fy.fsf@prep.synonet.com>, Paul Repacholi <prep@prep.synonet.com> writes:  D > > If I can ask a favour of anyone who may have an EMA with nothingA > > to do just yet :) And who would not mind running a few tests.e  F > > If you could lash up a set or 1,2,3,4,5 member raid sets, with allC > > the members of a set on ONE scsi bus, then see what you can get-E > > out of them. The historical constant for max performance has beengB > > 2, or rarely 3 drives per bus, but I'd like to have a check onF > > it. Last time I saw this sort of test done was about 8 years or so > > ago.  ; > > Cross posting to comp.arch.storage would not go astray.n  F > > Or just post the kit to the address below and I'll do the tests :)  D > Just curious as to why in the world would you want to do that?  ItE > doesn't make sense from a couple of directions.  First and foremost F > SCSI arbitration is an issue with *about* 3 or more active drives onF > a bus.  Second and most obvious, the bus goes, so goes your RAID setF > and you and I are too careful to do such things like that (have more/ > than 1 member of a RAID set on a single bus).o  # > What are you trying to "measure?"   F First off, I hope no one think I mean to set up a normal raid set thet way!!   ? Years ago, people would test the IO systems to find out the max-C performance configuration. This provides some guidelines if you are 6 adding drives if you should also get more controllers.  E The answers always came out as two or three, across many drive buses,eF except for the one reported set of numbers for Massbus. The last I wasD was SCSI-2 vintage, and no one seems to have done it with later SCSI of FC drive.  B Maybe SCSI arbitration has improved so 4 is the max, I don't know,1 and I don't know of anywhere that has the answer.i  F Not, setting up the stripe set is so you just run the one test against@ the stripe, and don't have to acount for the number of drives in@ it. Put them on one bus, cause that is what you want to measure.  B I think the peak performance will be for 2 or 3 drives/bus, still,( but it would be nice to know, not guess.   -- s< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.w@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov n   ------------------------------    Date: 11 Jun 2001 01:09:53 +0800, From: Paul Repacholi <prep@prep.synonet.com>? Subject: Re: Low level format of SCSI disk, so VMS can read it.e- Message-ID: <87bsnwqn1a.fsf@prep.synonet.com>t  * Terry Kennedy <terry@gate.tmk.com> writes:  0 > Paul Repacholi <prep@prep.synonet.com> writes:) > > Jim Agnew <Agnew@hsc.vcu.edu> writes:i1 > >> i'm told later versions vms can handle this.d > >m. > > Tolerate, with complaints would be nearer.  ? >   I'm pretty sure that the driver was changed to soft-set the E > changeable mode page if it found auto-reallocate set, not to ignore B > the fact that it was set. I'm pretty sure Glenn Everhart did the9 > change while at DEC, but I'm not 100% positive of that.c  G Glen made changes to allow mounting the drives, and the shadow code waso9 changed to warn you if you mounted one into a shadow set.h  G > > If the drive replaces the failed block, you have NO indication thataH > > the data is bogus. If the data can not be recovered and replace, VMSI > > marks the block as invalid till it is written to. This is *essential*i > > for data integrity.  > H >   I beg to differ. A drive that auto-reallocates data it can't read isL > not compliant with the SCSI 2 specification. See 9.3.3.6 Read-write error  > recovery page:  A Well, may I add the small postscript that even though that is notnD standard comformant, it is pretty 'industry standard'. Though I hope it is vanishing fast!a   ...   E >   Sure, there are broken implementations out there - but stuff likesA > broken TCQ support is *far* more common, and most of the broken F > drives have either had new firmware available or are older, obsolete	 > models.l   I'd agree with that.  G > Certainly, most vendors would not ship generic drives with these bits B > set-by-default if they were aware of implementation defects that8 > caused those drives to lose data during reallocations.  E You seem to have more faith in 'lowest possible cost' vendors than I.l  G BTW, I have had a read replace fail on a Seagate Barraduda 4GB. I drivefD that has been otherwise excelent. I did remove the failed block, butC did not replace it, thus leavinf a non-existant block in the middle E of the disk. :( I had to reformat to recover. It also totally screwed . up shadowing with no warning or OPCOM message!   -- t< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.y@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov s   ------------------------------  % Date: Sun, 10 Jun 2001 08:55:45 -0500o0 From: arturo saavedra <arturo.saavedra@wcom.com>+ Subject: RE: NCP to NCL command equivilantsoC Message-ID: <MOEAJKGGEIMGCCPEPJBHKEAFEMAA.arturo.saavedra@wcom.com>n  & To display corresponding NCL commands:   $ mcr net$mgmt  K When X-windows interface appears, go to Options, Default Actions and select K the box [] display ncl commands.  This will display the NCL commands on the D terminal session from where you executed the $ mcr net$mgmt command.       arturo saavedra/vmst       -----Original Message-----9 From: Robert Deininger [mailto:rdeininger@mindspring.com]r& Sent: Saturday, June 09, 2001 11:23 AM To: Info-VAX@Mvb.Saic.Com + Subject: Re: NCP to NCL command equivilants,    @ In article <VA.000003be.467b124d@sture.ch>, paul@sture.ch wrote:   > In articleA <rdeininger-0806012358450001@user-2ive75t.dialup.mindspring.com>,e > Robert Deininger wrote:n  J > > There's also the x-windows interface to NCL, which can be told to spit outS4 > > ncl command corresponding to whatever you click. > >oJ > Has anyone ever found that utility of any use except for read only mode?   Well, I haven't.   ..  H > I just went through this last week when I was setting up a development clusterlL > with multiple versions of VMS. I discovered that the MOP info had been setH > up to load the system image from DISK$LABEL: instead of SYS$SYSDEVICE: >oI > A real pain to work through the NET$CONFIGURE menu for even one client,i let = > alone multiple ones, so direct editing was the way to go...l  E It looks like the MOP part of DECnet-plus is gradually falling out ofeI favor.  The LANCP version of MOP seems to get more engineering attention,gD and is available even without DECnet.  I've stopped using the MOP in DECnet-plus.  H CLUSTER_CONFIG_LAN.COM (instead of CLUSTER_CONFIG.COM) uses LANCP behindF the scenes instead of NCL.  For small MOP jobs, like strange satelliteJ boots to upgrade firmware, I just use LANCP directly; a couple of commands$ are enough to support a simple boot.  I In a large-scale environment, DECnet-plus and NCL might be easier for MOPp stuff...  K > > I discovered fairly early that DECnet 5 is a lot easier to use once you L > > read the manuals.  Because it is slightly weird, it doesn't pay to startL > > in the middle of the manual like many in this group are probably used to
 > > doing. > > K > Presuming you do have the full manual set. But that last sentence is spoto on.e  H Well, it's on the VMS CDs, and most of the manuals are on the web.  FullD hardcopy sets are a bit scarce these days, like all the VMS manuals.@ DECnet-4 hardcopies are easy to get; just find an old gray wall.  DECnet-plus is post-gray wall...   -- Robert Deininger rdeininger@mindspring.comn   ------------------------------  % Date: Sun, 10 Jun 2001 10:20:08 +0200e) From: Christof Brass <brass@infopuls.com>  Subject: Re: The future of VMS, Message-ID: <3B232DB8.7C0E84A2@infopuls.com>   Jordan Henderson wrote:0 > . > In article <3B228FD6.50C86C16@infopuls.com>,- > Christof Brass  <brass@infopuls.com> wrote:i > >Carl Perkins wrote:   [SNIP   ? > >Maybe it is smarter to process the man page? Or combine bothsC > >sources? I never said this would be an easy task but it would be-A > >helpful because this task has to be done with every such port.r > ? > Someone earlier suggested making a getopts/CLD generator, andvA > I think that's probably the easiest task with the most benefit.p= > A lot of the GNU software uses getopts() and if it doesn't, 8 > could be made to and you'd have a good chance that the< > maintainer would take the mods.  You could put things that: > aren't taken into account in getopts() into some special? > comment convention that could be processed out of the source.n > ; > It doesn't cover 100%, but no other solution does either.a >  > -Jordan Hendersoni > jordan@greenapple.com   5 The getops solution has also the advantage to help in # establishing some sort of standard.l  @ I'm not quite sure if this conversion is all that has to be done> because what I want to achieve is to use the VMS parameter and> qualifier checking mechanism and not to let the ported program@ have them re-check because it doesn't know about running on VMS.  @ Using the man pages as a separate source for creating a CLD file& seems still to me a valuable approach.   ------------------------------    Date: 10 Jun 2001 08:49:08 -0400/ From: jordan@lisa.gemair.com (Jordan Henderson)n Subject: Re: The future of VMS* Message-ID: <9fvqc4$v08$1@lisa.gemair.com>  R In article <DC4745D1A85CA04180C83CDC706A9D186077A6@cthexc02.americas.cpqcorp.net>,* Main, Kerry <Kerry.Main@compaq.com> wrote: >Steve,  >pF >Just to expand on what you are saying about todays additional support >requirements -i >yJ >Even though the cars today are more sophisticated, faster, better qualityI >etc., you can still easily teach a 16 year old how to drive. That is noti# >much different than 10 years ago. k >rL >Similarily, the kids today can use faster systems to do certain things at a >certain level.e >i? >However, the mechanic fixing integration problems with onboardmG >microprocessors in many common vehicles today is very much a differentt8 >person than the mechanic working on cars 10 years ago.  >rL >GPS devices being able to locate cars and shutdown engines in case of theftH >. that is a totally different vehicle support requirement than tuning a >carburetor. >rL >Similarily, keeping an IT infrastructure up and running on a 24x7 basis andE >fixing things before they impact the business and complex middleware L >programming designed to be highly scalable, reliable, take advantage of theL >local platforms clustering features etc requires a whole different level ofJ >support than a simple college programmer (VB?) or admin person who has no >real experience.f >g  @ I think the car analogy is a poor one.  I believe, although I'm ? certainly no expert in this area, that maintaining today's carscA requires far fewer skill than those many years ago.  There aren't @ any carburetors anymore and things like GPS systems are all partA of the electronic system that are completely diagnosed by hookingo9 the car up to a diagnostic computer and reading the code.o  A Actually though, there does bring up a point.  Some vendors woulda@ have you believe that running their systems is no more difficult@ than reading the output of the diagnostic screens and performingA the action from the database recommendations, suggesting that thegA auto mechanic analogy is true (but flipped on it's head from whatu? you suggested).  We know that the reality of system maintenance'- today is somewhat more complicated than this.a   >:-) >l >Kerry Main  >Senior Consultant >Compaq Canada Inc.t >Professional Services >Voice: 613-592-4660 >Fax  :  819-772-7036  >Email: Kerry.Main@Compaq.comi >i >S > [big snip]   -Jordan Hendersonr jordan@greenapple.comh   ------------------------------    Date: 10 Jun 2001 11:12:19 -0400/ From: jordan@lisa.gemair.com (Jordan Henderson)n Subject: Re: The future of VMS* Message-ID: <9g02oj$6g1$1@lisa.gemair.com>  , In article <3B232DB8.7C0E84A2@infopuls.com>,+ Christof Brass  <brass@infopuls.com> wrote:o >Jordan Henderson wrote: >> t/ >> In article <3B228FD6.50C86C16@infopuls.com>,). >> Christof Brass  <brass@infopuls.com> wrote: >> >Carl Perkins wrote:h >  >[SNIP >o@ >> >Maybe it is smarter to process the man page? Or combine bothD >> >sources? I never said this would be an easy task but it would beB >> >helpful because this task has to be done with every such port. >> o@ >> Someone earlier suggested making a getopts/CLD generator, andB >> I think that's probably the easiest task with the most benefit.> >> A lot of the GNU software uses getopts() and if it doesn't,9 >> could be made to and you'd have a good chance that the,= >> maintainer would take the mods.  You could put things thatm; >> aren't taken into account in getopts() into some speciale@ >> comment convention that could be processed out of the source. >> n< >> It doesn't cover 100%, but no other solution does either. >>   >> -Jordan Henderson >> jordan@greenapple.com >)6 >The getops solution has also the advantage to help in$ >establishing some sort of standard. >oA >I'm not quite sure if this conversion is all that has to be donee? >because what I want to achieve is to use the VMS parameter and ? >qualifier checking mechanism and not to let the ported program A >have them re-check because it doesn't know about running on VMS.u >aA >Using the man pages as a separate source for creating a CLD file ' >seems still to me a valuable approach.r  A The man pages can't be counted on to have all the information you > need here.  I would recommend instead having a special comment? convention embedded in the source that would add the additional  qualifier checking.t  = Perhaps you could try to generate these special comments from @ man pages, but the coverage would be spotty at best.  Also, I've; learned from experience that specifications for programmingd> interfaces that are maintained separately from the source will? not be reliable.  Not that comments are always reliable either, A but they are more likely to be updated when the source is changed  than the man page is.    -Jordan Hendersonc jordan@greenapple.coml   ------------------------------  % Date: Sun, 10 Jun 2001 12:16:44 -0400i+ From: "Main, Kerry" <Kerry.Main@compaq.com>o Subject: RE: The future of VMSR Message-ID: <BE56C50EA024184DAF48F0B9A47F5CF48DBDDC@kaoexc01.americas.cpqcorp.net>  F >>> ..I think the car analogy is a poor one.  I believe, although I'm ? certainly no expert in this area, that maintaining today's cars 6 requires far fewer skill than those many years ago.>>>  J Ahhh .. just like computer systems today, individual components are easierI to fix (replacing a disk drive does not require a head alignment like the1G good ole days), but the overall system for auto's is becoming harder toa diagnose intermittent issues.   K Version incompatibilities, test equipment availability and how to keep themIK current, skilled resources that understand the new technologies (other thana0 the simple swap till you drop skill set) etc ...  H And like computer systems today, the auto dealers would have you believe- anyone with basic skills can fix their stuff.    :-),   Regardss    
