1 INFO-VAX	Tue, 19 Apr 2005	Volume 2005 : Issue 217       Contents:> Re: changing node name: RENAME/IDENTIFIER SYS$NODE_oldnodename> Re: changing node name: RENAME/IDENTIFIER SYS$NODE_oldnodename> Re: changing node name: RENAME/IDENTIFIER SYS$NODE_oldnodename Re: Compiler bug check Re: Compiler bug check Re: Compiler bug check Re: Compiler bug check DCPS B&W to a color LaserJet DECwindows: pointer policy Re: DECwindows: pointer policy Re: DECwindows: pointer policy Re: DECwindows: pointer policy Re: FA: VMS SW kits.! Re: indicator lights on SBB disks  Re: Login mystery  Re: Login mystery  Re: Login mystery  Re: Login mystery  Re: Login mystery  Re: Login mystery   Re: two questions about MINICOPYP Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot dataP Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot dataP Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot dataP Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot data1 Re: What do you expect from DIR/SIZ=(ALLO,USED) ?   F ----------------------------------------------------------------------  + Date: Mon, 18 Apr 2005 20:54:46 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)G Subject: Re: changing node name: RENAME/IDENTIFIER SYS$NODE_oldnodename $ Message-ID: <d416ql$4kh$3@online.de>  F In article <0_P8e.4048$LI.1508@news.cpqcorp.net>, hoff@hp.nospam (Hoff Hoffman) writes:    J > :I'm currently carefully planning how to change the node name.  The FAQ H > :is a starting point, but it doesn't cover everything, at least in my I > :case.  I'll post a summary here later today, and again after I change   > :the node name.   ; I'll still post a summary, but I need a bit more time.  :-|   K > :My situation is a bit different than that implicitly assumed in the FAQ. G > :In my cluster each node has its on system disk.  I want to add a new H > :node, with its own system disk, and use the system disk of one of the% > :current nodes as a starting point.  > D >   Identifiers are cluster-wide, and are shared across the cluster.D >   This is why the SYSUAF file must be shared or must be carefully A >   coordinated across the cluster, and why RIGHTSLIST is shared.   H SYSUAF (and the other files mentioned in SYLOGICALS.TEMPLATE) are all onB a non-system disk, mounted and used by all nodes, with appropriateD logicals defined (indirectly) in SYLOGICALS.TEMPLATE.  (Actually, inE SYLOGICALS.COM, if the machine is a cluster member, it tries to mount E this disk and, if successful, executes a procedure on that disk which I does the definitions.)  It's been this way for years.  However, I am not  F 100% sure that the common SYSUAF was "cleanly" constructed initially, " but I've not noticed any problems.  C >   Your situation does not differ from that assumed by the FAQ, in  >   other words. > < > :   o  Use the AUTHORIZE utility command RENAME/IDENTIFIER6 > :      to rename the SYS$NODE_oldnodename rightslist7 > :      identifier to match the new node name. (Do not 8 > :      change the binary value of this identifier, and' > :      do not delete the identifier.)  > : * > :Doesn't apply to my situation at all??? > J >   It applies often enough, in my experience, to warrant the mention that. >   it gets in the sequence listed in the FAQ.  G What I mean is that I now have, say, SYS$NODE_FOO.  I want to add a new A node, BAR.  To do this (mainly to avoid having to install all the D layered products), I want to start with a copy of the system disk ofB FOO.  There is just one SYSUAF, which is on neither system disk.  G Obviously I can't RENAME SYS$NODE_FOO, since it is still needed due to  ( the presence of node FOO in the cluster.  F > :I currently have a SYS$NODE_nodename identifier in the UAF for all L > :current nodes in the cluster, and also for a no-longer-used nodename.  I F > :have no problems at all that I am aware of which could possibly be  > :related to this.  > I >   If you and everyone else have *never* used this identifier *anywhere* H >   on any object and you are certain of it, then you can most certainly >   ignore it. > G >   If the identifier did get used and you don't want to rename it, you G >   can remove the old binary value and can add the new identifier back 8 >   in using the new binary value of the new identifier.  D What is confusing me is RENAME.  I want to use the new system disk, H which is the result of a node-name change on the copy of the old system E disk, in the same cluster.  