1 INFO-VAX	Mon, 16 Jun 2003	Volume 2003 : Issue 332       Contents: Re: 400 Mile Clustering??  Re: Chuckle  Re: Clustering Re: Clustering Re: Clustering Re: Clustering Re: Clustering2 Compaq Analyze incantation to analyse clue$fru.sys cxx performance  Re: cxx performance , Re: Do's and don'ts about cleaning keyboards, RE: Do's and don'ts about cleaning keyboards, RE: Do's and don'ts about cleaning keyboards/ Re: IBM won't buy VMS ... has NIH syndrome too! - Re: Knowledge of VMS tests for job candidates  manually setting the ProcessID" Re: manually setting the ProcessID" Re: manually setting the ProcessID" Re: manually setting the ProcessID" Re: manually setting the ProcessID" NFS Client to MS-Services for Unix Sun to be the next DEC! * Two day OpenVMS seminar in Ottawa, Ontario7 Re: VMS to Linux via NFS: remove of directory fails ... 1 WHY I WANT TO: Re: manually setting the ProcessID 5 Re: WHY I WANT TO: Re: manually setting the ProcessID / You say you want apps for VMS?  Here they come!   F ----------------------------------------------------------------------  % Date: Mon, 16 Jun 2003 13:05:43 +0100 ( From: Nic Clews <sendspamhere@127.0.0.1>" Subject: Re: 400 Mile Clustering??) Message-ID: <3EEDB297.DA1AEABE@127.0.0.1>    Gerald Marsh wrote:  > H > We are currently running a two-site FDDI cluster over approx 5 miles -H > An Alphaserver 8200 and DS20 on either end with HSZ50's and host-based% > volume shadowing. (Based in the UK)  > G > The cluster runs business critical applications but only from one the F > 8200's at any one time, although we do swap the live node frequentlyC > for testing. All nodes have one vote except the usually live node H > which has two. This would allow the cluster to continue without manual. > intervention if the "stand-by" site is lost. > G > We would like to extend the distance somewhat! The normal route would G > be 200 miles but, if we needed to use a contingency network route, it  > could be 400 miles.   D John Atoz at London talked about using clusters over 1000's of milesE over FibreChannel over satellite links. Effectively it is the latency G you have to put up with. I think this is more at the testing as opposed $ to being really delivered now stage.  H Current quote is 800 km, 500 miles for clustering 96 nodes but I believe= it is more down to what you can have from your network links.   E Your voting strategy sounds, um, interesting. Really you ought to see E voting as a protection system as opposed to one of denoting where the B services comes from, that is the job of quorum, which is something governed by circumstance.    --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------  # Date: Mon, 16 Jun 2003 17:25:21 GMT ' From: Don Sykes <anonymous@pacbell.net>  Subject: Re: Chuckle+ Message-ID: <3EEDFD7E.DB0BCAB8@pacbell.net>    JF Mezei wrote:  >  > Don Sykes wrote:@ > > > http://seattlepi.nwsource.com/horsey/viewbydate.asp?id=855 > > 5 > > I wouldn't be too supprised if this happens soon. 2 > > And no one would be happier to see it than me! > > Watch out Carly. > L > Imagine if HP's CEO and board were outsourced to Digital India :-) PerhapsP > then we'd see tons of VMS advertising, and the revival of ALL-IN-1 and all theH > other products that are now in maintenance mode by indian contractors.  D Too much to hope for. I'd settle for just a CEO outsource even if itE didn't mean a VMS revival, just to throw a serious scare into the CEO B pool. Stockholders need to realize you don't have to spend tens of/ millions of dollars a year for a competent CEO!    --     Have VMS, Will Travel  Wire paladin, San Francisco    (paladinATalphaseDOTcom)   ------------------------------  % Date: Mon, 16 Jun 2003 08:24:16 +0200 < From: "Martin Vorlaender" <martin.vorlaender@pdv-systeme.de> Subject: Re: Clustering 4 Message-ID: <bcjnrk$jm6b1$1@ID-56200.news.dfncis.de>   Andrew Robert wrote: > Insomnee wrote: < >> How do I tell which route my SCS traffic is taking? [...] >  > Try using SDA to do this.  > A > To do so, sign on as SYSTEM or a similar privileged account and  >  > $ ANAL/SYSTEM  > > SHOW NETWORK   It's SHOW LAN, really.  E > It will display several pages of information you will have to thumb   > through but the info is there.   Yep:   ... >                  -- EIA Device Summary 16-JUN-2003 06:18:47 --   LSB address  = 839A2000 - Device state = 00000103 RUN,INITED,RING_AVAIL   9 UCB      UCB Addr  Fmt   Value           Client     State < ---      --------  ---   -----           ------  ----------- EIA0     81584E80 J EIA2     8158A7C0  Eth   60-07           SCA     0017 STRTN,LEN,UNIQ,STRTD ...   