1 INFO-VAX	Mon, 25 Aug 2003	Volume 2003 : Issue 470       Contents: Re: 306GB drives!  Re: 306GB drives! % Re: allocation classes and satellites % Re: allocation classes and satellites % Re: allocation classes and satellites % Re: allocation classes and satellites % Re: allocation classes and satellites % Re: allocation classes and satellites % Re: allocation classes and satellites % Re: allocation classes and satellites % Re: allocation classes and satellites I Re: Another Endangered Account ( was Re: London Transport moving off VMS)  Re: AntiVirus for OpenVMS / Re: connect FALCO 5220e to ALPHAstation 255/233  Re: DCL Raise error  Re: DCL Raise error   Re: decw$mwm change process name  Re: decw$mwm change process name> Re: Difficulties with DPW 500au and StorageWorks components...> Re: Difficulties with DPW 500au and StorageWorks components...+ EncompassUS associate membership renewal???  Re: ftp access to itrc Re: ftp access to itrc Re: ftp access to itrc Ingres Error Re: Last week at HP World  Re: Last week at HP World < Re: Look for information on exercising third-party FC disks.( Re: M7940/DHV-11, TU58, and a VAX 11/750. Re: Mount DQB0 gets 'allready allocated' error$ Re: mounting remote disks on VAX/VMS) Re: OT: 64 bit desktop computing is here. ) Re: OT: 64 bit desktop computing is here. ) Re: OT: 64 bit desktop computing is here. I Re: overhead with host-based volume shadowing for a one-member shadow-set   Patching Ident field of an Image$ Re: Patching Ident field of an Image$ Re: Patching Ident field of an Image$ Re: Patching Ident field of an Image2 Re: Question about a free shell provider - VisTech+ Re: Reconstructing pdf file broken via mail + Re: Reconstructing pdf file broken via mail 2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server2 Re: setting up a LAT service via a terminal server< Re: Shadow or Raid.  - Don't do both!  WAS Re: 306GB drives!" Shadow_max Copy Was: 306GB drives! Re: SYSn Re: SYSn Re: SYSn Re: SYSn Re: SYSn Re: SYSn! Re: Tcl/Tk 8.3 or 8.4 for OpenVMS ( Re: Unicode client as Open VMS terminal.1 Re: Will OpenVMS I64 run on a Dell PowerEdge 3250 1 Re: Will OpenVMS I64 run on a Dell PowerEdge 3250 1 Re: Will OpenVMS I64 run on a Dell PowerEdge 3250   Re: XML Parser for VAX (OpenVMS)  F ----------------------------------------------------------------------    Date: 25 Aug 2003 07:46:07 -07001 From: keithparris_NOSPAM@yahoo.com (Keith Parris)  Subject: Re: 306GB drives!= Message-ID: <cf15391e.0308250646.44673746@posting.google.com>   h Kilgallen@SpamCop.net (Larry Kilgallen) wrote in message news:<hccQeRRXSEl5@eisner.encompasserve.org>...s > In article <cf15391e.0308221032.7fa596d6@posting.google.com>, keithparris_NOSPAM@yahoo.com (Keith Parris) writes: I > > There are disaster-tolerant VMS clusters in place which actually have I > > 3 datacenters (plus a quorum node in a 4th site), where each of the 3 D > > sites has a valid copy of the data, and the cluster can continueG > > operating despite loss of any 2 of those 3 datacenters without data % > > loss or downtime, thanks to HBVS.  > ( > So does the Quorum node have 2 votes ?  C Yes, that would indeed be an appropriate number of votes if each of  the 3 main sites had 1 vote.  F In the specific case I was thinking of, each of the 3 main sites has 3D servers with 1 vote each, so they set the quorum node's votes to 4. F The minimum value to keep quorum alive down to 1 site (plus the quorumD node) seems to be to give one more vote to the quorum node than thatF of one of the 3 sites, so that the sole surviving site plus the quorum4 node outweigh the two lost sites in number of votes.  C If you wanted to be able to continue with one site and still allow, C for example, one node at the suriving site to go down due to an ECO @ patch installation, VMS upgrade, or a crash, you could raise the< quorum node's votes, but at the expense of relatively higher dependence on the quorum node.  F Note that even with the minimum number of votes noted above, after youD lose 2 sites, the quorum node becomes a single point of failure (butF of course it has done its job at that point, and you have much more to0 worry about right then than just quorum issues).   ------------------------------    Date: 25 Aug 2003 09:58:11 -07001 From: keithparris_NOSPAM@yahoo.com (Keith Parris)  Subject: Re: 306GB drives!= Message-ID: <cf15391e.0308250858.6f5a171a@posting.google.com>    "David McKenzie" <david.mckenzie@paradigm-shift.biz> wrote in message news:<3f472abd$0$95049$c30e37c6@lon-reader.news.telstra.net>... N > I do wonder however about three site clusters. At some stage in the abstract: > I wonder at what stage complexity, replaces reliability.  ? That's a good point.  It's very true that adding complexity can C introduce factors that work against availability.  For example, the C addition of host-based Volume Shadowing software to cover potential C controller [pair] failures introduces some level of additional risk F due to things like bugs in that software itself, potential performanceD impact during recovery operations, as well as introducing additionalE system management work and risk of human error.  In practice, though, D the additional contribution to availability that HBVS provides seems3 to outweigh the additional risk for most customers.   B Where I see folks adding a quorum node at a 3rd site is when theirE Recovery Time Objectives are too stringent to allow for the delay due D to manual recovery of quorum in the event of a loss of either one ofE their two main sites or the inter-site link(s) between them.  (I also C see this introduced as a convenience when inter-site link costs are A minimal, such as when all 3 sites are in the same office campus.)   B Where I see folks using 3-member shadowsets in 3 separate sites is? cases where income levels dependent on the cluster are high and D prevention of any data loss is a very high priority.  In a 2-site DTE cluster, in practice, there are still some times when you are exposed > to the potential of data loss, even if for short periods.  ForF example, in the scenario we call Rolling Disaster, while you are doingE a full shadow copy from one site to another, you lose the source site E before the copy can be completed.  Or, in the Creeping Doom scenario, F where a failure taking out the inter-site link occurs in one site, andE then that site is chosen to continue operations, but then the initial @ failure (maybe a fire starting in the network closet) spreads to= consume the entire datacenter, causing you to lose the recent > transactions.  Having 3 copies of the data in 3 separate sites. prevents data loss in both of these scenarios.   ------------------------------  % Date: Mon, 25 Aug 2003 17:12:50 +1000 : From: "David McKenzie" <david.mckenzie@paradigm-shift.biz>. Subject: Re: allocation classes and satellitesC Message-ID: <3f49b729$0$95044$c30e37c6@lon-reader.news.telstra.net>   K As it used to be, if you had multiple nodes erving a disk from an HSC, then C all of the HSC's and hosts that were connected had to have the same K allocation class. The way problems turn up in this configuration is that if J a node not connected directly to the disk cannot see an MSCP host with the@ same allocation class as the HSC then it will not see the disks.  B This is critical, if you are relying on MSCP serving between sites   David McKenzie  
 Law and IT  @ David.mckenzie@paradigm-shift.biz.anomoly     remove the anomoly    < "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message! news:3F49803C.8F3E5556@fsi.net... 1 > Phillip Helbig---remove CLOTHES to reply wrote:  > > L > > In my hobbyist cluster, I use allocation classes, but only because I useL > > volume shadowing.  I don't have any dual-ported disks.  No node has more > > than one SCSI bus. > > G > > To ensure unique names, each node has a different allocation class.  > >  > > The documentation says:  > > E > >     1.  When serving satellites, the same nonzero node allocation A > >         class value must be assigned to the serving computers  > >         and controllers. > > L > > What does this mean?  Does it mean that if a satellite boots from a nodeD > > with allocation class X, that the satellite has to have the same@ > > allocation class X?  If so, why?  If not, what does it mean? > H > I take it as meaning, for example, if you have NODEA with ALLOCLASS 1,F > and it has CI paths to HSJ01/2 and HSJ03/4, then HSJ01/2 and HSJ03/4B > must use an allocation class of 1 for the disk units they serve. > J > My take on it. I can't comprhend why this would be necessary, but that's > how I read it... >  > --   > David J. Dachtera  > dba DJE Systems  > http://www.djesys.com/ > * > Unofficial Affordable OpenVMS Home Page:! > http://www.djesys.com/vms/soho/    ------------------------------    Date: 25 Aug 2003 00:30:02 -0700% From: Bart.Zorn@xs4all.nl (Bart Zorn) . Subject: Re: allocation classes and satellites< Message-ID: <a98cd882.0308242330.3787658@posting.google.com>  C The essential thing to keep in mind here is that you must configure @ all disks (and tapes) in your cluster in such a way that no nameA conflict can arise. No two different devices should have the same E name, or no single device should be visible under more than one name.   D Sometimes it is sufficient to just give all systems in the cluster aE different allocation class, but if you have a shared SCSI bus between C systems, that would not help. In that case you need port allocation  classes, a.k.a. device naming.  ; If you use ALLOCLASS=0, local SCSI device will get the name F node$DKdduuu. So for satelites in the cluster an allocation class of 0C is usually the best solution. You get device names which are easily 
 recognisable.   A If you have DSSI or CI devices, they get their alloclass from the C controller, not the host system. The alloclass of the host makes no  difference here.  F If you have an Alpha server with multiple SCSI busses, possibly sharedB with other Alpha's, I recommend using port allocation classes. SeeN also http://h71000.www7.hp.com/doc/731FINAL/4477/4477pro_009.html#index_x_487.A The node allocation class should again be zero. This also has the E advantage that the CDROM drive or the SCSI disk in the master PCI box  of a Wildfire system get the easiliy recognisable names.   C In short: once you use port allocation classes, you don't need node + allocation classes to be anything but zero.    HTH,  	 Bart Zorn       h Kilgallen@SpamCop.net (Larry Kilgallen) wrote in message news:<U7qyUqDbH$ZE@eisner.encompasserve.org>...y > In article <bib4q1$t29$1@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes: M > > In my hobbyist cluster, I use allocation classes, but only because I use  M > > volume shadowing.  I don't have any dual-ported disks.  No node has more   > > than one SCSI bus. > > G > > To ensure unique names, each node has a different allocation class.  > >  > > The documentation says:  > > E > >     1.  When serving satellites, the same nonzero node allocation A > >         class value must be assigned to the serving computers  > >         and controllers. > > M > > What does this mean?  Does it mean that if a satellite boots from a node  E > > with allocation class X, that the satellite has to have the same  @ > > allocation class X?  If so, why?  If not, what does it mean? > F > I presume it means that if multiple boot nodes are able to serve theG > system disk to a particular satellite, those boot nodes must have the & > same allocation class as each other.   ------------------------------  % Date: Mon, 25 Aug 2003 13:27:23 +0200 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl> . Subject: Re: allocation classes and satellites9 Message-ID: <bicrr0$7ra5c$1@ID-143435.news.uni-berlin.de>   L "Phillip Helbig---remove CLOTHES to reply" <helbig@astro.multiCLOTHESvax.de>1 schreef in bericht news:bib4q1$t29$1@online.de... J > In my hobbyist cluster, I use allocation classes, but only because I useJ > volume shadowing.  I don't have any dual-ported disks.  No node has more > than one SCSI bus. > E > To ensure unique names, each node has a different allocation class.  >  > The documentation says:  > C >     1.  When serving satellites, the same nonzero node allocation ? >         class value must be assigned to the serving computers  >         and controllers. > J > What does this mean?  Does it mean that if a satellite boots from a nodeB > with allocation class X, that the satellite has to have the same> > allocation class X?  If so, why?  If not, what does it mean?  J The text refers to dual hosted disks, say in a CI cluster with two HSC50'sJ where each RAnn disk has both paths 9A and B) connected to an HSC. In that< case these two HSC's must have the same value for ALLOCLASS.; The same for two VAX systems connected to the same disk(s).   