Obviously I can't RENAME anything in the  H (common) SYSUAF since the old node is still there.  I don't really want F to change the node name of a particular machine, but rather to change H the node name on the copy of a system disk so that I have a new node in  the same cluster.    ------------------------------  # Date: Mon, 18 Apr 2005 21:52:42 GMT # From: hoff@hp.nospam (Hoff Hoffman) G Subject: Re: changing node name: RENAME/IDENTIFIER SYS$NODE_oldnodename 1 Message-ID: <K_V8e.4096$BU.1984@news.cpqcorp.net>   w In article <d416ql$4kh$3@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes: H :What I mean is that I now have, say, SYS$NODE_FOO.  I want to add a new
 :node, BAR.     F   I will assume the new node is licensed for all products found on the   existing node.  F   There is extra effort involved in the parallel management of OpenVMSF   software installations; by running parallel system disks, of course.D   Officially, you'll want to perform a clean installation of OpenVMS   and of the layered products.  H   To (try to) replicate, add SYS$NODE_BAR to the shared RIGHTSLIST, thenF   shut down FOO::, BACKUP/IMAGE the system disk to another node, mountG   the target disk, disable the network startups for TCP/IP services and H   for DECnet Phase IV (and I'd blow away the network database files) andE   such, set NISCS_LOAD_PEA0 and CLUSTER parameters for non-clustered  G   operations, set SCSNODE to BAR, copy over clones of the shared SYSUAF I   and RIGHTSLIST files for the initial node operations, boot the disk on  I   the target host system, and then re-configure the host network stacks,  G   get rid of the temporary SYSUAF and RIGHTSLIST files, grab a copy of  F   the and CLUSTER_AUTHORIZE.DAT file, set NISCS_LOAD_PEA0 and CLUSTER C   system parameters for clustered operations, shut down and reboot.   *   I have probably missed a few steps here.  B   And do please note that the above is *NOT* officially supported.  G   I don't generally recommend cloning disks -- it usually works, but it H   is not supported.  It is often easier to go to a common system disk --H   which would be my preference, as managing multiple parallel individualH   system disks is, well, work -- or to perform the work and to reinstallI   the products on the new (parallel) system disk.  (Cloning system disks  H   means that the products are replicated as well as the existing cruft.)  I   Having the installation kits available, and having the COMMON area for  G   the core files makes managing fewer disks a whole lot more convenient F   and easy than managing and upgrading multiple parallel system disks.I   (The parallel disks also tends to mean that upgrades are performed less K   frequently, as multiple individual system disk upgrades are more involved A   and more effort than fewer system disks and a rolling upgrade.)   F   Now if there are no shared interconnects or such, well, I definitely   feel your pain.  :-)    N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com    ------------------------------    Date: 18 Apr 2005 18:23:57 -0700$ From: "AEF" <spamsink2001@yahoo.com>G Subject: Re: changing node name: RENAME/IDENTIFIER SYS$NODE_oldnodename C Message-ID: <1113873837.913993.227300@f14g2000cwb.googlegroups.com>   / Phillip Helbig---remove CLOTHES to reply wrote: B > In article <0_P8e.4048$LI.1508@news.cpqcorp.net>, hoff@hp.nospam (Hoff  > Hoffman) writes: > G > > :I'm currently carefully planning how to change the node name.  The  FAQ F > > :is a starting point, but it doesn't cover everything, at least in myC > > :case.  I'll post a summary here later today, and again after I  change > > :the node name.  [...] E > What is confusing me is RENAME.  I want to use the new system disk, B > which is the result of a node-name change on the copy of the old systemF > disk, in the same cluster.  Obviously I can't RENAME anything in theD > (common) SYSUAF since the old node is still there.  I don't really wantG > to change the node name of a particular machine, but rather to change   F > the node name on the copy of a system disk so that I have a new node in > the same cluster.     : The rename operation is appropriate if you are renaming anD already-in-use node. If you had only one node, and you used it for aF while, and then later you wanted to rename it from FRED to ALICE, thenG you would need to carry out the identifier-rename operation. If you had G a cluster containing TOM and TOM has already been in use, but later you C need to change it to MARY, then also in this case you would need to  rename the node identifier.   F But you are ADDING a node, in which case just leave it be, even thoughC you change a node name in the process. You are in effect creating a E duplicate node name and then renaming to fix it. No identifier rename 