B What's confusing me: Why doesn't this same info doesn't show up on   $ RUN SYS$SYSTEM:LANCP LANCP> SHOW CONFIG EIA0 /USERS  K All the other protocols are there (MOP, DNA, IP, ARP), but SCA is missing?!    cu,    Martin --  F   OpenVMS:                | Martin Vorlaender  |  VMS & WNT programmer3    The operating system   | work: mv@pdv-systeme.de F    God runs the           |   http://www.pdv-systeme.de/users/martinv/:    earth simulation on.   | home: martin@radiogaga.harz.de   ------------------------------  % Date: Mon, 16 Jun 2003 10:41:38 +0200 $ From: "Peter Flunger" <p-i-b@gmx.at> Subject: Re: Clustering 0 Message-ID: <bcjvs3$2sf$1@newsreader1.netway.at>  . "Insomnee" <robert.heyes@btinternet.com> wroteK > How do I tell which route my SCS traffic is taking? We have it configured  soI > that we have redundancy at the suggestion of our engineer, but he didnt  knowK > how to tell which path it is taking so we could test it. I would tell you H > how we have it configured at present, but I cant remember from memory.  K If you want to have an overview about which local interfaces have a path to < which remote interfaces, some counters a.s.o., you can use :	 $ANAL/SYS  SDA> show portC ( will show you one or more ports, one of the ports having Type=pe,  Device=PEA0 .   Note the PDT Address of that specific port )5 SDA> show port/address=<address-of-pea0-port>/vc/chan   8 will give you a few pages of rather usefull information.  
 HtH, Peter   ------------------------------  % Date: Mon, 16 Jun 2003 20:20:32 +1000  From: "Pip" <pip@ti.nl0.com> Subject: Re: Clustering < Message-ID: <3eed9a0a$0$24424$afc38c87@news.optusnet.com.au>  J If you're running v7.3, the SCACP utility makes having to sift through SDA! for SCA info a thing of the past. F SHOW CHANNEL shows all the channel states between the local and remoteK nodes, as well as error counts, priorities and most recent open/close time. K SHOW LAN also shows useful information, but more importantly SET LAN, START E LAN and STOP LAN allows you set the priority of a path, and makes the * LAVC$STOP_BUS and LAVC$START_BUS obsolete.   Pip.  9 "Insomnee" <robert.heyes@btinternet.com> wrote in message * news:bcidjf$2m3$1@sparta.btinternet.com...K > How do I tell which route my SCS traffic is taking? We have it configured  soI > that we have redundancy at the suggestion of our engineer, but he didnt  knowK > how to tell which path it is taking so we could test it. I would tell you H > how we have it configured at present, but I cant remember from memory. >  > Ta :o) >  >    ------------------------------  # Date: Mon, 16 Jun 2003 11:57:03 GMT - From: "labadie" <tonari_no_tottoro@127.0.0.1>  Subject: Re: Clustering 2 Message-ID: <jgiHa.2588$ch7.1817@news.cpqcorp.net>  ' "Pip" <pip@ti.nl0.com> wrote in message 6 news:3eed9a0a$0$24424$afc38c87@news.optusnet.com.au...L > If you're running v7.3, the SCACP utility makes having to sift through SDA# > for SCA info a thing of the past. H > SHOW CHANNEL shows all the channel states between the local and remoteG > nodes, as well as error counts, priorities and most recent open/close  time. G > SHOW LAN also shows useful information, but more importantly SET LAN,  START G > LAN and STOP LAN allows you set the priority of a path, and makes the , > LAVC$STOP_BUS and LAVC$START_BUS obsolete. >  > Pip.   Hello   J And Scacp lets you play with priorities, suppose you have 60-07 on EWA and FWA, you can do  mc scacp set lan_device FWA/prio=3  set lan_device EWA/prio=2    somewhere in the boot phase.   Regards    Grard   ------------------------------  % Date: Mon, 16 Jun 2003 12:36:24 +0100 ( From: Nic Clews <sendspamhere@127.0.0.1> Subject: Re: Clustering ) Message-ID: <3EEDABB8.307A5023@127.0.0.1>    Insomnee wrote:  > N > How do I tell which route my SCS traffic is taking? We have it configured soN > that we have redundancy at the suggestion of our engineer, but he didnt knowK > how to tell which path it is taking so we could test it. I would tell you H > how we have it configured at present, but I cant remember from memory.  @ You've had the good 7.3 information pointers in a previous post.  