G The "modern" equivalents are shared SCSI and shared DSSI clusters. Both B AXP's or VAXes on the same shared bus must have the same value forI ALLOCLASS. The NI connected nodes in the cluster must see the same device $ name, served by all cluster members.   Hans   ------------------------------  + Date: Mon, 25 Aug 2003 15:41:25 +0200 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> . Subject: Re: allocation classes and satellites; Message-ID: <01KZVTVZIZK2AOOLBD@sysdev.deutsche-boerse.com>   N > > Sounds like it may be related to an older restriction where a MSCP servingJ > > node must have the same allocation class as the disks it serves.  WhatK > > version are you running?  I fixed this a while ago (in V7.1+patches and  > > in all later versions).  >  > It's still in the 7.2-1 HELP.   I Two corrections: First: documentation, not HELP.  Second, it is still in  @ the 7.1 documentation.  (I think this was the last CD with full ( BOOKREADER docs, so I tend to use that.)   ------------------------------  + Date: Mon, 25 Aug 2003 15:43:03 +0200 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> . Subject: Re: allocation classes and satellites; Message-ID: <01KZVTY9G4EGAOOLBD@sysdev.deutsche-boerse.com>   M > As it used to be, if you had multiple nodes erving a disk from an HSC, then E > all of the HSC's and hosts that were connected had to have the same M > allocation class. The way problems turn up in this configuration is that if L > a node not connected directly to the disk cannot see an MSCP host with theB > same allocation class as the HSC then it will not see the disks. > D > This is critical, if you are relying on MSCP serving between sites  D A related question: My hobbyist cluster has 1 DSSI disk, inside of aC VAX.  At the moment, it has the same allocation class as the VAX.   = Everything seems to work, but in general is this good or bad?    ------------------------------  + Date: Mon, 25 Aug 2003 15:46:25 +0200 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> . Subject: Re: allocation classes and satellites; Message-ID: <01KZVU0ZCNKUAOOLBD@sysdev.deutsche-boerse.com>   E > The essential thing to keep in mind here is that you must configure B > all disks (and tapes) in your cluster in such a way that no nameC > conflict can arise. No two different devices should have the same G > name, or no single device should be visible under more than one name.    Right.  F > Sometimes it is sufficient to just give all systems in the cluster a > different allocation class,    Right, my solution.   + > but if you have a shared SCSI bus between E > systems, that would not help. In that case you need port allocation   > classes, a.k.a. device naming.   I don't have any shared buses.  = > If you use ALLOCLASS=0, local SCSI device will get the name H > node$DKdduuu. So for satelites in the cluster an allocation class of 0E > is usually the best solution. You get device names which are easily  > recognisable.   G I have to use non-zero allocation classes if at least one disk on that  E node is shadowed.  Also, for consistency, I'd prefer to have non-zero E allocation classes on all systems.  (I also might want to mount some  % disks on a satellite on other nodes.)   C > If you have DSSI or CI devices, they get their alloclass from the E > controller, not the host system. The alloclass of the host makes no  > difference here.  F I do have one DSSI disk, inside a VAX.  At the moment, its allocation I class is the same as that of that VAX (which also has a SCSI bus).  Good  ' or bad?  (Everything seems to work OK.)    ------------------------------  % Date: Mon, 25 Aug 2003 18:40:08 +0200 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl> . Subject: Re: allocation classes and satellites9 Message-ID: <bide5c$84m4q$1@ID-143435.news.uni-berlin.de>   H "Phillip Helbig" <HELBPHI@sysdev.deutsche-boerse.com> schreef in bericht5 news:01KZVTY9G4EGAOOLBD@sysdev.deutsche-boerse.com... J > > As it used to be, if you had multiple nodes erving a disk from an HSC, thenG > > all of the HSC's and hosts that were connected had to have the same L > > allocation class. The way problems turn up in this configuration is that ifJ > > a node not connected directly to the disk cannot see an MSCP host with the D > > same allocation class as the HSC then it will not see the disks. > > F > > This is critical, if you are relying on MSCP serving between sites > F > A related question: My hobbyist cluster has 1 DSSI disk, inside of aC > VAX.  At the moment, it has the same allocation class as the VAX. ? > Everything seems to work, but in general is this good or bad?   J It does not matter as long as you have one VAX (or AXP) on that bus. Other< cluster members should have a different value for ALLOCLASS.   ------------------------------    Date: 25 Aug 2003 12:21:34 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) . Subject: Re: allocation classes and satellites3 Message-ID: <eVouxWe94cDr@eisner.encompasserve.org>   w In article <01KZVTY9G4EGAOOLBD@sysdev.deutsche-boerse.com>, Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> writes:  > F > A related question: My hobbyist cluster has 1 DSSI disk, inside of aE > VAX.  At the moment, it has the same allocation class as the VAX.   ? > Everything seems to work, but in general is this good or bad?   D   In general this is OK.  It means the VAX can serve the disk to theJ   other members of the cluster (because the two allocation classes match).H   Since ther DSSI bus is only connected to one system you could also useJ   allocation class 0 on both (which make it the allocation class ignored).  H   You only have to use non-0 allocation classes when using a dual pathedK   bus (such as a dual ended DSSI) where more than one system can serve the  L   disk to the cluster and that disk therefor must have the same device name    on both systems.  G   If the system's allocation class does not match the disk's allocation F   class then only that system can see the disk, it cannot be served to   the cluster.   ------------------------------  + Date: Mon, 25 Aug 2003 19:29:48 +0200 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> . Subject: Re: allocation classes and satellites; Message-ID: <01KZW1R85AX2AMSOR1@sysdev.deutsche-boerse.com>   H > > A related question: My hobbyist cluster has 1 DSSI disk, inside of aG > > VAX.  At the moment, it has the same allocation class as the VAX.   A > > Everything seems to work, but in general is this good or bad?  > J > In general this is OK.  It means the VAX can serve the disk to the otherE > members of the cluster (because the two allocation classes match).    C Presumably, then, if the allocation classes were different, then I  G COULDN'T serve the disk to other members of the cluster.  Since I want  H to do that, the allocation classes should match.  So I suppose I should ? keep them the same so that I can serve the disk to other nodes.    > Since B > ther DSSI bus is only connected to one system you could also useJ > allocation class 0 on both (which make it the allocation class ignored).  ? Yes, from that point of view, but since HBVS requires non-zero  H allocation classes (only to make the disk names short enough, as far as 9 I can tell!), I have to have non-zero allocation classes.   H > You only have to use non-0 allocation classes when using a dual pathedJ > bus (such as a dual ended DSSI) where more than one system can serve theF > disk to the cluster and that disk therefor must have the same device > name on both systems.   B Right, but also if one uses HBVS.  This really confused me at the F beginning, since ONE motivation for HBVS is the ability to access the I same "disk" from more than one node!  It appears the only reason this is  E required is to make the disk name short enough for the HBVS software!   G > If the system's allocation class does not match the disk's allocation J > class then only that system can see the disk, it cannot be served to the > cluster.     OK.  Makes sense in hindsight!   ------------------------------  # Date: Mon, 25 Aug 2003 13:54:46 GMT # From: "John Smith" <a@nonymous.com> R Subject: Re: Another Endangered Account ( was Re: London Transport moving off VMS)I Message-ID: <Gyo2b.276981$hOa.11825@news02.bloor.is.net.cable.rogers.com>   < "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message! news:3F4572AB.56113C3F@fsi.net...  > Tim Llewellyn wrote: > >  > > Fellow VMS users,  > > B > > I have received information that would suggest that if you are planning to B > > visit London then you might consider sooner rather than later. > > E > > London Transport are currently recruiting VMS ops (for peanuts, I 
 would makeC > > a net loss if I relocated for these positions, shift work too).  However, I did9 > > hear that they are already planning the move off VMS.  > >  > > You have been warned.  >  > Sue?     David,  B Sh*t like this wouldn't be happening at many soon-to-be former VMS? accounts if HP took marketing and advertising of VMS seriously.   D Marketing and advertising consists of getting your value propositionB in front of the complete decision-making chain - from the Board ofB Directors, the head of application development, the chief securityB officer, the CTO, the head of computer operations (ie. data centerF management), the application architects, and finally the grunts in the- field who actually do the coding and testing.   D HP does none of this for VMS with any conviction or elan, or at all.  B HP seems to think that advertising $59 inkjet cartridges will helpC companies make data center and disaster tolerence decisions go HP's  way.  " As Homer Simpson might say, "Doh!"   ------------------------------  % Date: Mon, 25 Aug 2003 10:57:36 -0400 0 From: "Brian Tillman" <Tillman@sparkingwire.com>" Subject: Re: AntiVirus for OpenVMS$ Message-ID: <3f4a23eb$1@news.si.com>  F >A client is being forced by its security chappies to (a) implement anE >AV solution on their VMS boxes, or (b) come up with some irrefutable " >reasons why this is not required.  L VMS Mail is not an active agent; i.e. it does not try to execute whatever isF contained in the messages it receives.  Also, most users on a properlyL configured VMS system have no privilege to modify important system files, soH viruses have no ability to attack.  The worst case is that a virus mightK alter a user's own files or affect a user's own process.  Recovery is easy: J delete the offending program and restore damaged files from backup.  ThereJ is no registry in VMS that specifies files to be automatically run at bootJ time.  That's done by a read-only (to the general user) file that is human4 readable and to any modification can be easily seen.  D >I know Sophos do some sort of AV solution, but I'm note sure of (a)5 >its scope of functionality, (b) how expensive it is.   K Sophos' VMS-based antivirus does not scan for VMS-hosted viruses.  It scans : for PC-hosted viruses being passed through the VMS system. --  I Brian Tillman         Internet: Brian.Tillman at smiths-aerospace dot com 5 Smiths Aerospace  Addresses modified to prevent SPAM. D 3290 Patterson Ave. SE, MS 1B3 Replace "at" with "@", "dot" with "." Grand Rapids, MI 49512-1991 8        This opinion doesn't represent that of my company   ------------------------------  % Date: Mon, 25 Aug 2003 13:18:50 +0200 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl> 8 Subject: Re: connect FALCO 5220e to ALPHAstation 255/2339 Message-ID: <bicrau$78s84$1@ID-143435.news.uni-berlin.de>   L "Phillip Helbig---remove CLOTHES to reply" <helbig@astro.multiCLOTHESvax.de>1 schreef in bericht news:bi8utb$2f1$1@online.de... G > Several years ago, I used to have a FALCO 5220e terminal connected to J > the serial port (OPA0 or TTA0) of an ALPHAstation 255/233.  Now, I can't > get it to work.  > H > The serial ports on the ALPHAstation work OK; a VT320 connects up withJ > no problem.  Thus, I suspect that the problem is with the FALCO terminal > and/or with the cables.  > H > The cables are PHYSICALLY OK, since other connections using them work. > J > When I had things working several years ago, I had a couple of very longC > serial cables connected together and used that as the connection. H > Presumably, the reason was that a standard serial cable wouldn't work,? > and the ones which did had some non-standard pin connections.  > F > Does the FALCO require a different sort of cable (apart from gender)E > than a VT320 (the FALCO has a female RS232 connector on the back (2 G > actually, one for each port) whereas the VT320 has a male connector)? I > There are so many options in the setup that even if it were possible to G > change the nature of the connection, I doubt I could find it.  Anyone  > know if this is possible?  > F > What should the setup values be for the FALCO to use it as a console! > terminal with the ALPHAstation?  > H > I have a serial cable where I can flip switches to turn the individualH > pins on and off at each end.  