 is needed.  > Of course, anything on the new system that references the nodeE identifier will have to be modified to use the new identifier. If you F build your new node from scratch by installing VMS fresh on the system? disk then there is no need to worry about renaming such things.    JMHO   ------------------------------  % Date: Mon, 18 Apr 2005 13:51:45 -0700 , From: "Paul Dembry" <pade@nospam.trifox.com> Subject: Re: Compiler bug check 1 Message-ID: <bdqdndirJeDugPnfRVn-3Q@megapath.net>   H >    I've seen lots of layered products behave this way when running out >    of PGFLQUOTA.K I doubled the pgflquota, no luck. It even dumps when I specify /noobject. I  am using dec c v 5.2-003.  Thanks,  Paul   ------------------------------  % Date: Mon, 18 Apr 2005 17:15:15 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com>  Subject: Re: Compiler bug check B Message-ID: <1113858912.28bae7cec64fa19aa92e281153a3140f@teranews>   Bob Koehler wrote:@ > > %VCG-F-BUGCHECK, Compiler bug check during middle end phase.  H >    I've seen lots of layered products behave this way when running out >    of PGFLQUOTA.    G On VAX, with DEC-C, when I trried to first compile Kermit, the compiler A complained about not enough virtual memory. It didn't bugcheck. I + increases pgfilquo and then it worked fine.    ------------------------------  # Date: Mon, 18 Apr 2005 21:33:57 GMT # From: hoff@hp.nospam (Hoff Hoffman)  Subject: Re: Compiler bug check 0 Message-ID: <9JV8e.4095$BU.932@news.cpqcorp.net>  ` In article <bdqdndirJeDugPnfRVn-3Q@megapath.net>, "Paul Dembry" <pade@nospam.trifox.com> writes:I :>    I've seen lots of layered products behave this way when running out  :>    of PGFLQUOTA. L :I doubled the pgflquota, no luck. It even dumps when I specify /noobject. I :am using dec c v 5.2-003.  J   V5.2 dates back to 1995, and various errors have been identified in thatJ   version, and resolved.  I'd suggest an upgrade to a more recent version,.   both for the fixes and for the enhancements.  L   As of this writing, V6.4 is the current compiler for OpenVMS VAX, V6.5 for1   OpenVMS Alpha, and (IIRC) V7.1 for OpenVMS I64.   H   If this is not the process quotas -- and I'd not immediately assume a K   compiler bugcheck error was due to insufficient quotas -- then I'd expect I   the problem is in the processing of the source code itself.  Obviously. H   This means that shuffling the source code around or changing it might H   help, if you can identify the particular construct involved -- but theG   compiler upgrade is probably the best approach, as would be expected.   D   If folks are to pursue this, details of the source code and of the(   compiler bugcheck will be of interest.  N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com    ------------------------------  % Date: Mon, 18 Apr 2005 16:28:48 -0700 , From: "Paul Dembry" <pade@nospam.trifox.com> Subject: Re: Compiler bug check 1 Message-ID: <H8adnXiSfobf3_nfRVn-gg@megapath.net>   L >   V5.2 dates back to 1995, and various errors have been identified in thatL >   version, and resolved.  I'd suggest an upgrade to a more recent version,0 >   both for the fixes and for the enhancements. > J >   As of this writing, V6.4 is the current compiler for OpenVMS VAX, V6.5 for 3 >   OpenVMS Alpha, and (IIRC) V7.1 for OpenVMS I64. K Is v 6.4 compatible with OpenVMS 6.1? I'll also rearrange the code a bit to F see if that works. It's irritating that this has compiled in the past. Thanks,  Paul   ------------------------------  # Date: Tue, 19 Apr 2005 00:11:33 GMT + From: Jack Patteeuw <jjpatteeuw@nospam.net> % Subject: DCPS B&W to a color LaserJet B Message-ID: <V0Y8e.9774$go4.9100@newsread2.news.atl.earthlink.net>  I Is there a way to configure DCPS such that when it is converting an ANSI  I file into PostScript and sending it to a color LaserJet (5550 IRRC) that  I I can tell the printer to skip the CMY passes and therefore speed up the   overall print rate.    In other words, just print B&W.    