G SDA is one of the places to look, but does need some interpretation and C understanding. SHOW PORT is the basic starting point, and also SHOW D CLUSTER/CONT and using the ADD FIELD in conjunction with reading the4 help as to what is being displayed should also help.  B I appreciate it is not immediately obvious that paths are working,H partly because VMS fails them over transparently. There is an article inD the WIS database discussing how to determine the path being used forA cluster communications which will help navigate SDA, available to B contract customers and possibly on the web with ask openvms (Still3 going? last I checked didn't give what I expected.)   E The error log file also contains details of virtual circuit closures, H particularly in the LAN environment. The console will also give you someE details of the status of virtual circuits as the open and close. From H the early days I believe the "noise" from CNXMAN to the console has beenA reduced, unless there's a way of switching it all back on, Keith?   H The proof would be, seeming as you are at early implementation stage, ofG actually simulating failures and watching what happens to your cluster, E and that it does what you want it to do. Also a good time to document  procedures in eventualities.   --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------  % Date: Mon, 16 Jun 2003 14:30:48 +0100  From: Roy Omond <Roy@Omond.net> ; Subject: Compaq Analyze incantation to analyse clue$fru.sys 4 Message-ID: <bckgr2$ib4$1$830fa78d@news.demon.co.uk>   Gentle colleagues,  ; I'm struggling with Compaq Analyze (what a *&^&^%%^ crock), 8 to generate output from a CLUE$FRU.SYS (itself generated from:    	$ analyze/system  	SDA> clue fru   I've tried:   * 	$ ca n tra input clue$fru.sys out x.x all  ! which gives the following output:    Event:       1G Description: Configuration Event at Fri Jun 13 17:57:13 GMT+00:00 2003   from XE  File:        clue$fru.sys > ==============================================================  2 OS_Type                 2           -- OpenVMS AXP, Hardware_Arch           4           -- Alpha; CEH_Vendor_ID           3,564       -- Compaq Computer Corp = Hdwr_Sys_Type           35          -- GS40/80/160/320 Series = Logging_CPU             0           -- CPU Logging this Event  CPUs_In_Active_Set      0 : Entry_Type              110         -- Configuration Event> DSR_Msg_Num             1,967       -- Compaq AlphaServer GS802 Chip_Type               8           -- EV6 - 21264 CEH_Device              0 ! CEH_Device_ID_0        x0000 0000 ! CEH_Device_ID_1        x0000 0000 ! CEH_Device_ID_2        x0000 0000  Unique_ID_Count         1  Unique_ID_Prefix        0 / TLV_DSR_String          AlphaServer GS80 6/1224 1 TLV_Sys_Serial_Num      xxxxxxxx (removed by me!) 7 TLV_Time_as_Local       Fri, 13 Jun 2003 17:57:13 +0000   TLV_OS_Version          V7.2-1H1 TLV_Computer_Name       XE                          NOTE 9     -  CONFIGURATION ENTRY encountered in Event Log File. :     -  A Decomposed Configuration Tree Report is available9        for this event, and may be selected separately for %        display in certain user modes.       B So what's the magic incantation to get the "CONFIGURATION ENTRY" ?   Any help much appreciated,  	 Roy Omond  Blue Bubble Ltd.   ------------------------------  # Date: Mon, 16 Jun 2003 12:33:32 GMT 4 From: "Johan Nilsson" <johan.nilsson@esrange.ssc.se> Subject: cxx performance@ Message-ID: <342184c0f69157b4e419b32529e9e64a@free.teranews.com>   Hi,   G I've been struggling for some time now to increase the performance when F running the DEC (aka COMPAQ/HP) C++ compiler. I'm doing a (perhaps not3 completely fair) comparison of building CppUnit 1.8 K (http://cppunit.sourceforge.net) on a Windows XP machine and an OpenVMS 7.1 I PWS au/433 machine and I'm really wondering what's going on. Here are the  machine configurations:   ; -----------------------------------------------------------  1) Windows platform  H/W: Intel PIII 550MHz Memory: 512 MB OS: Windows XP SP1' Compiler: Microsoft Visual C++.NET 2003    2) OpenVMS platform ( H/W: Digital Personal Workstation 433 au Memory: 256 MB OS: OpenVMS Alpha 7.1-1H1  Compiler: DEC C++ 6.5   E And the compilation times (full rebuild from scratch including link): : ----------------------------------------------------------  J 1) Windows machine: 01:32 mins (static library, precompiled headers turned _off_, debug build) B 2) OpenVMS machine: 07:35 mins (!) (object library, local template instantiations, /debug, /noopt)   J I simply can't figure out why using the DEC C++ compiler takes such a longJ time. Please note that it _is_ the compiler that's taking its time and notJ the linker (the linking phase runs pretty fast by comparison). Here's what( I've tried to do on the OpenVMS machine:  A - Install more physical memory (originally was 128 MB, now 256MB) J - (Using AUTHORIZE) Increased the wsdef, wsquo and wsext to ~64MB (same as7 WSMAX) for the user that used for software development. J - Invoking the compiler only once for compiling all source files (e.g. cxxL /bla File1.cpp,File2.cpp,File3.cpp ...) instead of one invocation per source file.  - Defragmented the disk(s).   I I've also been trying to use the MONITOR utility to find out if there are J any performance bottlenecks in the system (I'm admittedly not an expert in? this area). I've tried to monitor using the following commands:   H - MONITOR/PAGE    => Well over 99.9% of all page faults during build was from memory, and not from diskK - MONITOR/FILE_SYSTEM_CACHE => Continuously at ~100% hit rate for directory  and file hdr caches   K Am I having unrealistic expectations here? I would expect the builds on the I VMS system to be slightly slower than on the XP machine - but not to this I extent (considering that none of the two systems are exactly new). As for L the difference in physical memory available I've checked the memory usage ofK "CL.EXE" (the compiler) on the Windows machine and it rarely uses more than I 10-12MB of memory, and increasing the working set parameters for the OVMS J account from ~32MB to ~64MB resulted in no noticable performance increase.  G Perhaps I'm looking at the wrong parameters - could anyone give me some   hints on where to start looking.     Thanks for any inputs // Johan  K (By the way - why is there no precompiled header option for the OpenVMS cxx ' compiler - the Tru64 version has it ? )    ------------------------------  % Date: Mon, 16 Jun 2003 14:28:28 +0100 ( From: Nic Clews <sendspamhere@127.0.0.1> Subject: Re: cxx performance) Message-ID: <3EEDC5FC.C324019D@127.0.0.1>    Johan Nilsson wrote:  I > I've been struggling for some time now to increase the performance when H > running the DEC (aka COMPAQ/HP) C++ compiler. I'm doing a (perhaps not5 > completely fair) comparison of building CppUnit 1.8 M > (http://cppunit.sourceforge.net) on a Windows XP machine and an OpenVMS 7.1 K > PWS au/433 machine and I'm really wondering what's going on. Here are the  > machine configurations:  > G > And the compilation times (full rebuild from scratch including link): < > ---------------------------------------------------------- > L > 1) Windows machine: 01:32 mins (static library, precompiled headers turned > _off_, debug build) D > 2) OpenVMS machine: 07:35 mins (!) (object library, local template! > instantiations, /debug, /noopt)     C Why is this important? No, I don't think you're being fair. You are G comparing a 64 bit architecture RISC compiler with a 32 bit, um, thingy G compiler. Surely compilation is a one-off process, the real performance G you're looking for is in the execution, and that could be anything from A program design (which also affects compilation times) to external H factors such as network latencies and device reading and writing and the@ use of caches as to "how fast it goes". I'm surprised you've not* compared size of the delivered executable.  E I expect that Itanium compilations will take even longer, because the 8 compiler is going to spend time looking for parallelism.  F It is possible for someone experienced in performance to tune a systemD as a compilation engine. If you want some parity, I'd switch off theE default write caching in XP. Some folks I know got a huge performance G boost in compilations using the XFC features of 7.3 of VMS over 7.1, so H I don't believe you're really on a level playing field here. When was XPE released? And 7.1? This is set as a task for you and hopefully you'll F find some reasons to consider 7.3-1. But no matter what you do to yourG apple or your orange, they'll still taste different. Then again, one of & them does bruise a little more easily.   --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------  % Date: Mon, 16 Jun 2003 12:24:54 +0100 ( From: Nic Clews <sendspamhere@127.0.0.1>5 Subject: Re: Do's and don'ts about cleaning keyboards ) Message-ID: <3EEDA906.52ECC327@127.0.0.1>    Paul Sture wrote:  > I > My heartful thanks to the person who recommended putting my LK keyboard  > through the dishwasher.  >  > It is beatifully clean now.  > ( > But a pity it doesn't work any more...  H After 1 or two episodes of cleaning out keyboards I've talked about hereH before, you can also dismantle the keyboard for a proper electrical fix.B It is a matter of removing all the keys gently and opening all theB catches under the keyboard and between the keys, then the case andF membranes can be easily separated and cleaned, but don't rub too hard.C Make sure it is absolutely bone dry before reassembling, and having > another keyboard to look at while putting the keys back helps.  G If the kb still refuses to work, it could be track or other damage that G is beyond repair. Track type damage often affects a row or two, or line  of keys.   --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------  % Date: Mon, 16 Jun 2003 17:19:44 +0100 * From: Andrew Robinson <arobinson@hspg.com>5 Subject: RE: Do's and don'ts about cleaning keyboards O Message-ID: <CDA4BAD1E10ED41181AC00508B6051D3023CD015@grumpy.internal.hspg.com>   > >From: p_sture@elias.decus.ch [mailto:p_sture@elias.decus.ch] H >My heartful thanks to the person who recommended putting my LK keyboard through the dishwasher.? >  >It is beatifully clean now. > ( >But a pity it doesn't work any more...  > I Pity you didn't remove the electrics first - less than two minutes with a K flat bladed screwdriver. If you want a spare I have between 50 - 60 VT420's K & 520's + keyboards cluttering up pallet spaces in the warehouse. MD though L they were Russian TV's & told us to 'upgrade' them to PC's running emulation> :-( So now have lots of spares for the workshops, or Hobbyists   Regards    Andrew R   ------------------------------    Date: 16 Jun 2003 12:19:58 -0500B From: clubley@remove_me.eisner.decus.org-Earth.UFP (Simon Clubley)5 Subject: RE: Do's and don'ts about cleaning keyboards 3 Message-ID: <ebti8c8GrfXf@eisner.encompasserve.org>   | In article <CDA4BAD1E10ED41181AC00508B6051D3023CD015@grumpy.internal.hspg.com>, Andrew Robinson <arobinson@hspg.com> writes:K > Pity you didn't remove the electrics first - less than two minutes with a M > flat bladed screwdriver. If you want a spare I have between 50 - 60 VT420's M > & 520's + keyboards cluttering up pallet spaces in the warehouse. MD though N > they were Russian TV's & told us to 'upgrade' them to PC's running emulation@ > :-( So now have lots of spares for the workshops, or Hobbyists >   L In case anybody wants to take you up on that offer (not me BTW, I've alreadyK got enough hobbyist equipment), on which part of the planet are you located  and how much are you asking ?    Simon.   --  B Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP       L VMS advocate: One who makes a Mac advocate look like a beginner at advocacy.   ------------------------------    Date: 16 Jun 2003 07:37:23 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) 8 Subject: Re: IBM won't buy VMS ... has NIH syndrome too!3 Message-ID: <ANOXJKC+64q1@eisner.encompasserve.org>   V In article <3EEB8CD0.9F5B3452@istop.com>, JF Mezei <jfmezei.spamnot@istop.com> writes:M > When you management comes to you and tell you that they have decided to run O > SAP or some other application that is no longer available on VMS, what do you 
 > tell them ?   H    I tell them stay away from SAP.  They don't always listen, but then I$    don't have to fix thier mistakes.   ------------------------------   Date: 16 Jun 03 18:23:30 +0200) From: p_sture@elias.decus.ch (Paul Sture) 6 Subject: Re: Knowledge of VMS tests for job candidates) Message-ID: <zHMtihp0Hk4S@elias.decus.ch>   V In article <3EEA116D.D11B54AA@istop.com>, JF Mezei <jfmezei.spamnot@istop.com> writes: > Jeff Cameron wrote: O >> My Web pages on VMS has a System Manager's Quiz. It was intended for fun but H >> can separate the novices from the real VMS Admins. It is located at :' >> http://www.jcameron.com/vms/quiz.htm  > O > I started to answer that quiz. I wouldn't think it was a system manager quiz, L > but rather a VMS engineer quiz. Also, I found it quite hard to explain theO > difference between a logical name and a symbol since they are so different on G > all aspects, one could write a novel to describe the differences :-)   >   K I thought that was a useful question, as I recall beginners getting the two G confused. I tend to agree with your assessment that some questions were K more applicable for a VMS engineer than a system manager. Everyone probably " has their own bias here of course.   --  