What should the settings be for a VT320? > For the FALCO? > 6 Try jumpering/shorting pins 4+5 and 6+20 on the Falco.   ------------------------------  + Date: Mon, 25 Aug 2003 16:24:14 +0200 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>  Subject: Re: DCL Raise error; Message-ID: <01KZVVEGT12WAOOLBD@sysdev.deutsche-boerse.com>   N > > Might seems a silly question, but is there an easy way to raise a specific > > error in DCL?  >  > $	spawn exit 20    My favourite?      $  SPAWN EXIT 36   ------------------------------  + Date: Mon, 25 Aug 2003 16:23:24 +0200 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>  Subject: Re: DCL Raise error; Message-ID: <01KZVVE8E3DUAOOLBD@sysdev.deutsche-boerse.com>   8 From:	IN%"Kilgallen@SpamCop.net" 22-AUG-2003 20:41:01.02 To:	IN%"Info-VAX@Mvb.Saic.Com" CC:	 Subj:	RE: DCL Raise error   , Return-path: <Info-VAX-Request@Mvb.Saic.Com>> Received: from CONVERSION-DAEMON by sysdev.deutsche-boerse.comO  (PMDF V5.2-32 #40435) id <01KZRXIMUE34ANRO56@sysdev.deutsche-boerse.com>; Fri, !  22 Aug 2003 20:40:59 +0200 (MET) : Received: from mailp2.deutsche-boerse.de ([172.20.124.50])4  by sysdev.deutsche-boerse.com (PMDF V5.2-32 #40433)D  with ESMTP id <01KZRXILYL7EA9XQHG@sysdev.deutsche-boerse.com>; Fri,!  22 Aug 2003 20:40:56 +0200 (MET) L Received: from deimos1.deutsche-boerse.de () by mailhub.exchange.de (V1.3.2)1  id h7MIetA23720; Fri, 22 Aug 2003 20:40:55 +0200 M Received: (from uucp@localhost) by mail2.exchange.de (V1.3) id UAA16939; Fri, %  22 Aug 2003 20:40:54 +0200 (MET DST) M Received: from unknown(198.151.12.104) by deimos1.deutsche-boerse.de via smap 5  (V5.0)	id xmab16840; Fri, 22 Aug 2003 20:40:22 +0200 % Date: Fri, 22 Aug 2003 13:10:42 -0500 - From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: DCL Raise error To: Info-VAX@Mvb.Saic.Com 1 Reply-to: Kilgallen@SpamCop.net (Larry Kilgallen) 3 Message-id: <DEZUcnUsbBXk@eisner.encompasserve.org>  Organization: LJK Software Content-type: TEXT/PLAIN8 NNTP-Posting-Date: Fri, 22 Aug 2003 18:09:16 +0000 (UTC)I X-Trace: grandcanyon.binc.net 1061575756 18546 192.135.80.34 (22 Aug 2003   18:09:16 GMT) X-Complaints-To: abuse@binc.net  X-Gateway-From: mvb.saic.com	 Lines: 13 H X-Authentication-warning: deimos1.deutsche-boerse.de: uucp set sender to)  <Info-VAX-Request@Mvb.Saic.Com> using -f = X-Comment: Message Virus scanned by sysdev.deutsche-boerse.de  X-Gateway-source-info: USENET  X-Newsgroups: comp.os.vms   p In article <93b50805.0308220958.1536c12c@posting.google.com>, chris@townleyc.demon.co.uk (Chris Townley) writes:L > Might seems a silly question, but is there an easy way to raise a specific > error in DCL?    $	spawn exit 20 + %DCL-S-SPAWNED, process KILGALLEN_1 spawned = %DCL-S-ATTACHED, terminal now attached to process KILGALLEN_1 ' %SYSTEM-F-BADPARAM, bad parameter value 3 %DCL-S-RETURNED, control returned to process _RTA3: ' %SYSTEM-F-BADPARAM, bad parameter value  $	show symbol $status    $STATUS == "%X00000014"  $    ------------------------------  % Date: Mon, 25 Aug 2003 10:25:42 -0400 0 From: "Brian Tillman" <Tillman@sparkingwire.com>) Subject: Re: decw$mwm change process name $ Message-ID: <3f4a1c70$1@news.si.com>  ! >I found that DECW$MWM was by far L >more robust than the window manager(s) that come with Exceed, at least when >dealing with Decwindows apps.  G Why would you need to use ANY window manager with Exceed, then Windows, ' itself, can act as the windows manager?  --  I Brian Tillman         Internet: Brian.Tillman at smiths-aerospace dot com 5 Smiths Aerospace  Addresses modified to prevent SPAM. D 3290 Patterson Ave. SE, MS 1B3 Replace "at" with "@", "dot" with "." Grand Rapids, MI 49512-1991 8        This opinion doesn't represent that of my company   ------------------------------  + Date: Mon, 25 Aug 2003 14:45:53 +0000 (UTC) , From: lewis@mazda.mitre.org (Keith A. Lewis)) Subject: Re: decw$mwm change process name . Message-ID: <bid7f1$aob$1@newslocal.mitre.org>  | "Brian Tillman" <Tillman@sparkingwire.com> writes in article <3f4a1c70$1@news.si.com> dated Mon, 25 Aug 2003 10:25:42 -0400:" >>I found that DECW$MWM was by farM >>more robust than the window manager(s) that come with Exceed, at least when  >>dealing with Decwindows apps.  > H >Why would you need to use ANY window manager with Exceed, then Windows,( >itself, can act as the windows manager?  D Promised the customer a Motif look-and-feel.  The app was originally? developed using X terminals, and that's what they were used to.   I Aside from that, some non-Motif X functions simply don't work right using  Windows windows.  % --Thundermaker$yahoo.com (Spud Demon) > The above may not (yet) represent the opinions of my employer.   ------------------------------  % Date: Mon, 25 Aug 2003 07:45:05 +0200  From: Dirk Munk <munk@home.nl>G Subject: Re: Difficulties with DPW 500au and StorageWorks components... 2 Message-ID: <bic7r5$7pf$1@news2.tilbu1.nb.home.nl>   ohm62 wrote:K > Well, as a matter of fact, the SCSI DIFF inscription and the differential J > SCSI logo <<>- are stamped on a white sticker next to the connector, not > really in the metal....   6 Maybe it is stamped in the metal under the sticker ???   > D > The main Qlogic chip on board says ISP1040B and the following linkK > http://www.qlogic.com/products/ultra_series/ultra_one_tech.asp  says that N > processor is used in both QLA1040 (single ended) and QLA1041 (differential),L > then mentions 3 connectors, 1 external 2 internals for QLA1040, and only 1K > external connector for QLA1041... My board definitely has 3 connectors!!!    So has my adaptor.   > J > Now I am really confused!!!  Could that sticker for differential type beN > mistakenly applied there?  Am I faced with an alien replica?  How can I tellJ > these things apart for sure?  Does one of you have a QLA1041 or KZPBA-CB; > or -CY, who could tell me the number of connectors on it?  > L > The vendor proposes a QLA1080 (capable of both Ultra wide Single Ended andG > Ultra2 LVD...) as a replacement or my money back... I am unsure it is B > compatible with the PWS though, even with recent SRM versions.    P The bios of the QLA1020/1040 adaptors is part of the SRM console, and is loaded J by the SRM console. I'm quite sure it is a Digital / Compaq / HP specific M version that differs from the normal Qlogic version. If I'm not mistaken the  G change from W to UW for these adaptors was done by upgrading this bios.l  K Taking these things in account I seems to me that it is very likely that a lQ QLA1080 will not function properly in your PWS. These adaptors are nut supported a by VMS as far as I know.  Q Now when you search sys$system:sys$config.dat for NCR, Adaptec and SCSI you will eP find entries that suggest certain adapters of these brands are supported, but I ( can't garantuee that. (haven't tried it)  " As I see it you can do two things:   1. find a S.E. QLA1040, or! 2. find a diff. personality unit.3      	 There are K > some warnings in older posts in this news group about its incompatibilitycL > with OpenVMS, but maybe things have evolved...?   I don't know what is the" > right thing to do at this point! >  > Thank You! >  >   -- Olivier.p >  >    ------------------------------    Date: 24 Aug 2003 23:50:30 -0700% From: Bart.Zorn@xs4all.nl (Bart Zorn)sG Subject: Re: Difficulties with DPW 500au and StorageWorks components...f= Message-ID: <a98cd882.0308242250.7a05db2d@posting.google.com>H  E A difference between the SE and the diff versions of the KZPBA that IrC know of, is that the diff version (the -CY version) has an array ofuE sockets for termination resistors near the connector panel, where theb SE (-CX) has not.g   HTHe  	 Bart Zorne  b "ohm62" <ohm62@coldmail.com> wrote in message news:<1Vd2b.15997$kF1.15290@nwrdny02.gnilink.net>...K > Well, as a matter of fact, the SCSI DIFF inscription and the differential J > SCSI logo <<>- are stamped on a white sticker next to the connector, not > really in the metal....g > D > The main Qlogic chip on board says ISP1040B and the following linkK > http://www.qlogic.com/products/ultra_series/ultra_one_tech.asp  says thatiN > processor is used in both QLA1040 (single ended) and QLA1041 (differential),L > then mentions 3 connectors, 1 external 2 internals for QLA1040, and only 1K > external connector for QLA1041... My board definitely has 3 connectors!!!e > J > Now I am really confused!!!  Could that sticker for differential type beN > mistakenly applied there?  Am I faced with an alien replica?  How can I tellJ > these things apart for sure?  Does one of you have a QLA1041 or KZPBA-CB; > or -CY, who could tell me the number of connectors on it?Y > L > The vendor proposes a QLA1080 (capable of both Ultra wide Single Ended andG > Ultra2 LVD...) as a replacement or my money back... I am unsure it iseK > compatible with the PWS though, even with recent SRM versions.  There arewK > some warnings in older posts in this news group about its incompatibility-L > with OpenVMS, but maybe things have evolved...?   I don't know what is the" > right thing to do at this point! >  > Thank You! >  >   -- Olivier.i   ------------------------------  + Date: Mon, 25 Aug 2003 11:23:10 +0000 (UTC)@. From: John Forkosh <john@SeeSigForAddress.com>4 Subject: EncompassUS associate membership renewal???, Message-ID: <bicriu$i62$1@reader2.panix.com>  ( After renewing my vms hobbyist licenses,. I also went to Encompass to renew my associate3 membership.  But associate member renewal, per se,  3 doesn't seem to be a possible option.  Is associater0 renewal necessary?  Or should I just join again? Or what?  Thanks,s -- a> John Forkosh  ( mailto:  j@f.com  where j=john and f=forkosh )   ------------------------------    Date: 24 Aug 2003 23:56:57 -0700% From: Bart.Zorn@xs4all.nl (Bart Zorn)  Subject: Re: ftp access to itrco= Message-ID: <a98cd882.0308242256.6c45ca8f@posting.google.com>.  C Indeed, it works. But on my DS10 with Mozilla the directory listings looks hopelessly garbled.h  	 Bart Zorn     k brad@.gateway.2wire.net (Bradford J. Hamilton) wrote in message news:<%uc2b.248008$o%2.112142@sccrnsc02>...mP > An earlier post in this NG (I think) complained of lack of access to the patchQ > list via FTP at ftp.itrc.hp.com.  When I saw the post, I checked, and found theiN > same problem that the poster complained about; no directories were available > for viewing. > I > I just checked the site again a few minutes ago, and all seems to be asaL > expected; directories are available for viewing, and the patch tree can be > navigated as usual.v > P > I wanted to follow up to the original thread, but could not find it in Google,, > so I am starting a new thread.  Apologies. > R > ________________________________________________________________________________C > Bradford J. Hamilton                    "All opinions are my own"uM > bMradAhamiPltSon-at-coMmcAast.nPeSt     "Lose the MAPS, and replace '-at-' :2 >                                          with @"   ------------------------------  # Date: Mon, 25 Aug 2003 10:50:07 GMT54 From: brad@.gateway.2wire.net (Bradford J. Hamilton) Subject: Re: ftp access to itrce> Message-ID: <zRl2b.191338$It4.93408@rwcrnsc51.ops.asp.att.net>  e In article <a98cd882.0308242256.6c45ca8f@posting.google.com>, Bart.Zorn@xs4all.nl (Bart Zorn) writes:aD !Indeed, it works. But on my DS10 with Mozilla the directory listing !looks hopelessly garbled. !m