 Kerry Main Senior Consultantu Compaq Canada Inc. Professional Servicesn Voice: 613-592-4660i Fax  :  819-772-7036 Email: Kerry.Main@Compaq.com     -----Original Message-----< From: jordan@lisa.gemair.com [mailto:jordan@lisa.gemair.com] Sent: June 10, 2001 8:49 AM  To: Info-VAX@Mvb.Saic.Comw Subject: Re: The future of VMS    
 In articleG <DC4745D1A85CA04180C83CDC706A9D186077A6@cthexc02.americas.cpqcorp.net>,s* Main, Kerry <Kerry.Main@compaq.com> wrote: >Steve,- >-F >Just to expand on what you are saying about todays additional support >requirements -y >eJ >Even though the cars today are more sophisticated, faster, better qualityI >etc., you can still easily teach a 16 year old how to drive. That is notY# >much different than 10 years ago. q >yL >Similarily, the kids today can use faster systems to do certain things at a >certain level.v > ? >However, the mechanic fixing integration problems with onboard G >microprocessors in many common vehicles today is very much a differenta8 >person than the mechanic working on cars 10 years ago.  > L >GPS devices being able to locate cars and shutdown engines in case of theftH >. that is a totally different vehicle support requirement than tuning a >carburetor. >lL >Similarily, keeping an IT infrastructure up and running on a 24x7 basis andE >fixing things before they impact the business and complex middlewareuL >programming designed to be highly scalable, reliable, take advantage of theL >local platforms clustering features etc requires a whole different level ofJ >support than a simple college programmer (VB?) or admin person who has no >real experience.a >n  @ I think the car analogy is a poor one.  I believe, although I'm ? certainly no expert in this area, that maintaining today's carsoA requires far fewer skill than those many years ago.  There aren'th@ any carburetors anymore and things like GPS systems are all partA of the electronic system that are completely diagnosed by hookingu9 the car up to a diagnostic computer and reading the code.   A Actually though, there does bring up a point.  Some vendors wouldT@ have you believe that running their systems is no more difficult@ than reading the output of the diagnostic screens and performingA the action from the database recommendations, suggesting that the A auto mechanic analogy is true (but flipped on it's head from what ? you suggested).  We know that the reality of system maintenanceb- today is somewhat more complicated than this.,   >:-) >o >Kerry Main' >Senior Consultant >Compaq Canada Inc.  >Professional Services >Voice: 613-592-4660 >Fax  :  819-772-7036n >Email: Kerry.Main@Compaq.comt >c >t > [big snip]   -Jordan Hendersonu jordan@greenapple.come   ------------------------------    Date: 10 Jun 2001 23:09:17 +0800, From: Paul Repacholi <prep@prep.synonet.com> Subject: Re: The future of VMS- Message-ID: <87elsss76q.fsf@prep.synonet.com>   ' paddy.o'brien@zzz.tg.nsw.gov.au writes:   F > We employ younger staff for the whizz-bang GUI stuff written in C et@ > al., that our engineers have really little use for -- but it's+ > conventional to have GUI stuff available.c  ; > Not only do you employ the program language that fits then> > requirement, but you can employ the programmers to write the5 > programs in the language that fits the environment.e  ? The answer used to be that these people would never even get an  interview, let alone a job..   Serious question;t  ? Does anyoe here consider a 'one language' programmer has even a< moderate idea of programming?   5 Would you be happy to have them bugfixing non-C code?    -- e< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.e@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov h   ------------------------------    Date: 10 Jun 2001 23:24:50 +0800, From: Paul Repacholi <prep@prep.synonet.com> Subject: Re: The future of VMS- Message-ID: <87ae3gs6gt.fsf@prep.synonet.com>e  3 bill@triangle.cs.uofs.edu (Bill Gunshannon) writes:a  A > OK, as the person who started this particular thread, I'm back.d   > My list starts out with: >   A Bourne Shell clone a& >   A C-shell clone (most likely tcsh)  F Are you intending these to be CLIs? They will have to reverse engineerD the CLI hooks, and learn to link it for P1 space. And will DCL be in there as well?  