 Paul Sture   ------------------------------  % Date: Mon, 16 Jun 2003 11:30:11 +0200 3 From: "Tim Aisthorpe-Buckley" <Tim_A_B@Hotmail.com> ' Subject: manually setting the ProcessID / Message-ID: <ver4iuqt8jg20e@corp.supernews.com>   G Does anybody know if it is possible to manually set the process id of a  process when it is created?    Thanks   Tim    ------------------------------  + Date: Mon, 16 Jun 2003 12:40:01 +0100 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> + Subject: Re: manually setting the ProcessID ; Message-ID: <01KX5V2TI4U4AOKN0V@sysdev.deutsche-boerse.com>   I > Does anybody know if it is possible to manually set the process id of a  > process when it is created?   : I don't know.  However, what are you trying to accomplish?  F One can, of course, set the NAME of a process.  Is there a case where , this won't help you but setting the ID will?  I This reminds me of the story that an early VAX was up so long that there  F was danger than soon a process with the PID 00000000 would be created F (i.e. the number would wrap around); DEC delivered a patch in time to  prevent the problem.   Anyone know details of this?   ------------------------------  # Date: Mon, 16 Jun 2003 11:14:36 GMT , From: "Ed Dennison" <ed.dennison@compaq.com>+ Subject: Re: manually setting the ProcessID 1 Message-ID: <wEhHa.2581$oA4.652@news.cpqcorp.net>    Tim,  E     Only if you don't mind the system (possibly) crashing. The pid is J assigned by VMS at process creation. The high order digits relate to whichH node you are on in the cluster (or zero if it is not clustered). The lowB order digits are an index into a table - ('the PCB vector array').  K     So to manually set the pid you would need a privileged program (CMKRNL) G which altered you position in the PCB vector array - which would not be J 'too' difficult. But setting the upper digits might pose more of a problem if the systems were clustered.  -     Might be best to say - No, can't be done.            Ed Dennison > "Tim Aisthorpe-Buckley" <Tim_A_B@Hotmail.com> wrote in message) news:ver4iuqt8jg20e@corp.supernews.com... I > Does anybody know if it is possible to manually set the process id of a  > process when it is created?  >  > Thanks >  > Tim  >  >    ------------------------------  % Date: Mon, 16 Jun 2003 12:00:19 +0100  From: Dave Brennan <> + Subject: Re: manually setting the ProcessID 8 Message-ID: <ne8rev0iqc7sknatlt38gmrc6t1adi185g@4ax.com>  : Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com> wrote:  J >> Does anybody know if it is possible to manually set the process id of a >> process when it is created?   > ; >I don't know.  However, what are you trying to accomplish?  > G >One can, of course, set the NAME of a process.  Is there a case where  - >this won't help you but setting the ID will?  > J >This reminds me of the story that an early VAX was up so long that there G >was danger than soon a process with the PID 00000000 would be created  G >(i.e. the number would wrap around); DEC delivered a patch in time to   >prevent the problem.  >  >Anyone know details of this?   F I think the process ID is a hash of the process slot, cluster node andE a counter that increments on reuse. I think it was this reuse counter  that could overflow.  F The real question is why you want to change it, as givenn the name you can map the process id.    ------------------------------  % Date: Mon, 16 Jun 2003 12:48:17 +0100 ( From: Nic Clews <sendspamhere@127.0.0.1>+ Subject: Re: manually setting the ProcessID ) Message-ID: <3EEDAE81.D86568A0@127.0.0.1>    Tim Aisthorpe-Buckley wrote: > I > Does anybody know if it is possible to manually set the process id of a  > process when it is created?   A It has already been mentioned you can change the process name, at F command level or programmatically. Within UIC groups, the process name/ can be used in many of the same ways a PID can.   > I'd advise you to read "internals and datastructures" books toH understand the relevance and importance of the PID and how it is used inA many aspects of the life of a process on a system, and if you did + achieve it, what the consequences would be.   