 !Bart Zorn !n  G I found that the post I was attempting to repond to was actually in then openvms.org forum.  Oops.w  L Bart, if you don't mind my asking, why access an FTP site with Mozilla, when= you (presumably) have "native" FTP capability from your DS10?e  J __________________________________________________________________________A Bradford J. Hamilton                    "All opinions are my own"rK bMradAhamiPltSon-at-coMmcAast.nPeSt     "Lose the MAPS, and replace '-at-' H0                                          with @"   ------------------------------    Date: 25 Aug 2003 05:59:08 -0700% From: Bart.Zorn@xs4all.nl (Bart Zorn)b Subject: Re: ftp access to itrco= Message-ID: <a98cd882.0308250459.147a8ec1@posting.google.com>   E Well, there is a simple explanation for that: via Mozilla I can reach)- the site, but using command line ftp I can't.s  D I did not bother to find out why that is yet. Maybe I will sometime. Could be a firewall issue.   Bart  z brad@.gateway.2wire.net (Bradford J. Hamilton) wrote in message news:<zRl2b.191338$It4.93408@rwcrnsc51.ops.asp.att.net>...g > In article <a98cd882.0308242256.6c45ca8f@posting.google.com>, Bart.Zorn@xs4all.nl (Bart Zorn) writes:iF > !Indeed, it works. But on my DS10 with Mozilla the directory listing > !looks hopelessly garbled. > !o > !Bart Zorn > !i > I > I found that the post I was attempting to repond to was actually in they > openvms.org forum.  Oops.u > N > Bart, if you don't mind my asking, why access an FTP site with Mozilla, when? > you (presumably) have "native" FTP capability from your DS10?  > L > __________________________________________________________________________C > Bradford J. Hamilton                    "All opinions are my own" M > bMradAhamiPltSon-at-coMmcAast.nPeSt     "Lose the MAPS, and replace '-at-' s2 >                                          with @"   ------------------------------    Date: 25 Aug 2003 00:06:00 -0700& From: smohapatra@hss.hns.com (Sukanta) Subject: Ingres Errorv= Message-ID: <5c75209e.0308242306.2eeb56f4@posting.google.com>a   Hi,o  / I am getting following error in the '.out file'5  C SYSTEM ERROR: II_gettupdesc(2) : Requested 264 bytes from pipe, buth read only 237 bytes.  2 What is the possible way to find out the solution?  & Advance Thanks, for any input on this.     regdsi Sukanta    ------------------------------  % Date: Mon, 25 Aug 2003 10:29:06 -0400e0 From: "Brian Tillman" <Tillman@sparkingwire.com>" Subject: Re: Last week at HP World$ Message-ID: <3f4a1d3c$1@news.si.com>  ) Mark Schafer <mark.schafer@hp.com> wrote:S  8 >Congrats for receiving the Encompass recognition award!   No surprise here.g  G Mark, why did you include the ENTIRE content of Sue's post in your own?.% Doesn't cut and paste work on HP PCs?3 -- 3I Brian Tillman         Internet: Brian.Tillman at smiths-aerospace dot com 5 Smiths Aerospace  Addresses modified to prevent SPAM.sD 3290 Patterson Ave. SE, MS 1B3 Replace "at" with "@", "dot" with "." Grand Rapids, MI 49512-1991 8        This opinion doesn't represent that of my company   ------------------------------  % Date: Mon, 25 Aug 2003 16:50:58 +0200t$ From: Michael Unger <unger@decus.de>" Subject: Re: Last week at HP World9 Message-ID: <bid83u$7uubn$1@ID-152801.news.uni-berlin.de>   * On 25-Aug-2003 16:29, Brian Tillman wrote:  + > Mark Schafer <mark.schafer@hp.com> wrote:  > 9 >>Congrats for receiving the Encompass recognition award!  >  > No surprise here.t > I > Mark, why did you include the ENTIRE content of Sue's post in your own? ' > Doesn't cut and paste work on HP PCs? &                                 ^^^^^^  E According to his "user-agent" header he is running Mozilla V1.3 on an3 AlphaServer DS10 466 MHz ...   Michaelo   -- i; Real names enhance the probability of getting real answers.w@ Please do *not* send "Security Patch Notifications" or "SecurityA Updates"; this system isn't running a Micro$oft operating system..= And don't annoy me <mailto:postmaster@[127.0.0.1]> please ;-)w   ------------------------------    Date: 25 Aug 2003 01:02:42 -0500+ From: young_r@encompasserve.org (Rob Young)aE Subject: Re: Look for information on exercising third-party FC disks.s3 Message-ID: <PRhols6iuG1b@eisner.encompasserve.org>o  T In article <3F4852A1.9D58ABBF@telusplanet.net>, Lee <lytmah@telusplanet.net> writes: > Thanks, Rob.F > I realize comparing I/O of the DGA and DUA disks would be fruitless.N > I'm more interested in evaluating the compatibility of the DGA disks on VMS.  ; 	I am not sure what you mean by compatability.  Do you meang< 	do they work?  Sure they work.  Can you break them or throw> 	enough IO at them to break them?  I wouldn't know how or why.  N > I wanted to know if I could add more to the current exerciser mix, which forL > the first five minutes of the run has 18 concurrent jobs (12 CONVERT and 6I > sequential copies).  Even assuming that VMS CONVERT is efficient, woulds3 > these 18 jobs not throw off sequential pre-fetch?e  < 	I'm not sure.  How are you doing it?  Each convert going to; 	a different DGA?  You will prefetch.  Two or more convertsc: 	going to the DGA?  I'd wager the subsystem recognizes two6 	separate IO streams and prefetches.  But I'm not sure& 	what you are attempting to determine.  J > An interesting note is that the DGA devices have so far racked up errorsH > in bunches.  No errors have been logged on the corresponding HSJ disksC > I reformatted to have the same number of blocks as the DGA disks.I@ > What do you make of this?  (Only L is connected to FC so far.)  C 	I don't know.  I haven't seen any errors.  Maybe you've discoveredsC 	a way to make them spit errors.  What does ANALYZE/ERROR report?   + 	Current on all firmware, HBAs and Console?.   > 	 > Site 1:lF > $1$DGA142:         (L)  ShadowSetMember     20  (member of DSA1001:)F > $1$DGA191:         (L)  ShadowSetMember     20  (member of DSA1003:)F > $1$DGA200:         (L)  ShadowSetMember     20  (member of DSA1005:)F > $1$DGA444:         (L)  ShadowSetMember     19  (member of DSA1008:) > 	 > Site 2:tF > $1$DGA4270:        (L)  ShadowSetMember      3  (member of DSA1001:)F > $1$DGA4271:        (L)  ShadowSetMember      4  (member of DSA1003:)F > $1$DGA4272:        (L)  ShadowSetMember      5  (member of DSA1006:)F > $1$DGA4273:        (L)  ShadowSetMember      6  (member of DSA1007:) >   9 	Something seems fishy about the above display.  You haveo; 	exactly 20 errors on DSA1001:, DSA1003: Site 1 members and > 	yet 3 errors on Site 2 member of DSA1001:, 4 errors on Site 2E 	member of DSA1003:.  The fact that 3 separate shadowsets - DSA1001,  @ 	1003, 1005 Site 1 members have 20 errors each hint at a system  	related issue.t? 	(Otherwise - let's pretend they are disk errors - you wouldn'tlD 	have matching numbers.  Error numbers of 21, 27, 134 would be more I 	real.  What do your switches look like - are they reporting any errors,  B 	dropped packets, retransmits?  MountVerifys?  Connect/reconnects? 	OPCOM messages?)n  H > Re directing I/O to a particular DGA, could I not get the same numbers. > working with a single-member DGA shadow set?  ? 	Actually - you should probably get higher numbers.  Definitelyo1 	higher numbers with sequential IO.  But remembera> 	that shadowing is mostly about availability - not performance; 	in many situations.  BUT let's get in our time machine andS= 	go back a few years.  Many of us were shadowing single disksn@ 	across HSJ controller pairs.  In that case, you would certainly< 	get better random IO performance as the shadow driver would; 	either round robin reads to shadow members or queue lengthr< 	would take over and shdriver would go to least busy.  Also,: 	since there was no sequential prefetch - throwing largish< 	(64K or 32K) IO at single 7200 RPM drives doesn't take much? 	to saturate, the second drive to read from helps a great deal. > 	So performance was often greatly improved two drives shadowed9 	versus a single drive.  (Of course the next argument wascH 	always:  What about shadowing versus RAID?)  However, now that many of @ 	us are shadowing RAID and the backend can prefetch, Maybe this 	 	applies:h   --------  Z http://groups.google.com/groups?selm=tK1NSCJOuvIE%40eisner.encompasserve.org&output=gplain  ; 	One final comment.  How about tweeking the crusty SHDRIVERo@ 	algorithm?  The idea there would be pick a target on sequential= 	IO.  If the IO response gets dramatically better after 3 IOsa= 	(or so), then continue to send to that target.  If not, moveg9 	on to the next target, send 3 IOs, if not better move ont> 	to the 3rd target.  If after testing all three targets and noF 	improved IO, go into round robin mode.  Reasons are you may have HSJ E 	shadowed to HSG or good pre-fetching 3rd party storage.  Or you may iF 	have good third party storage or HSG80 storage and round-robining on < 	sequential IO doesn't make sense if you can force pre-fetch; 	to contoller cache.  The ugly work around for those in ther, 	know is to fiddle with READ_COST - painful.   -----   E 	All that to say: you get better results if only one of those drives lE 	are fulfilling read requests as prefetch logic kicks in.  And yet...-B 	I'm reading recently where vendors are improving their algorithmsE 	to recognize out of sequence sequential requests (stepped?  I forget > 	the terminology) and are prefetching on those.  I believe EVA= 	does that.  Adaptive Read-Ahead caching it is called in EVA:   D http://h18006.www1.hp.com/products/quickspecs/11008_na/11008_na.html    2 > BTW, how can I show the read costs of a device,   > 	You can view it.  But remember - you determine it and set it.: 	It isn't as if you will find that one device has a higher 	or a lower read_cost.     	This is an excellent piece:  : http://h71000.www7.hp.com/openvms/fibre/fc_hbvs_dtc_wp.pdf  J The new /READ_COST qualifier to the SET DEVICE command lets you modify anyK members read cost on a per-system basis. This qualifier can be used to loadoL balance the applications read I/O in a variety of ways. The /READ_COST valueN assigned to a device can be determined by using the SDA (System Dump Analyzer) command SHOW DEVICE DSAnnnn.  J For example, in a configuration with two members at one site (connected toM different controller pairs) and a third member at the other site, the defaulthN values will evenly distribute the read requests between the two local members.I If there is a need to offload a specific local controller for performanceeK reasons, the ability to adjust the /READ_COST qualifier on one member makes L this possible. The SET DEVICE /READ_COST command can be used to simulate theO I/O performance characteristics of the application when the local device is not N available or when that member is no longer in the shadow set. To revert shadowN set members to their default values after such changes have been done, use theK command: SET DEVICE /READ_COST = 1 DSAnnnn. This command will cause HVBS toS= restore the default values for all of the shadow set members.r    ; 	What I would think in your case (assuming it isn't privatet8 	fibre but shared comm site<->site link) you may want to@ 	minimize Site-to-Site traffic.  Judicious use of READ_COST willA 	ensure all READs stay local (of course writes go to both sites).y  , > and how did you display  the test results?  2 	You mean the IOT stuff?  I ran IOT against a DGA.  - 	I make this statement:  "It's all about IO."y= 	Co-workers like to remind me of that statement and poke some F 	mild fun at me.  It really is all about IO (unless of course you are C 	crunching numbers - but we live in the business world - it is all m 	about IO).l   				Robl   ------------------------------  % Date: Mon, 25 Aug 2003 11:19:50 +0200r+ From: "Hans Vlems" <hvlems.nieuw@zonnet.nl>-1 Subject: Re: M7940/DHV-11, TU58, and a VAX 11/75039 Message-ID: <bickbo$7sh5s$1@ID-143435.news.uni-berlin.de>l  ) <misc@vectorgames.org> schreef in berichtl' news:bianvs$tvd@library2.airnews.net...3 > Hans Vlems wrote:  >  > >> > >lI > > IIRC the 11/750 could boot directly from RL02 and RA80's depending onG theO > > selector switch position..K > > Provided that the bootblock was properly setup with writeboot. That way, youn/ > > don't need a TU58, they're not that fast...t > >d >  >eJ > Unfortunately I don't have (I don't think....) a bootable RL02 with VMS.H > I will check again, though. I may have a working RL02 diskpack from an > 8600-series. > C > This 11/750 is being brought-up "cold" - meaning I don't have anyi. > bootable media for it yet to start into VMS. >gH > It's a rescue project - all of the components of it and my 11/780 wereE > rescued from a computer scrapyard and restored. The 750 seems to beeL > working correctly now, and the next step is to make it part of my cluster. >sH An RL02 is not big enough to hold VMS. If you have an 11/780, how do you boot VMS on that system?J The reason I ask is this: should you have an RA class drive then you couldJ dual home that drive and connect the UDA-50 of the 11/750 to the "B" port.H Assuming you do have a UDA-50 in the 750, right? AFAIK there's no way to boot an 11/750 across a LAN.   ------------------------------  % Date: Mon, 25 Aug 2003 09:41:30 +0200e: From: Karl Rohwedder <extern.karl.rohwedder@volkswagen.de>7 Subject: Re: Mount DQB0 gets 'allready allocated' errors, Message-ID: <biceip$oqm5@doiweb4.wob.vw.vwg>   Karl Rohwedder wrote:h  F > I try to mount a CDROM (DQB0) on a DS10 and get the 'device already % > allocated to another user' - error,g/ > though SHOW DEV /FULL display an empty owner:e > 
 > $ allo dqb0e> > %SYSTEM-W-DEVALLOC, device already allocated to another user > " > $  moun/over=id/media=cdrom dqb0; > %MOUNT-I-OPRQST, device already allocated to another useroJ > %MOUNT-I-OPRQST, device _$1$DQB0: (QAK01) is not available for mounting. >  > $ sh dev dqb0/fu > . > Locks queued to device lock "SYS$_$1$DQB0:": > H >    - Lock 7A002A8A,  length of items 00180018,  $ENQ status 00000001, 1 > value block 00000000 00000000 00000000 00000000m9 >    -   LOCKID     PID      SYSID   (node)     REMLKID  s+ > REMSYSID           RQMODE  GRMODE   STATEsB >    -  7A002A8A  000060B5  00000000 ()        7A002A8A  00000000 9 > ()          00      00     01     [ NL / NL / GRANTED ], > C > Disk $1$DQB0: (QAK01), device type Compaq  CRD-8402B, is online, d > file-orientedmH >     device, shareable, available to cluster, error logging is enabled. > 5 >     Error count                    0    Operations s > completed                907I >     Owner process                 ""    Owner UIC                       