Thanks   ------------------------------  + Date: Mon, 18 Apr 2005 21:32:03 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)# Subject: DECwindows: pointer policy $ Message-ID: <d4190j$a33$1@online.de>  I Using the traditional (as opposed to CDE) DECwindows stuff, on a VAX, is  I it possible to have the active window be the one where the mouse pointer  F currently is, as opposed to being the one which has been activated by  clicking on it?   F Which reminds me---with the other pointer policy, why does one have to@ click on the frame to raise a window?  Surely one could have theD position of the pointer determine which window is active, and click F anywhere in it when raising it, as with the other pointer policy.  In E other words, why do the two things (pointer determines active window/ G clicked-on window is active and click on frame/click anywhere to raise  * window) come only in certain combinations?   ------------------------------  % Date: Mon, 18 Apr 2005 17:57:16 -0400  From: John Doe <jdoe@doe.org> ' Subject: Re: DECwindows: pointer policy B Message-ID: <1113861428.cf9af51ffbe58973d1b8b1a0251175ad@teranews>  / Phillip Helbig---remove CLOTHES to reply wrote: J > Using the traditional (as opposed to CDE) DECwindows stuff, on a VAX, isJ > it possible to have the active window be the one where the mouse pointerG > currently is, as opposed to being the one which has been activated by  > clicking on it?   D This is a window manager issue. I know that with TWM on a MAC, I canB have that policy and it is extremely annoying. You can'ty move theH cursor off a window to let you type in peace. And as you move the cursor; around, windows rapidly change focus as you move over them.   H Similarly MOSAIC originally came with a policy that the cursor had to beD a on a field, it wasn't enough to just click on the field to give it! focus and that too was annoying.     I fixed this by adding: # Mosaic*KeyboardFocusPolicy: pointer     @ For the window manager, it is located iin DECW$MWM.DAT either inG decw$system:defaults and overriden by parameterd in DECW$USER_DEFAULTS:   M In the system one, there is some documentation on the various focus policies.     H > Which reminds me---with the other pointer policy, why does one have to( > click on the frame to raise a window?   D On my system, clicking anywhere in a window brings it to foreground.    3 And I did not remove my clothes to reply to you ;-)    ------------------------------  % Date: Mon, 18 Apr 2005 14:59:31 -0700 , From: Ken Fairfield <my.full.name@intel.com>' Subject: Re: DECwindows: pointer policy + Message-ID: <d41ak3$dpd$1@news01.intel.com>   / Phillip Helbig---remove CLOTHES to reply wrote: K > Using the traditional (as opposed to CDE) DECwindows stuff, on a VAX, is  K > it possible to have the active window be the one where the mouse pointer  H > currently is, as opposed to being the one which has been activated by  > clicking on it?   E      Certainly, I've been using "focus follows pointer" for ages. :-) E Unfortunately, I'll need to check my home system to find the resource A that needs to be defined/modified.  I'll report back later, but I A expect you'll find various settings in DECW$SMB_POINTER.DAT.  You D may also find these settings by right-clicking on the background and- checking the various menu items that pops up.   H > Which reminds me---with the other pointer policy, why does one have toB > click on the frame to raise a window?  Surely one could have theF > position of the pointer determine which window is active, and click H > anywhere in it when raising it, as with the other pointer policy.  In G > other words, why do the two things (pointer determines active window/ I > clicked-on window is active and click on frame/click anywhere to raise  , > window) come only in certain combinations?  G      You mean, make it work like MS Windows!?  Actually, I'm mixed over H which method I prefer...  A colleague of mine used to run with a settingH that focus follows pointer *and* pointer-focus-raises-the-window...after@ a short delay (about 2 secs or so).  That policy drove me batty!  D      Anyway, I think the reason you can't click on, for example, theE main DECterm window (rather than the frame) is that any text that was E selected will be de-selected when you click.  It's just a Motif thing 
 I think...  	      -Ken  --  6 I don't speak for Intel, Intel doesn't speak for me...  