A You don't allude to why you want to do this, there are techniques G available which will allow this "uncontrolled" information being useful F from a single interface into something else. I'm speculating a processH monitoring or control function based on the PID where a common referenceG point such as a logical name can be used to similar effect as a "fixed"  PID.  F PID numbering and reuse of is a fascinating topic, the PID is a numberF generated from the process index as seen in the SDA, with implicationsF for general process management and control on a system, not to mention< system parameter settings. The PID is a cluster wide entity. --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------  % Date: Mon, 16 Jun 2003 10:08:00 -0600 6 From: "Michael D. Ober" <obermd-@-alum-mit-edu-nospam>+ Subject: NFS Client to MS-Services for Unix 0 Message-ID: <AXlHa.19$7H5.32303@news.uswest.net>  J Anyone gotten this to work?  If so, how?  HP (DEC) Support has been unable to figure this out.    TIA,
 Mike Ober.   ------------------------------    Date: 16 Jun 2003 10:15:59 -0700( From: bob@instantwhip.com (Bob Ceculski)  Subject: Sun to be the next DEC!= Message-ID: <d7791aa1.0306160915.37f9047b@posting.google.com>   3 gee Andrew, I hope this means we willn't be hearing 3 from you anymore :) ... there is still time to join  the OpenVMS Itanium sales team!   K http://techupdate.zdnet.com/techupdate/stories/main/0,14179,2914032,00.html    ------------------------------  % Date: Mon, 16 Jun 2003 11:52:16 -0400 ) From: "Neil Rieck" <n.rieck@sympatico.ca> 3 Subject: Two day OpenVMS seminar in Ottawa, Ontario 7 Message-ID: <QIlHa.710$5d.212412@news20.bellglobal.com>   M The Canadian Association of Compaq Users, in collaboration with HP Canada and L HP Corporation, will be hosting an OpenVMS 2-day technical seminar in Ottawa= on June 25 and 26, 2003. (on the University of Ottawa Campus)   H All members, users and customers of HP technology from across Canada areF invited to attend this information-packed seminar for IT ProfessionalsK involved in HP technology platforms. Registration is now open. Please visit ; our seminar pages regularly to find out the latest updates.    http://www.canacu.org/) http://www.communitywerke.com/seminar.htm   
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada.! http://www3.sympatico.ca/n.rieck/    ------------------------------    Date: 16 Jun 2003 02:53:49 -0700- From: norbert.hanowski@dillinger.de (Norbert) @ Subject: Re: VMS to Linux via NFS: remove of directory fails ...< Message-ID: <6f378eb9.0306160153.154c2ae@posting.google.com>   Hi John   > Thank You - this was the option I searched. It works fine now.   Norbert    "John Gemignani, Jr." <jon-nope@thiswontworkossc.net> wrote in message news:<wsmGa.4865$Jw6.3263284@news1.news.adelphia.net>... N > First of all, you can export your files so that the .DIR's don't appear.  Do > this with: > 1 >     $TCPIP ADD EXPORT "/name" /OPTIONS=TYPELESS  > N > Which version of UCX/TCPIP are you using, and are you up on your ECO levels? > Do you get any messages? >  > -John  > < > "Norbert" <norbert.hanowski@dillinger.de> wrote in message9 > news:6f378eb9.0306130439.725fd1d2@posting.google.com...  > > Hi, I > > I mount a VMS-device via NFS from a Linux-server. My NT-client on the = > > Linux-server works fine with the data on the VMS-machine. E > > The only Problem: I can create a new device on the VMS machine by I > > Windows but I can't remove or rename this directory by the PC. In the J > > WNT Explorer all directories on the VMS-device have the extension .dir > > .  > > What is wrong? > > 
 > > Thank you  > >  > >  > > Norbert    ------------------------------  % Date: Mon, 16 Jun 2003 14:07:48 +0200 3 From: "Tim Aisthorpe-Buckley" <Tim_A_B@Hotmail.com> : Subject: WHY I WANT TO: Re: manually setting the ProcessID/ Message-ID: <vercp6i159do7d@corp.supernews.com>    Thanks for all the suggestions,   H The reason why I want to do this is that we have a VMS box that has someJ automated software on it. So nobody really needs to touch the VMS server -+ and nobody knows how to on the client site.   H The trouble is that there are several different versions of the softwareI that can be run - each does something different - yet only one can run at L once. So it is neccessary to be able to switch programs without going to theK server room. We tried it with telnet, but that doesn't work. I then wrote a L java socket, put that on the server and that doesn't work. The reason why itK doesn't work is because the start and stop scripts that were written for us K use the PID of the invoking program to find out what needs to be started or K stopped - therefore the only way to stop something you have started is from J within the same terminal on the server as they were started, or by closing3 all of the GUI applications on the server manually.   