 > [SYSTEM]> >     Owner process ID        00000000    Dev Prot             > S:RWPL,O:RWPL,G:R,Wg9 >     Reference count                1    Default buffer t > size                 5126 >     Total blocks             1290232    Sectors per  > track                     8t5 >     Total cylinders            40320    Tracks per e > cylinder                   4& >     Allocation class               1 >  > PID 60B5 is myself.a > . > Has somebody the same problem or a solution? >    The problem is solved:  E The user tried to create a Pathworks Share on this CDROM, but failed. G The PWRL$LMSRV process has a channel open on DQB0. Restarting Pathworksa freed the device..   -- ,  + mit freundlichen Gren | with best regardse    Karl Rohwedder		iT-Ingenieurteam$ Ellernbruch 11		D-38112 Braunschweig  E mailto:it-ingteam(at)t-online.de | mailto:rohwedder(at)decus.decus.de - mailto:extern.karl.rohwedder(at)volkswagen.dee   ------------------------------  % Date: Mon, 25 Aug 2003 11:51:02 -0400a0 From: "Brian Tillman" <Tillman@sparkingwire.com>- Subject: Re: mounting remote disks on VAX/VMSe$ Message-ID: <3f4a3071$1@news.si.com>  D >One person at work indicates that this cannot be done, that you canE >only mount drives from another VAX box, not one from a Linux or Unix  >box.r  J That person is clueless.  I have VMS-hosted drives mounted on Unix systems$ via NFS and on PC systems via Samba. -- aI Brian Tillman         Internet: Brian.Tillman at smiths-aerospace dot comt5 Smiths Aerospace  Addresses modified to prevent SPAM.tD 3290 Patterson Ave. SE, MS 1B3 Replace "at" with "@", "dot" with "." Grand Rapids, MI 49512-1991-8        This opinion doesn't represent that of my company   ------------------------------  # Date: Mon, 25 Aug 2003 07:50:34 GMT / From: Milton <mbhewitt@SPAMBLOCK.optonline.net> 2 Subject: Re: OT: 64 bit desktop computing is here.8 Message-ID: <vpejkvsj7h4frq3b0m99ro06au557didjs@4ax.com>  H On Sun, 24 Aug 2003 20:31:21 -0400, JF Mezei <jfmezei.spamnot@istop.com> wrote:  M >While people were arguing whether AMD might survive long enough to introducet* >its 64 bit 8086, Apple was hard at work.  >tJ >They have now introduced the G5, which is a 64 bit Power4 core with addedD >gadgets such as Apple's velocity engine. 13nm copper process. 2Ghz. >  >http://www.apple.com/ca/G5  >iO >Oh, and it can run 32 bit MAC applications natively with no performance hit or  >need for an emulator. > 
 >Power Mac G5  >Dual 2GHz PowerPC G5 - >SPECint_base2000: Integer calculations: 	800g2 >SPECfp_base2000: Floating-point calculations: 8402 >SPECint_rate_base2000: Integer calculations: 17.28 >SPECfp_rate_base2000: Floating-point calculations: 15.7 >iL >In the performance white paper, they only compare it to the 3ghz 8086. TheyL >also compare real world applications perforance such as photoshop. It wouldE >have been interesting to see comparisons with Alpha EV7 and whateverp* >version-du-jour IA64 has available today. >tL >Can anyone comment on whether Apple's G5 truly ranks as the world's fastestP >desktop processor or will it be outranked by AMD's offering when it comes out ? >97 >And how does it fare against IA64 based workstations ?a >dL >Also, how does the 3mhz 8086 compare to whatever is currently available for+ >IA64 if you exclude non-CPU differences ?   >,G >Interestingly, the G5 achieves its perforance with only 512k L2 cache..   Read5 http://www.amdzone.com/articleview.cfm?articleid=1296$  E Here's the real SPEC CPU2000 and the false results from Apple, all ona2 the same graph. It also contains opteron results. < http://www.amdzone.com/articleview.cfm?articleid=1296&page=2     Cheers,M Milton   ------------------------------  % Date: Mon, 25 Aug 2003 10:59:17 +0200e$ From: Michael Unger <unger@decus.de>2 Subject: Re: OT: 64 bit desktop computing is here.9 Message-ID: <biclbp$7m23l$1@ID-152801.news.uni-berlin.de>   % On 25-Aug-2003 02:31, JF Mezei wrote:e  N > While people were arguing whether AMD might survive long enough to introduce+ > its 64 bit 8086, Apple was hard at work. r > K > They have now introduced the G5, which is a 64 bit Power4 core with addediE > gadgets such as Apple's velocity engine. 13nm copper process. 2Ghz.d/                                            ^^^^e >  > [...]   9 That would be an "extremely deep UV chip process" ... ;-)e I guess you meant "130 nm".    Michaelf   -- i; Real names enhance the probability of getting real answers.s@ Please do *not* send "Security Patch Notifications" or "SecurityA Updates"; this system isn't running a Micro$oft operating system.-= And don't annoy me <mailto:postmaster@[127.0.0.1]> please ;-)o   ------------------------------  % Date: Mon, 25 Aug 2003 09:49:55 -04000* From: JF Mezei <jfmezei.spamnot@istop.com>2 Subject: Re: OT: 64 bit desktop computing is here.) Message-ID: <3F4A1402.212763FF@istop.com>o  
 Milton wrote:-7 > http://www.amdzone.com/articleview.cfm?articleid=1296:  O Well, a web site dedicated to AMD isn't exactly the most likely to be unbiased.L   ------------------------------    Date: 25 Aug 2003 10:22:44 -07001 From: keithparris_NOSPAM@yahoo.com (Keith Parris)CR Subject: Re: overhead with host-based volume shadowing for a one-member shadow-set= Message-ID: <cf15391e.0308250922.77d215a0@posting.google.com>   | helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) wrote in message news:<bi8jo1$pss$1@online.de>...F > My mid-term goal is to have every logical disk be a two-membered (orF > even three-membered) shadow set.  To make things easier when working> > towards this goal (i.e. less to think about when editing theG > disk-mounting procedure which is run at startup), I'm thinking about  I > making ALL disks a shadow set, even if at the moment there is only one n	 > member.h > , > How much additional overhead will this be?  F Not much.  Basically SHDRIVER will take an I/O request and re-queue itB directly to the disk driver handling the only member present.  Not/ much to do, so not many CPU instructions there.r  K > Also, I now mount all disks on all nodes, even if these disks are rarely (H > accessed from a node other than the one they are connected to via the G > SCSI bus (for example, system disks).  It can be convenient, though, eI > for example when running a procedure to show information about all the nE > disks.  Again, how much additional overhead is this (i.e. the MSCP   > serving).   ? When there's no actual I/O going on, as I recall, about all the F overhead you get is an MSCP Get Unit Status message exchange from eachB node to each MSCP-serving node about every 15 seconds.  I wouldn't worry about it.C   ------------------------------    Date: 25 Aug 2003 04:39:13 -0700* From: Anders.Wallin@om.com (Anders Wallin)) Subject: Patching Ident field of an Imagei= Message-ID: <79de16e3.0308250339.2aaf3fda@posting.google.com>L   Hello,  @ I would like to stamp the "Ident" field of an image with a valueE reflecting a released system version. This information is however notp? available during linking so I can't place it in an option file.e  7 Does anyone have a utility that can do this job for me?   $ I'm using Alpha, VMS 7.2 and higher.   Regardsm
 Anders Walling   ------------------------------    Date: 25 Aug 2003 07:18:51 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)-- Subject: Re: Patching Ident field of an Image"3 Message-ID: <0dvg6iF2fdRE@eisner.encompasserve.org>o  j In article <79de16e3.0308250339.2aaf3fda@posting.google.com>, Anders.Wallin@om.com (Anders Wallin) writes:  B > I would like to stamp the "Ident" field of an image with a valueG > reflecting a released system version. This information is however not A > available during linking so I can't place it in an option file.. > 9 > Does anyone have a utility that can do this job for me?t > & > I'm using Alpha, VMS 7.2 and higher.  F Instead of the Image File Identification, would you be able to use the! Image File Build Identification ?l  ? I believe that field can be set by the PRODUCT PACKAGE command.5; If you create a Reference kit rather than a Sequential kit,e< the modified image will be readily available to you, and you; can discard the byproducts of that PRODUCT PACKAGE command.c   ------------------------------    Date: 25 Aug 2003 08:07:01 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)l- Subject: Re: Patching Ident field of an Image:3 Message-ID: <vOsWYXZjQFSK@eisner.encompasserve.org>o  j In article <79de16e3.0308250339.2aaf3fda@posting.google.com>, Anders.Wallin@om.com (Anders Wallin) writes: > Hello, > B > I would like to stamp the "Ident" field of an image with a valueG > reflecting a released system version. This information is however not8A > available during linking so I can't place it in an option file.  > 9 > Does anyone have a utility that can do this job for me?r  A    If it's not available during link, how and when does it becomee%    available?  IMHO, I'd change that.M   ------------------------------   Date: 25 Aug 03 09:46:39 PST From: mckinneyj@cpva.saic.com - Subject: Re: Patching Ident field of an Image.( Message-ID: <Qq5JIpJydbsZ@cpva.saic.com>  = In article <79de16e3.0308250339.2aaf3fda@posting.google.com>,s-  Anders.Wallin@om.com (Anders Wallin) writes:f > Hello, > B > I would like to stamp the "Ident" field of an image with a valueG > reflecting a released system version. This information is however notSA > available during linking so I can't place it in an option file.r > 9 > Does anyone have a utility that can do this job for me?  > & > I'm using Alpha, VMS 7.2 and higher. > 	 > Regards  > Anders Wallin-  * Maybe you can just use the editor (TPU)...  D The ident is a counted ascii string with the count beginning at %xF0A of the first record in the image file. Use DUMP/PAGE on your .EXE>* and you'll see what ANALYZE/IMAGE reports.  E So, assign some value to the ident field during your build to make it E easy to identify in the editor. (Try and make sure the length of thismE string is the same as the one that you'll use to replace it or you'llgF have to use a spec-ins sequence in the editor to alter the count valueF that precedes the ident string too). Edit the file, change the string,H and exit. Then you'll need to CONVERT/FDL the resultant file back to itsF binary format (you can $ ANA/RMS/FDL the original .EXE to help provide input to CONVERT).   ------------------------------    Date: 25 Aug 2003 10:26:47 -07001 From: keithparris_NOSPAM@yahoo.com (Keith Parris)u; Subject: Re: Question about a free shell provider - VisTech$= Message-ID: <cf15391e.0308250926.79ed0d46@posting.google.com>w  r djcameron60616@yahoo.com (stg-delfuego) wrote in message news:<6b28e43f.0308231600.2ba19449@posting.google.com>...H > 3.  Anyone know of a good site offering free user accounts on VMS that > is available?>  E I recommend ENCOMPASServe (aka DECUServe).  Lots of good information, 5 and a friendly community of users as well.  TELNET to 9 encompasserve.org and log in under username REGISTRATION.a   ------------------------------  % Date: Mon, 25 Aug 2003 19:06:43 +0200h, From: Albrecht Schlosser <ajs856@tiscali.de>4 Subject: Re: Reconstructing pdf file broken via mail, Message-ID: <7mfdib.p58.ln@news.hus-soft.de>   JF Mezei wrote:  > H > Someone at a government tried to send me a PDF file. Unfortunatly, theR > contents were sent from some microsoft box, obviously not adhering to standards. > N > It was sent as application/octet-stream, and instead of encoded as base64 or3 > uuencode, it was sent as quoted-printable "text".6 > P > The email was picked up by an application of mine from a pop mailbox. It opens > the VMS mail file with:pP > fopen("thefile","w","rfm=var","rat=cr") and uses fputs() to write each line to > the file.m > N > the MAIL application barfed on the message due to a record being longer thanN > some 5000 bytes, so I was not able to extract it. But I was able to find the# > actual VMS file and play with it.s  E One _possible_ solution: If it was sent as text, would you be able to H forward the mail to a pc as text? If this can be done, you might be ableG to store the file on the pc as text. Hopefully, this is the same as thes? original file. Then, use it or ftp it (binary) to your VMS box.C  H I know, this has only a small success probability, but I would give it a try ...i  