C Plus, there is at least one big clash. Will they do the globing? oroC can standard VMS images use the rtl routines still to handle files?nB I'm sure there are a whole heap of other things that are of a like nature.t  F On thinking about it, I have come to the conclusion that they are veryB good topics for a course. For a change, they will have to do solidC thinking and get to understand the differences between two systems.e< And they may learn that some things just should not be done.   >   make  D That one is reasonable. Plus there are at least 3 equivalents extant that they can compare to.i   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.a@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov a   ------------------------------    Date: 10 Jun 2001 23:36:24 +0800, From: Paul Repacholi <prep@prep.synonet.com> Subject: Re: The future of VMS- Message-ID: <8766e4s5xj.fsf@prep.synonet.com>   + Christof Brass <brass@infopuls.com> writes:t  C > ... I don't see any big difference in most apps like browser, webe@ > server, spread sheet, word processor, graphics package, searchD > engine.  From what I read there is a demand of that kind of SW forD > VMS.  And I don't see any advantage in different PLs for that type > of programming.   @ > Would you like to provide a few examples why certain PLs excelA > on certain tasks within the range of app development (includingp > servers)?m  F Well, take the spread sheet and the graphics package. I think you willF agree that they not only will use a fair chunk of math, but that theirE answers and behaviour must also be 'least suprise' compared to normalx maths operations.t  B Now consider that Cs operations on unsigned and signed numbers areF most like very different math numeric types, then you have to say that2 as a result, C is less well suited for these apps.  E For the search engine, either a DB or a language like Snobol, or SCAN<A or SQL with plain files would seem to be a better idea. If for noe2 other reason than being able to compile the query.  D Note, that this totally ignores the issue of existing code that does large parts of the job.w   -- n< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.y@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov s   ------------------------------  % Date: Sun, 10 Jun 2001 13:38:41 -0400:- From: Michael Austin <miaustin@bellsouth.net>F6 Subject: Re: Urgent: Oracle 8 parallel server on VMS 7- Message-ID: <3B23B0A1.D5478D8B@bellsouth.net>   d Personally, if I were going to run an Oracle database on OpenVMS it would not be Oracle Classic, butg Oracle Rdb.   Rdb has been running on muliptle nodes in a true cluster very successfully for many yearsh` now.   (For those in the comp.database.oracle.server newsgroup, Digital and OpenVMS invented theb cluster, not as some of you have been told that it was Sun or NT and has been around since '81-'82f timeframe) While OPS on VMS has also been around a while, Rdb seems to have done it right  and it just works.   Michael Austin` DBA Consultant (Oracle Rdb and Oracle "Classic" on OpenVMS and most popular flavours of Eunichs)  
 chan01 wrote:o   > Hi, K >  I think it is not the problem with OPS Oracle, it is depending up on the D > database architecture. slow because of the LMD. may be you have to: > change your archi. what kind of system you were running. >a > Thankse > "Harald Thienel" <harald@h-thienel.de> wrote in message news:<9fr6la$6jh$03$1@news.t-online.com>....P > > Since one week we are running Oracle 8.1.7 on VMS-Alpha-Cluster (3 machines)2 > > in parallel server mode. VMS Version is 7.2.1.P > > there is one instance of the database on every machine. The database is used3 > > by 200-300 users distributed on the 3 machines.e > >,O > > First problem is that Oracle 8 seems to be several times slower than Oracle: > > 7 (maybe 4 times). > >gI > > But the big problem is that after some hours (1 to 3), Oracle becomesnM > > getting slower and slower and then stops processing completely. All usersoJ > > are waiting forever. This seems to be a problem in the lock managementI > > (LMD-process). Sometimes this state can be cured by shutting down oneeO > > instance of the database (then the other two resume working), sometimes alleG > > three instances must be shut down and restarted. In a manufacturingnC > > environment like ours, these shutdowns are not very wellcome...t > >fN > > We are in contact with Oracle-Support and tweeking around with parameters,' > > but this didn't really help so far.t > > ; > > The situation is escalating - any help is very welcome! 