I So if I want to use a Java socket (which is nice because the client could K then use a client socket on their desktop PCs to stop and start as required J without going to the server room) then I need to be able to set the PID ofH the Java Socket, so that if the socket goes down, I can then start it upH again later with the same PID. I do not know if this will achieve what IJ want to, as the script is third party and I'm not a VMS person, so I don'tL really understand what it is doing with the PID of the invoking process. But I think it is worth a try.   Regards    Tim   > "Tim Aisthorpe-Buckley" <Tim_A_B@Hotmail.com> wrote in message) news:ver4iuqt8jg20e@corp.supernews.com... I > Does anybody know if it is possible to manually set the process id of a  > process when it is created?J >2 > Thanks >0 > Timu >3 >    ------------------------------  % Date: Mon, 16 Jun 2003 13:25:48 +0100e+ From: John Laird <john@laird-towers.org.uk> > Subject: Re: WHY I WANT TO: Re: manually setting the ProcessID8 Message-ID: <o8drev41bnubijld5rj4jj7l2s4da5jm15@4ax.com>  ; On Mon, 16 Jun 2003 14:07:48 +0200, "Tim Aisthorpe-Buckley"i <Tim_A_B@Hotmail.com> wrote:    >Thanks for all the suggestions, >oI >The reason why I want to do this is that we have a VMS box that has somecK >automated software on it. So nobody really needs to touch the VMS server -D, >and nobody knows how to on the client site. >yI >The trouble is that there are several different versions of the softwareSJ >that can be run - each does something different - yet only one can run atM >once. So it is neccessary to be able to switch programs without going to theAL >server room. We tried it with telnet, but that doesn't work. I then wrote aM >java socket, put that on the server and that doesn't work. The reason why it-L >doesn't work is because the start and stop scripts that were written for usL >use the PID of the invoking program to find out what needs to be started orL >stopped - therefore the only way to stop something you have started is fromK >within the same terminal on the server as they were started, or by closinge4 >all of the GUI applications on the server manually.  L I would heartily recommend you get your "start and stop scripts" fixed.  TheL PID is irrelevant to any form of start request or process creation, being anI output from the event, and if the stop request is nothing more than a DCLoK STOP command, then you can simply switch from the (presumed) STOP/ID= style K to STOP "process name" if the process is suitably named and the scripts areiL running in the same group.  Giving a new process a unique name which you can: then refer to, in order to stop it, is not rocket science.  J >So if I want to use a Java socket (which is nice because the client couldL >then use a client socket on their desktop PCs to stop and start as requiredK >without going to the server room) then I need to be able to set the PID ofwI >the Java Socket, so that if the socket goes down, I can then start it upeI >again later with the same PID. I do not know if this will achieve what IsK >want to, as the script is third party and I'm not a VMS person, so I don't M >really understand what it is doing with the PID of the invoking process. Buti >I think it is worth a try.o  J To be honest, learning enough to change the controlling procedures to makeD them work far more robustly is going to be childs play compared withL attempting to create processes with a pid of your choice (VMS doesn't have a@ kernel you can dicker about with in this area, praise the lord).     	John   J PS  You wouldn't be related to a Peter A-B who lives in the New Forest, by any chance ?   ------------------------------    Date: 16 Jun 2003 06:29:41 -0700( From: bob@instantwhip.com (Bob Ceculski)8 Subject: You say you want apps for VMS?  Here they come!= Message-ID: <d7791aa1.0306160529.4c037deb@posting.google.com>   ( they're on the way says the inquirer ...  ) http://www.theinquirer.net/?article=10022g   ------------------------------   End of INFO-VAX 2003.332 ************************