 Good luck.   Albrecht   ------------------------------  % Date: Mon, 25 Aug 2003 13:20:50 -0400o* From: JF Mezei <jfmezei.spamnot@istop.com>4 Subject: Re: Reconstructing pdf file broken via mail) Message-ID: <3F4A456E.28052605@istop.com>g   Albrecht Schlosser wrote:)G > One _possible_ solution: If it was sent as text, would you be able tot# > forward the mail to a pc as text?M  M I don't have a PC... Forwarding it through SMTP would increase the likelyhood K of CR/LF screwups because the file would be trated by VMS and possibly some  other SMTP server.  M The problem is that some data may have bene permanently lost during receptionfM from the pop server to the VMS host when I treat incoming data from the TCPIPe< pipe and store it as a text file in variable record format.   F I know that the file does contains some CR characters in the middle ofM vbariable length records. However, I have no idea how much damage to the filenI may have occured during transport between the government and myself (manyt servers involved).  J It is surprising that some microsoft software would/could be configured soI poorly as to send binary attachements without any RFC compliant encoding.i   ------------------------------  % Date: Mon, 25 Aug 2003 14:07:58 +0200 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl> ; Subject: Re: setting up a LAT service via a terminal serverw9 Message-ID: <bicu72$7ksum$1@ID-143435.news.uni-berlin.de>n  L "Phillip Helbig---remove CLOTHES to reply" <helbig@astro.multiCLOTHESvax.de>1 schreef in bericht news:bi7sj9$3la$1@online.de...nI > Apart from the consoles, I have a VT320 connected to a serial port of aoE > node in my hobbyist cluster.  If that node is down for some reason,cC > however, then of course one can't log in via that terminal.  (TheAG > terminal in question is in my son's bedroom.  How many 9-year-olds doyJ > you know who know how to use EDT?  One of my biggest triumphs was when IE > heard him say "Mama's PC doesn't have a PROPER keyboard"!)  Thus, IoH > would like to have the terminal connect to the CLUSTER instead of to a > specific node.   :-)c   >eH > It seems the way to do this would be to connect via a terminal server.G > I have a DECserver 250 and a DECserver 200/MC.  Presumably these willgI > work fine for what I want to do.  I also have the necessary cables etc.  >tH > Unfortunately, this stuff is located 500 km from where I spend most ofE > my time, so I don't have much time to experiment.  What I need is anD > step-by-step guide to setting up what I need.  Presumably, the LATG > service has to be created somehow, perhaps the terminal server has toa > boot from somewhere etc. >   A I have no knowledge of the DS250, it was IIRC intended to connect  lineprinters to ethernet.AJ A DS200 gets its microcode from a load host. The DS200 uses PR0801ENG.SYS,E the DS100 uses PS0801ENG.SYS. I can email them to you, they're small.eH They reside in SYS$SYSROOT:[MOM$SYSTEM], the logical name that points to that is MOM$LOAD.U3 ou can put the download incantations in NCP, like:l   $ mc ncp sho node astaat charu8 Node Volatile Characteristics as of 25-AUG-2003 13:31:15 Remote node =   1.8 (ASTAAT)    Service circuit          = EWA-0& Hardware address   = 08-00-2B-06-3A-E83 Load file                  = MOM$LOAD:PS0801ENG.SYS.  F The decserver 100 has a DECnet address only to put the data in the ncpK database. DECnet is not used, only MOP in the communications. Note that the,: circuit (here EWA-0) must have its service cicuit enabled:   $ mc ncp sho circ ewa-0 char  ; Circuit Volatile Characteristics as of 25-AUG-2003 13:33:21i   Circuit = EWA-0i   State                    = ono" Service                  = enabled) Designated router        =  1.39 (KALIUM)B Cost                     = 4  I When the DS200 runs, you can use NCP to connect to the server:  $  mc ncps connect node astaatt  C You get the 3 prompt in return. Type the password ACCESS to get in.m1 Once in you must make yourself a privileged user:-   Local> set prive SYSTEM  L If the DS200 was in use somewhere then it may have different values. you canG reset the DS200 to factory settings as follows: Make sure the device iscG power off. Push the square red button on the back and connect the powern: cord. Once the DS200 has booted, let go of the red button.  H > I have no experience with LAT other than using SET HOST/LAT to connectI > to other nodes (in the cluster) and no experience with terminal serversf > (except as an end-user). > J > Ideally, switching on the VT and hitting RETURN would bring up the login, > sequence from some machine in the cluster.  G The DS200 uses define and set commands in the same way as NCP does, fordI permanent and volatile storage. The DS200 has a pretty comprehensive helpeL tool built in. HELP DEFINE PORT will give you a pretty good idea what to do.L The ports of the DS200 may be configured one by one or all at the same time:  0 Local> define port all session limit 1 speed 300) Local> define port 1 speed 9600 type ansi_  I Port 1 is (by default) the console port. Put a VT on it when you power up , the DS200, it'll tell you what is happening.K The DS200 has only 64 kB memory. Most of it is used to buffer the sessions.:I So memory is spent on speed and sessions. If you want to conserve memory,-L reduce the session limit to 1 and the speed to 300 baud for all ports. LaterG you can set the attributes for the port(s) that you are going to use at  reasonable values, like:  F Local> define port 2 type ansi preferred service MYCLUSTER autoconnect enable  D The last command will connect that port to the LAT service MYCLUSTER automatically.     >aH > At the moment, the cluster has a VAX at 7.3, a VAX at 7.2 and an ALPHAG > at 7.2-1.  There is no DECnet running.  LAT is configured so that one H > can SET HOST/LAT to and from any machine in the cluster.  With TCPIP IH > have a cluster alias so that to the outside world, the cluster behavesE > as one machine.  Ideally, I'd like the same functionality with LAT._    On each node in the cluster use:  # $ mc latcp create service MYCLUSTERo      (In> > fact, since the connection to the outside world is via a DSLG > switch/router which does PAT and forwards incoming connections to the"@ > cluster alias, not only is there just one address for INcomingI > connections, but just one address for OUTgoing connections as well, therH > external address of the router.  (Normally, a cluster alias applies toG > just incoming connections, whereas each node will use its own addressi > for outgoing connections.))I >nG > Load-balancing would be nice just to get some experience with it, butmG > for my purposes failover (like with the TCPIP cluster alias) would ber	 > enough.o >wH > While I'm at it, I have an LNO3 sitting around.  Can I connect that to > the terminal server as well? >.L Yes.  You need the LATsymbiont to connect to printers. And create a LAT port on the VMS system, like:   $       mc latcp set node/conn=both set node/service_responder cre port lta8:/applic/log 0 set port lta8:/queue/node=astaat/port=port_8/log exit  : I'm sure this is not complete but it should get you going.   Hans   ------------------------------  % Date: Mon, 25 Aug 2003 12:14:30 -0400p0 From: "Brian Tillman" <Tillman@sparkingwire.com>; Subject: Re: setting up a LAT service via a terminal server_$ Message-ID: <3f4a35f1$1@news.si.com>  E >> You can set up a terminal server to allow you access to all of thelH >> "services"/nodes that are out there using the LAT protocal. You first= >> connect to the Terminal Server, and then issue the CONNECT- >> {nodename/service} command. >0I >OK, I remember doing this years ago at the university, but how do I "setv! >up the service" on the VMS side?2   In LAT$SYSTARTUP.COM:m   ...i $       lcpt !g) !           Define cluster alias servicesT !E)             create service SWDEV /nolog -s2                 /ident="VAXcluster - Agvax,Benzie"*             create service ENGDEV /nolog -2                 /ident="VAXcluster - Agvax,Benzie" ...t  K where "SWDEV" is the service offered by each node of the cluster that is to  participate.  Put this prior toa   $! $!      Start the LAT protocol $! $       lcp set node /state=on    6 On the terminal service side (with privileges enabled)  ; define port xxx preferred service SWDEV autoconnect enabled  logout port xxxg   Booting the server requires:   $ mcr ncp sho nod dsv034 char     8 Node Volatile Characteristics as of 25-AUG-2003 11:59:38   Remote node =   8.34 (DSV034).    Service circuit          = ISA-0, Hardware address         = 08-00-2B-05-9D-9B> Load file                = SYS$COMMON:[DECSERVER]PR0801ENG.SYS> Dump file                = SYS$COMMON:[DECSERVER]DS2DSV034.DMP  G so you have to have PRO801ENG.SYS (for a DECserver 200) in the MOM$LOAD 0 directory.  DECserver 250 requires DP0601ENG.SYS --  I Brian Tillman         Internet: Brian.Tillman at smiths-aerospace dot comd5 Smiths Aerospace  Addresses modified to prevent SPAM.-D 3290 Patterson Ave. SE, MS 1B3 Replace "at" with "@", "dot" with "." Grand Rapids, MI 49512-1991r8        This opinion doesn't represent that of my company   ------------------------------  % Date: Mon, 25 Aug 2003 18:21:44 +0200o, From: Albrecht Schlosser <ajs856@tiscali.de>; Subject: Re: setting up a LAT service via a terminal serverd, Message-ID: <r1ddib.ta7.ln@news.hus-soft.de>   Phillip Helbig wrote:S > I > Apart from the consoles, I have a VT320 connected to a serial port of aoE > node in my hobbyist cluster.  If that node is down for some reason,tC > however, then of course one can't log in via that terminal.  (ThepG > terminal in question is in my son's bedroom.  How many 9-year-olds dosJ > you know who know how to use EDT?  One of my biggest triumphs was when IE > heard him say "Mama's PC doesn't have a PROPER keyboard"!)  Thus, InH > would like to have the terminal connect to the CLUSTER instead of to a > specific node. > H > It seems the way to do this would be to connect via a terminal server.G > I have a DECserver 250 and a DECserver 200/MC.  Presumably these willaI > work fine for what I want to do.  I also have the necessary cables etc.o > [...]t > H > At the moment, the cluster has a VAX at 7.3, a VAX at 7.2 and an ALPHA) > at 7.2-1.  There is no DECnet running. k  F If you don't have DECnet running, then you need LANCP to configure theH MOP loading. As others wrote about the DECserver, LANCP also has set andF define commands. To enable MOP (aka DLL), you should do something like   $ lancp :== $lancp4 $ lancp def dev ewb0 /mop=(enable,exclusive,noknown) $ lancp list dev ewb0 /mop    # Device Listing, permanent database:u;               --- Downline Load Service Characteristics ---EA   Device     State   Access Mode      Clients           Data Size A   ------     -----   -----------      -------           ---------zB   EWB0      Enabled  Exclusive    NoKnownClientsOnly     246 bytes $   D (Replace ewb0 with your network device, see "LANCP show device). See% LANCP HELP (or the docs) for details.   D This enables MOP loading for all clients, even "unknown" clients, ifH they request a specific file (PR0801ENG.SYS for DECserver 200). You need@ only this file to get a DS-200 started. For LANCP, the load file1 directory is defined by the logical name LAN$DLL:'   $ sho log lan$dlly<    "LAN$DLL" = "SYS$SYSROOT:[MOM$SYSTEM]" (LNM$SYSTEM_TABLE)   Usee   $ lancp sho mop    LAN MOP DLL Status:    EWA0 disabled1F   EWB0 enabled in exclusive mode for all requests, data size 246 bytes  C        #Loads  Packets   Bytes     Last load time       Last loaded1G        ------  -------   -----  --------------------  -----------------D   EWA0      0        0       0G   EWB0      1     1712  207763  20-AUG-2003 17:18:17  08-00-2B-16-5F-58)  ? to see the MOP/DLL status (MOP and DLL are synonyms wrt LANCP).   F As you can see, I have LANCP DLL enabled on EWB0 only, because EWA0 is served by DECnet MOP/DLL (***):e   $ ncp sho circ ewa-0 char   ; Circuit Volatile Characteristics as of 25-AUG-2003 18:20:30   m Circuit = EWA-0y  o State                    = ono( Service                  = enabled (***)   > LAT is configured so that one H > can SET HOST/LAT to and from any machine in the cluster.  With TCPIP IH > have a cluster alias so that to the outside world, the cluster behavesE > as one machine.  Ideally, I'd like the same functionality with LAT.   F As others stated, more than one node can offer the same service. ThereH are examples in SYS$MANAGER:LAT$SYSTARTUP.TEMPLATE. If you have not doneH so, copy this file to SYS$MANAGER:LAT$SYSTARTUP.COM and modify it as youF need. In fact, this file is called by LAT$STARTUP.com, which is called from SYSTARTUP_VMS.COM.l  B > (In fact, since the connection to the outside world is via a DSLG > switch/router which does PAT and forwards incoming connections to the @ > cluster alias, not only is there just one address for INcomingI > connections, but just one address for OUTgoing connections as well, thetH > external address of the router.  (Normally, a cluster alias applies toG > just incoming connections, whereas each node will use its own address  > for outgoing connections.))  > G > Load-balancing would be nice just to get some experience with it, butoG > for my purposes failover (like with the TCPIP cluster alias) would bee	 > enough.t  C You don't want to route LAT over the internet, do you? LAT can't beg0 routed (LAN only), but you surely know this :-).  