 > > Thanks
 > > Harald   ------------------------------  % Date: Sun, 10 Jun 2001 17:26:53 +0100e, From: "Richard Maher" <maher_rj@hotmail.c*m>E Subject: Re: Use of SYSMAN to keep password file in sync on 8 systems 2 Message-ID: <9g07j2$q1$2@neptunium.btinternet.com>   Hi,e  B What would happen if you did a $set file/ru_journal on sysuaf.dat?  J I'm by no means recommending it, just curious. If there were no dodgy sideJ effects then It would be pretty easy to knock up a program that would readK the file directly and update the new password on all eight systems with theaF prophylactic of a DECdtm 2PC. And then of course there's *parallelism*) Update all eight in the same time as two.b  K One thing is for sure, if you update the file directly through RMS then you4D have the luxury of using manual record locking to freeze your targetL username between the read and write calls. (Something which is sadly lacking between $getuai and $setuai)   Regards Richard Maherq  9 PS> What macro holds the latest record layout for SYSUAF?e  = Hoff Hoffman <hoffman@xdelta.zko.dec.nospam> wrote in messagei- news:8QNT6.1156$fi2.29820@news.cpqcorp.net...hE > In article <4aktht43ijudqghfnjj7k8cv0tk3ju599a@4ax.com>, Norman Wooy <nwoo@videotron.ca> writes:6 >aE > :We have 8 NON-CLUSTERED systems (4 x Compaq Alphas running OpenVMS F > :7.1-2 and 4 x Digital VAX running VMS 5.5-2) distributed throughoutG > :the country.  Users can basically log onto any of these 8 systems to  > :do their work...m >mG >   I would tend to encourage an OpenVMS upgrade, particularly to thosepI >   OpenVMS VAX V5.5-2 systems.  (That software release shipped in 1992.)i >aF > :Their accounts are captive in the sense that they cannot get at the > :system command prompt.p >tA >   That would certainly meet the definition of a captive user...a >TI > :system, they are into the application.  The application is the same onhI > :all 8 systems.  One of the issues we need to resolve is when the users G > :change their passwords when logging onto anyone of the 8 systems, werC > :would like to have this password updated on the other 7 systems.  >pF >   Since these are captive users, I might suggest moving the password# >   mechanism into the application.a >oH > :We looked at the VAX Clustering solution but found it was too costly. >tF >   Inapplicable, too, given the distances involved between the nodes. >TI > :We also looked at Compaq DCE but DCE only works on OpenVMS 7.0 and up.a > I >   PATHWORKS-based (Advanced Server) domain (distributed) authentication G >   is available on OpenVMS V7.1 and later, and Kerberos authenticationeI >   is available on V7.3 and later.  (Kerberos would likely be a nice fitn) >   into your application environment...)r >iH > :We also looked at third party solution (from Ideous and Symantec) but) > :were also costly (We have 1000+ users)  >gE >   Lock in the passwords and don't worry about it, or use a passwordoF >   specific to the application and maintained within the application. >mH > :Doing some digging into the manuals, we saw a potential: SYSMAN.  Any: > :comments as to this solution? Or any other suggestions? >1H >   SYSMAN simply issues commands on other nodes, and is not specific toH >   this situation.  You could just as easily use DECnet task-to-task or< >   other similar approach to issue commands on other nodes. >oE >   There have been regular discussions of keeping a set of passwordsS >   synchronized.o >gJ >   Most any hack solution -- based on SYSMAN or task-to-task or otherwiseH >   -- could be gotten to work here.  The downside involves what happensJ >   when there is a lack of connectivity.  If you are willing to implementF >   the retransmit logic, or are looking to use a reliable transactionA >   mechanism such as RTR, or are otherwise willing to accept the?C >   potential for skewed passwords, well, that makes this easier...  >aI >   Another downside: any mechanism you implement to distribute passwordsrJ >   will be an obvious target for someone looking to crack your system(s). >eK >   There are a dozen or more discussions of external authentication in the(I >   OpenVMS Ask The Wizard area -- search for "authen" or other string...  >,( >  ---------------------------- #include' <rtfaq.h> -----------------------------nL >       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com, >  --------------------------- pure personal# opinion ---------------------------o1 >    Hoff (Stephen) Hoffman   OpenVMS Engineeringa hoffman#xdelta.zko.dec.com >p   ------------------------------    Date: 11 Jun 2001 01:27:04 +0800, From: Paul Repacholi <prep@prep.synonet.com>. Subject: VDE on FreeWare 4. Manuals unreadable- Message-ID: <873d98qm8n.fsf@prep.synonet.com>e  = Has anyone been able to read any of the docos for VDE on FW4? @ The copys on the web site seem to have been eaten by an internal9 FTP or something. The PS files have lost all the <lf>s...o   -- i< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.o@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov     ------------------------------    Date: 11 Jun 2001 00:36:32 +0800, From: Paul Repacholi <prep@prep.synonet.com>8 Subject: Re: VMS Installation and set up (from a newbie)- Message-ID: <87g0d8qokv.fsf@prep.synonet.com>m  ' kaliushkin@att.net (Dan Kalish) writes:y  
 > Hi, all!  ? > I've had a lot of experience with IBM mainframes and IBM PS/2 H > microcomputers, as a programmer, and concerning the latter, hardware, . > installation of operating systems, LAN, etc.  C > I could get a PDP-11 but would have to build a neighborhood power E > generator and probably reinforce the floor.  :-)  I could pick up ae   Leave the 11 till a bit later.  , > VAXstation on eBay: currently they've got . >  Digital VAXstation 4000/60 Unix Workstation  C Pretty ok. Fast enough, and you are not tight on space for the SCSIy@ drive as you are with a VLC. A 90 would be nicer. They won't run@ Ultrix BTW, but net/bsd is fine on them if that rocks your boat.   >  DEC 3100-M38 SYSTEMG >  DEC VAX STATION 3100 with Floppy.  (What are the differences betweenaH > these three Vaxen?).  I could run the Ersatz-11 emulator on one of the > PS/2s, under PC-DOS.  & They are very slow. Pass is my advise.  D > Once I get such a computer, where do I find VMS?  In fact how do IE > set up a VAXstation?  How do I add disk drives, tape backup, floppy,A > drives, and CD-ROM drives?  How about ethernet cards?  Does VMSc" > support peer-to-peer networking?  C Get Hoff's FAQ, and read it. Join DECUS... You can buy the hobbiestn CD, and away you go.  < Oh, VMS tends to be a bit fussy about CDs, so that may be an6 agrivation, and you will need a sync-on-green monitor.  @ Try really hard to forget everything you know about IBM, etc. It will only mislead you.   -- u< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov y   ------------------------------    Date: 10 Jun 2001 23:50:58 +0800, From: Paul Repacholi <prep@prep.synonet.com>2 Subject: Re: [OT] Current Microsoft v DOJ status ?- Message-ID: <871yoss599.fsf@prep.synonet.com>   3 nothome@spammers.are.scum (Malcolm Dunnett) writes:w  C >    I haven't seen any figures on the other side of the equation -c? > how much does an airbag enhance the chances of survival for as? > passenger who is already using a properly adjusted seat belt?n  E Summarie figures I saw (so beware...) showed no difference for a 'oneqB hit' accident, and the air bags as much worse in a multi-impact orE side impacts. They added the note that the numbers indicated that therF severity of the accident was less in a significant number of accidentsF for the seat-belt case, and attributed this to the driver being better able to maintain control.r  < BTW, neither get anywhere near the saftey of a full harness.   -- r< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.t@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov o   ------------------------------    Date: 10 Jun 2001 23:54:43 +0800, From: Paul Repacholi <prep@prep.synonet.com>2 Subject: Re: [OT] Current Microsoft v DOJ status ?- Message-ID: <87wv6kqqik.fsf@prep.synonet.com>h  % system@niuhep.physics.niu.edu writes:a  F > If they were still available and if it weren't for the environmentalF > impact I would kill for an old-style U.S. taxicab.  Huge space, easyE > to get in, the engines last 100s of thousands of miles in miserablew > city driving.   ! Aren't Checkers is in production?w   -- e< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.V@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.H Spam-To: uce@ftc.gov,enforcement@sec.gov,sness@fcc.gov,hfurchtg@fcc.gov,#   mpowell@fcc.gov,gtristan@fcc.gov R   ------------------------------   End of INFO-VAX 2001.320 ************************