H > While I'm at it, I have an LNO3 sitting around.  Can I connect that to > the terminal server as well?   Yes.   ------------------------------  + Date: Mon, 25 Aug 2003 18:30:34 +0200 (MET)a9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> ; Subject: Re: setting up a LAT service via a terminal serverc; Message-ID: <01KZVZPR1LTAAMSOR1@sysdev.deutsche-boerse.com>p  H Thanks to you and Brian for the LANCP examples; I should have mentioned ' that I don't (yet) have DECnet running.o  ! > > LAT is configured so that onegJ > > can SET HOST/LAT to and from any machine in the cluster.  With TCPIP IJ > > have a cluster alias so that to the outside world, the cluster behavesG > > as one machine.  Ideally, I'd like the same functionality with LAT.  > H > As others stated, more than one node can offer the same service. ThereJ > are examples in SYS$MANAGER:LAT$SYSTARTUP.TEMPLATE. If you have not doneJ > so, copy this file to SYS$MANAGER:LAT$SYSTARTUP.COM and modify it as youH > need. In fact, this file is called by LAT$STARTUP.com, which is called > from SYSTARTUP_VMS.COM.  > D > > (In fact, since the connection to the outside world is via a DSLI > > switch/router which does PAT and forwards incoming connections to the B > > cluster alias, not only is there just one address for INcomingK > > connections, but just one address for OUTgoing connections as well, the-J > > external address of the router.  (Normally, a cluster alias applies toI > > just incoming connections, whereas each node will use its own addressi > > for outgoing connections.))o > > I > > Load-balancing would be nice just to get some experience with it, but I > > for my purposes failover (like with the TCPIP cluster alias) would be  > > enough.c > E > You don't want to route LAT over the internet, do you? LAT can't be 2 > routed (LAN only), but you surely know this :-).  G The TCPIP cluster alias was just an example for failover as opposed to aG load balancing.  With respect to LAT, "outside world" means logging in _F from a terminal on the LAN.  All I need is failover; does LAT do load  balancing by default?e   ------------------------------  % Date: Mon, 25 Aug 2003 18:46:15 +0200 , From: Albrecht Schlosser <ajs856@tiscali.de>; Subject: Re: setting up a LAT service via a terminal serverd, Message-ID: <rfedib.kq7.ln@news.hus-soft.de>   Phillip Helbig wrote:  > 1 > Thanks to you and Brian for the LANCP examples;n   welcomen   > I should have mentionedr) > that I don't (yet) have DECnet running.e  1 You did, that's why I added the LANCP reply  :-).d   > [...]eH > The TCPIP cluster alias was just an example for failover as opposed toH > load balancing.  With respect to LAT, "outside world" means logging inG > from a terminal on the LAN.  All I need is failover; does LAT do loadO > balancing by default?   4 LAT _can_ do it, there is something called "Rating":   $ lcp sho servi alpha1    A Service Name:     ALPHA1                   Service Type:  GeneralsB Service Status:   Available                Connections:   Enabled  ...g  8 Node Name            Status      Rating   IdentificationH ALPHA1               On            82 D   Alpha1 (AlphaServer 800) *NEU*  H AFAIK, the rating of a specific service is calculated somehow related to9 the cpu power and load of the machine. I never used this.e   Albrecht   ------------------------------  % Date: Mon, 25 Aug 2003 12:50:38 -0400l* From: JF Mezei <jfmezei.spamnot@istop.com>; Subject: Re: setting up a LAT service via a terminal servers) Message-ID: <3F4A3E5D.5E51EB34@istop.com>a   Phillip Helbig wrote:IG > from a terminal on the LAN.  All I need is failover; does LAT do loadm > balancing by default?o  L LAT does load balancing by default when you have a cluster. Each node in theJ cluster sends a regular "I am not busy" or "I am busy" or "I am very busy"H message to servers. Then, when a user requests a connection to a serviceL served by many nodes, the terminal server choose the one that is least busy.K However, once connected, the terminal server doesn't dynamically swicth you3 from one node to another.   J However, if your terminal server port is set to auto connect to a service,N when the terminal server realises that the node to whcih you are connected hasH failed, it will disconnect you , remove that node from the nodes able toN provide that service, and then automatically reconnect yo to the service whichL will get you to a different node. However, from there, you must re-login and starts from scratch.   ------------------------------  + Date: Mon, 25 Aug 2003 19:19:53 +0200 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> ; Subject: Re: setting up a LAT service via a terminal serverh; Message-ID: <01KZW1GNFDKAAMSOR1@sysdev.deutsche-boerse.com>   I > > from a terminal on the LAN.  All I need is failover; does LAT do loade > > balancing by default?y > J > LAT does load balancing by default when you have a cluster. Each node inJ > the cluster sends a regular "I am not busy" or "I am busy" or "I am veryH > busy" message to servers. Then, when a user requests a connection to aJ > service served by many nodes, the terminal server choose the one that isB > least busy. However, once connected, the terminal server doesn't3 > dynamically swicth you from one node to another. K  * Wow!  This feature is, what, 20 years old?  D The load broker makes something similar possible with TCPIP.  (I've A never used this; since the outside world only has one address to  F connect to through my router, I could only use this internally, which  wouldn't be of much use.)w  C > However, if your terminal server port is set to auto connect to a1G > service, when the terminal server realises that the node to whcih you J > are connected has failed, it will disconnect you , remove that node fromJ > the nodes able to provide that service, and then automatically reconnectI > yo to the service which will get you to a different node. However, fromu4 > there, you must re-login and starts from scratch.   F I don't know if the TCPIP load balancing will automatically present a 0 new login prompt if the connected service fails.   ------------------------------  % Date: Mon, 25 Aug 2003 13:13:56 -04000* From: JF Mezei <jfmezei.spamnot@istop.com>; Subject: Re: setting up a LAT service via a terminal server ) Message-ID: <3F4A43D1.86874676@istop.com>A   Phillip Helbig wrote:0G > from a terminal on the LAN.  All I need is failover; does LAT do load/ > balancing by default?i   MC LATCPK LATCP> HELP SET SERVICE /STATIC_RATING gives you a good explanation of what  happens by default.a   ------------------------------  # Date: Mon, 25 Aug 2003 17:25:18 GMT L From: winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr"); Subject: Re: setting up a LAT service via a terminal server 6 Message-ID: <00A24E7B.A51DED9E@SSRL.SLAC.STANFORD.EDU>  w In article <01KZW1GNFDKAAMSOR1@sysdev.deutsche-boerse.com>, Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> writes:  >5E >The load broker makes something similar possible with TCPIP.  (I've  B >never used this; since the outside world only has one address to G >connect to through my router, I could only use this internally, which   >wouldn't be of much use.) >aD >> However, if your terminal server port is set to auto connect to aH >> service, when the terminal server realises that the node to whcih youK >> are connected has failed, it will disconnect you , remove that node fromaK >> the nodes able to provide that service, and then automatically reconnecteJ >> yo to the service which will get you to a different node. However, from5 >> there, you must re-login and starts from scratch. d >(G >I don't know if the TCPIP load balancing will automatically present a -1 >new login prompt if the connected service fails.3  J If memory serves - and it may not - TCPIP load balancing goes through the I DNS, so it's up to the client to go back to the DNS for a new numeric IP  G if the server fails.  (Which means you need pretty short TTL on the DNSo translations.)   -- Alane   -- oO ===============================================================================h0  Alan Winston --- WINSTON@SSRL.SLAC.STANFORD.EDUM  Disclaimer: I speak only for myself, not SLAC or SSRL   Phone:  650/926-3056tM  Paper mail to: SSRL -- SLAC BIN 99, 2575 Sand Hill Rd, Menlo Park CA   94025MO ===============================================================================a   ------------------------------    Date: 25 Aug 2003 10:07:37 -07001 From: keithparris_NOSPAM@yahoo.com (Keith Parris) E Subject: Re: Shadow or Raid.  - Don't do both!  WAS Re: 306GB drives!i= Message-ID: <cf15391e.0308250907.700da1c2@posting.google.com>e  e "Mike Naime" <mnaime@kc.rr.com> wrote in message news:<0Nz1b.7630$re.167999@twister.rdc-kc.rr.com>...fH > I had my first drive failure in our old CI cluster that uses HSJ's and > shadowing.= > It froze the whole cluster instead of autosparing the disk!   ? Obviously a bug of some sort.  I hope you reported the problem,e) escalated it if necessary, and got a fix.s  5 > On the HSG's a drive failure is a non-event for me./  > I wish HBVS had the equivalent of controller sparesets and and> automatic replacement of failed members.  Folks have to cobble together DCL to do that today.  J > Basically you are stating here that shadowing and RAID technology DO NOT	 > MIX!!!!p. > Either SHADOW, or RAID.  but DO NOT DO BOTH!  > No, I frequently recommend that customers use controller-basedB mirrorsets at each site within a disaster-tolerant VMS cluster, toD provide protection against disk failures after the loss of one site.  J > 2 data centers connected by dark fibre.  I will take EVA's over HBVS any > day!  ? It's obvious that using EVAs and Continuous Access is the rightaE solution for your needs.  Far be it from me to say that you're wrong.r   ------------------------------  % Date: Tue, 26 Aug 2003 00:43:12 +0800w, From: Paul Repacholi <prep@prep.synonet.com>+ Subject: Shadow_max Copy Was: 306GB drives!,0 Message-ID: <87vfsl8z8f.fsf_-_@prep.synonet.com>  3 keithparris_NOSPAM@yahoo.com (Keith Parris) writes:e  B > 2) Ensure that enough SHADOW_MAX_COPY SYSGEN parameter units areD > available so merge threads can be in progress on all shadowsets atE > once.  With insufficient values of SHADOW_MAX_COPY (and the default D > value of 4 per node is often insufficient, especially after you'veD > just lost one node due to the crash that started the merges in theB > first place), a critical disk may have to wait to even start its= > merge operation until another shadowset finishes its merge.c  C Keith, is the number of threads per cluster? IE, if I have a 3 node$B cluster, many shadow sets and all node set at the default 4, if on< node crashes do the other nodes do 4 each, or is it 4 total?   -- e< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.b@                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Mon, 25 Aug 2003 07:49:31 +0200i From: Dirk Munk <munk@home.nl> Subject: Re: SYSnd2 Message-ID: <bic83f$8pf$1@news2.tilbu1.nb.home.nl>   Sean O'Banion wrote:C > SYSF and SYSE are used for Standalone Backup on VAXen, and systemmD > upgrades, respectively:  using these roots (on VAX or, in my case,F > Alpha) for anything else causes problems.  I found this out the hardA > way at HP ETS last year while setting up the Encompass Hands ongF > Workshops.  As a workaround, I used SYSFF and SYSEE for the 15th andE > 16th system: I had named the systems including 00 through 12 hex intG > the name, and intended to have that reflect the root that that system- > booted from. > H > Like you are doing, I configured one disk with all the roots I needed,G > created an image saveset of that disk, did a MOP-based satellite bootmH > as "diskless" (no Votes) into the Cluster, and restored the saveset toH > a local disk on the satellite.  When booting locally, I did not createG > or join a Cluster.  This allowed each system to be nearly identicallyrD > configured, and I only needed to install VMS once, but did need toH > boot from each root to configure DECnet and TCP/IP before creating the
 > saveset. > E > Host-based Shadowing would have worked as well, but would have beenrB > slower, since each block of the destination disk would have been  > examined and possibly written. > H > I think the recommended limit on the number of nodes in one cluster is& > 96, so that might be a useful limit. >  > Q As far as I know the 96 node cluster was tried once and it worked. It is not the     architectural limit.   ------------------------------  + Date: Mon, 25 Aug 2003 15:52:03 +0200 (MET)n9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>t Subject: Re: SYSne; Message-ID: <01KZVU79DEAQAOOLBD@sysdev.deutsche-boerse.com>)  H > The VMS installation CD goes up to SYSFF, so 255 cluster nodes. Should% > be sufficient for a hobby cluster. i  E Actually, I don't want to actually use anything close to the maximum v? number of roots.  Rather, my "master plan" is to intelligently  F coordinate "numbers" such as SCSSYSTEMID, DECnet address, IP address, D ALLOCLASS, numerical suffixes to make volume labels of system disks I unique and so on as well as use the numbers to code up other information 2F (such as architecture, system speed etc).  Thus, I might want to use, E say, 135 not because I have 134 other roots, but because that number a  codes up some other information.   ------------------------------  + Date: Mon, 25 Aug 2003 16:04:05 +0200 (MET)g9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>p Subject: Re: SYSnl; Message-ID: <01KZVUCQQQQ0AOOLBD@sysdev.deutsche-boerse.com>   H > I think the recommended limit on the number of nodes in one cluster is& > 96, so that might be a useful limit.  G That's the largest supported number, yes, but presumably the number of iI system-disk roots coould be less (and, for performance reasons, probably f should be).e   ------------------------------  + Date: Mon, 25 Aug 2003 16:08:32 +0200 (MET)09 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>e Subject: Re: SYSn ; Message-ID: <01KZVUVF5V3IAOOLBD@sysdev.deutsche-boerse.com>a  J > As far as I know the 96 node cluster was tried once and it worked. It is > not the architectural limit.    6 I believe it is the highest number which is supported.   ------------------------------  % Date: Mon, 25 Aug 2003 10:09:19 -0400a) From: Brainwave Surfer <B.S.@hsc.vcu.edu>n Subject: Re: SYSno' Message-ID: <bid5ak$su3$1@mira.vcu.edu>e  G hhmm... sounds like an interesting scheme...  also testing Mozilla, so m0 if this post makes no sense, that's ok.. Jim ...   Phillip Helbig wrote:o  H >>The VMS installation CD goes up to SYSFF, so 255 cluster nodes. Should% >>be sufficient for a hobby cluster. c >  > G > Actually, I don't want to actually use anything close to the maximum hA > number of roots.  Rather, my "master plan" is to intelligently tH > coordinate "numbers" such as SCSSYSTEMID, DECnet address, IP address, F > ALLOCLASS, numerical suffixes to make volume labels of system disks K > unique and so on as well as use the numbers to code up other information eH > (such as architecture, system speed etc).  Thus, I might want to use, G > say, 135 not because I have 134 other roots, but because that number l" > codes up some other information.   -- eG "3,000 years ago I made a mistake."  Elrond Half-Elven, in "Fellowship I of the Ring"  G "I try not to be right any more than necessary". -- Larry Wall, author c of the Perl Language   ------------------------------    Date: 25 Aug 2003 12:27:58 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)m Subject: Re: SYSnQ3 Message-ID: <uztwkoXZSLoc@eisner.encompasserve.org>0  w In article <01KZVU79DEAQAOOLBD@sysdev.deutsche-boerse.com>, Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> writes:m  G > Actually, I don't want to actually use anything close to the maximum mA > number of roots.  Rather, my "master plan" is to intelligently iH > coordinate "numbers" such as SCSSYSTEMID, DECnet address, IP address, F > ALLOCLASS, numerical suffixes to make volume labels of system disks K > unique and so on as well as use the numbers to code up other information TH > (such as architecture, system speed etc).  Thus, I might want to use, G > say, 135 not because I have 134 other roots, but because that number  " > codes up some other information.  G   Been there, done that.  Each boot node used SYS0, each satellite usedeC   SYSx0, x starting at 1.  Never had a problem with not rememberingfG   which satellite was which root because they were named similar to theuA   boot node, but the name ended in x.  This worked for me becauseuF   multiple organizations using the same cluster generally had one bootC   node and several satellites, and that part of the cluster was alleF   named for thier project.  And I didn't have any dual pathed disks atE   first so I didn't have any need for multiple non-satellite roots ona   the same system disk.  7  I   I did use SYS1 and SYS2 for some experimental boot setups, such as one  I   node which could boot from SYS1 as a standalone node, but defaulted to e   SYS0 and entered the cluster.a   ------------------------------  % Date: Mon, 25 Aug 2003 12:15:33 +02006' From: JOUKJ <joukj@hrem.stm.tudelft.nl> * Subject: Re: Tcl/Tk 8.3 or 8.4 for OpenVMS2 Message-ID: <3F49E1C5.1060104@hrem.stm.tudelft.nl>   Petros Dafniotis wrote:S > Jouk-o > ? > cannot ftp to any of the two possible machines you mentioned.  > hrem.stm.tudelft.nlvH This is only for E-mails. It can be used again. My university reapaired  the broken mx-recors.    > and1 > nchrem.tnw.tudelft.nlr> The ftp-servers crashed on 2 of my systems. I do not know why. I restarted it.i    	      Jouk-   >    ------------------------------  % Date: Mon, 25 Aug 2003 15:17:38 +0200r. From: "Hans M. Aus" <aus@vim.uni-wuerzburg.de>1 Subject: Re: Unicode client as Open VMS terminal. B Message-ID: <aus-08675B.15173825082003@wrzx08.rz.uni-wuerzburg.de>   Paul,s  4 Please explain how you got your accented characters.  B From the MacOS Terminal, when I try ssh -l aus nnn.xxx.yyy.1, the  response is:  B ssh: connect to address nnn.xxx.yyy.1 port 22: Connection refused.    F A port scan on the VMS 7.2-1 Alpha reveals that port 22 is not active:    Port   Type   Description  10     TCP    10b  21     TCP    ftp  23     TCP    telnetl  25     TCP    smtp0  111    TCP    sunrpcb  118    TCP    118    * The TCPIP Server configuration looks like:  I   Compaq TCP/IP Services for OpenVMS Server Components Configuration Menub     Configuration options:  I     1 - BIND         Disabled Stopped      11 - NTP      Disabled StoppednI     2 - BOOTP        Disabled Stopped      12 - PC-NFS   Disabled StoppeduI     3 - DHCP         Disabled Stopped      13 - POP      Disabled Stopped I     4 - FINGER       Disabled Stopped      14 - PORTMAPPER Enabled StartecI     5 - FTP          Enabled  Started      15 - RLOGIN   Enabled  Started0I     6 - LBROKER      Disabled Stopped      16 - RMT      Disabled Stopped I     7 - LPR/LPD      Disabled Stopped      17 - SNMP     Disabled StoppedrI     8 - METRIC       Disabled Stopped      18 - TELNET   Enabled  StartedwI     9 - NFS          Enabled  Started      19 - TFTP     Disabled StoppedaI    10 - LOCKD/STATD  Disabled Stopped      20 - XDM      Disabled Stoppeds      ) In article <P87ZeBF8kIMg@elias.decus.ch>,n+  p_sture@elias.decus.ch (Paul Sture) wrote:r   >... tJ > Bingo. After quite a bit of messing around in Terminal, visiting the MacH > Help Center, you suggestion prompted me to abandoned it in favour of a > VMS solution :-) > D > The following got me there for VMS (although not in Terminal - I'd4 > like this to work at the Mac command prompt too*). > I > Login to VMS (with ssh, I have telnet disabled, so haven't tested that)d > % > $ set terminal/device=vt300/eighbitm > . > and I get my accented characters :-) :-) :-) > J > * my mistake all along had been to try getting _Terminal_ to display the! > accented characters on its own.o >  > VMS wins again :-)   -- -B Cheers, Hans M. Aus, Wuerzburg, Germany,  aus@vim.uni-wuerzburg.de   ------------------------------  # Date: Mon, 25 Aug 2003 13:59:04 GMTv# From: "John Smith" <a@nonymous.com>D: Subject: Re: Will OpenVMS I64 run on a Dell PowerEdge 3250I Message-ID: <ICo2b.277179$hOa.89074@news02.bloor.is.net.cable.rogers.com>e  3 "John Travell" <john@jomatech.com> wrote in messager3 news:bi2t7m$4nf99$1@ID-120847.news.uni-berlin.de... % > "Andrew Harrison SUNUK Consultancy"e' <Andrew_No.Harrison_No@nospamn.sun.com>c= > wrote in message news:bi27em$t5a$1@new-usenet.uk.sun.com...  > > Charlie Hammond wrote:9 > > > "John Reagan" <john.reagan@hp.com> wrote in message 1 > > > news:i0r0b.2704$I03.634@news.cpqcorp.net...i > > >a > > >>Alan wrote:i > > >>6 > > >>>Will OpenVMS I64 run on a Dell PowerEdge 3250 ? > > >> > > >a9 > > > The short term answer is almost certainly "NO", butp< > > > Dell's response to this question would be interesting. > > >  > >c > >  > > Why ???r > > = > > Dell needs some formal mechanism probably with supportinge9 > > test suites to allow them to certify OpenVMS on theirh
 > > platform.l > >c@ > > Without this Dell could try OpenVMS on an Itanium Poweredege< > > but there would be no certainty that OpenVMS was running > > correctly. > > @ > > The best example I can give you is Solaris x86. Sun suppliesA > > and supports it on our own x86 boxes and but we also supply ae> > > Hardware Certification Test Suite to 3rd party x86 systems; > > vendors which they can use to self certify their systeme > > with Solaris x86.0 > >s) > > http://www.sun.com/bigadmin/hcl/hcts/u > >R0 > > Microsoft have similar programs for Windows. > > ) > > Dell would need the same for OpenVMS.n > >S >OC > This actually not a bad idea, the objective would be to sell more  VMSa > licenses.n  F And this are precisely the reasons why HP won't do it. Winkler et. al. won't permit it to happen.   ------------------------------  % Date: Tue, 26 Aug 2003 01:19:59 +0800 , From: Paul Repacholi <prep@prep.synonet.com>: Subject: Re: Will OpenVMS I64 run on a Dell PowerEdge 3250- Message-ID: <87n0dx8xj4.fsf@prep.synonet.com>   ( jlsue <jefflsxxxz@sbcglobal.net> writes:  = > I suspect the difference might be that the customers would,cE > ultimately, expect HP to generate patches/fixes for any issues thatcB > crop on those 3rd-party systems.  I've seen similar expectationsE > with other 3rd-party hardware connected to current OpenVMS systems.c  D > Note, too, that if they were adding said systems into a VMSclusterD > with (or even without) HP systems, this could get very tricky, and > very complicated to qualify.  B But why? They are all industry *STANDARD* remember? Isn't that why the whole mess is happening?  = Or is itanic the standard you have when you are non-standard.t   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.i@                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  + Date: Mon, 25 Aug 2003 19:57:23 +0200 (MET)r9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>s: Subject: Re: Will OpenVMS I64 run on a Dell PowerEdge 3250; Message-ID: <01KZW2SD88M4AMSOR1@sysdev.deutsche-boerse.com>   ? > > I suspect the difference might be that the customers would,mG > > ultimately, expect HP to generate patches/fixes for any issues thatoD > > crop on those 3rd-party systems.  I've seen similar expectationsG > > with other 3rd-party hardware connected to current OpenVMS systems.a > F > > Note, too, that if they were adding said systems into a VMSclusterF > > with (or even without) HP systems, this could get very tricky, and  > > very complicated to qualify. > D > But why? They are all industry *STANDARD* remember? Isn't that why > the whole mess is happening? > ? > Or is itanic the standard you have when you are non-standard.w  H Even in other areas where the term "industry-standard" is used, I don't C think that this is a substitute for a vendor officially supporting ) something.    E Presumably, being "industry-standard" implies that volume will drive nA prices down.  Whether this happens or not, of course, is another aE question, but I've never heard anyone claim (and wouldn't expect to)  G that VMS will run on anything which has an "Itanium inside" sticker on nI it or whatever---well, run, perhaps, but certainly with no guarantee and   no support.E   ------------------------------  # Date: Mon, 25 Aug 2003 16:03:05 GMTa0 From: Randy Park <rjpark@mindspring.nospaam.com>) Subject: Re: XML Parser for VAX (OpenVMS) 8 Message-ID: <vnckkvkrqahf4pto2br27ldiv72jvb4dav@4ax.com>  2 On Sun, 24 Aug 2003 21:26:10 GMT, "Craig A. Berry"& <craigberry@mac.com.spamfooler> wrote:  9 >In article <mm9ikvknijs0d3pm2m7bv9fd0keggn1ggl@4ax.com>,c3 > Randy Park <rjpark@mindspring.nospaam.com> wrote:  >l5 >> Does anyone know of any XML parsers, callable fromo5 >> a VMS application, that are available for both Vaxc3 >> and Alpha systems.  I have C and Basic compilerso9 >> currently available.  I have found numerous referencesn2 >> to parsers for OpenVMS Alpha using C++ or Java, >> but nothing for Vax.n >t6 >There's a non-validating parser in C called expat at @ ><http://sourceforge.net/projects/expat>.  The Perl XML::Parser E >extension is based on expat.  There's also a validating parser in C i@ >called libxml2 that's part of the Gnome project.  More info at  ><http://www.xmlsoft.org/>.  . >oE >Both of these parsers have been known to work on VMS.  There may be nG >additional porting required for VAX support (or maybe not -- I simply s  >haven't tried to build in VAX).  # Thank you.  I will give them a try.s   ------------------------------   End of INFO-VAX 2003.470 ************************