 Ken Fairfield ! D1C Automation VMS System Support " who:   kenneth dot h dot fairfield where: intel dot com   ------------------------------  % Date: Mon, 18 Apr 2005 19:49:47 -0700 , From: Ken Fairfield <my.full.name@intel.com>' Subject: Re: DECwindows: pointer policy + Message-ID: <d41rkc$mbh$1@news01.intel.com>   / Phillip Helbig---remove CLOTHES to reply wrote: K > Using the traditional (as opposed to CDE) DECwindows stuff, on a VAX, is  K > it possible to have the active window be the one where the mouse pointer  H > currently is, as opposed to being the one which has been activated by  > clicking on it?   D     OK, here's the answer.  In your DECW$MWM.DAT file (create one if8 you don't already have one), set the following resource:  " 	Mwm*keyboardFocusPolicy:  pointer  % The other value for this resource is:   # 	Mwm*keyboardFocusPolicy:  explicit   * which gives the click-to-select behaviour.  D     Also, you can make this and other changes by right-click-an-holdB on the background, then move the pointer to "Options" and release.G Under Options, select Workspace, then choose the options as you desire.    > H > Which reminds me---with the other pointer policy, why does one have toB > click on the frame to raise a window?  Surely one could have theF > position of the pointer determine which window is active, and click H > anywhere in it when raising it, as with the other pointer policy.  In G > other words, why do the two things (pointer determines active window/ I > clicked-on window is active and click on frame/click anywhere to raise  , > window) come only in certain combinations?  A     There's another reousrce, also settable through the Workspace @ menu, that controls raising the window on focus.  That would be:   	Mwm*focusAutoRaise: true   ) This resource is normally set to "false".   6      BTW, how are these features controlled under CDE?   	-Ken  --  6 I don't speak for Intel, Intel doesn't speak for me...  
 Ken Fairfield ! D1C Automation VMS System Support " who:   kenneth dot h dot fairfield where: intel dot com   ------------------------------  + Date: Mon, 18 Apr 2005 20:37:32 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) Subject: Re: FA: VMS SW kits. $ Message-ID: <d415qc$4kh$1@online.de>  7 In article <3cf7j9F6miqsiU1@individual.net>, Paul Sture  <paul.sture@decus.ch> writes:   K > But back to the real topic. My fear is that blatant abuse of the current  K > scheme could endanger the Hobbyist program. As someone who tried out the  I > first CHARON-VAX Hobbyist programme, I was very disappointed to see it  K > withdrawn after someone published a hack (or was it a valid license key?   > I forget.) in the wild.   B I don't really see that as an issue.  It's not like because of theA Hobbyist programme HP is making millions selling VMS CDs.  A more H serious problem would be abuse of hobbyist licenses, i.e. using them for commercial purposes.    I I believe that it has been stated here, more or less officially, that it  F is OK to borrow media for an installation, IIRC even for a commercial ? one.  It would be silly to allow that and disallow selling CDs.   H I believe there was a hobbyist PDP programme which was cancelled due to  abuse.   ------------------------------  + Date: Mon, 18 Apr 2005 18:05:17 +0000 (UTC) , From: cdl@deeptow.ucsd.edu (Carl Lowenstein)* Subject: Re: indicator lights on SBB disks) Message-ID: <d40sst$gmc$1@news1.ucsd.edu>   0 In article <1151jifgutuu449@corp.supernews.com>,) Dave Froble  <davef@tsoft-inc.com> wrote:  >Tom Linden wrote:D >> On Sun, 3 Apr 2005 20:23:10 +0000 (UTC), Phillip Helbig---remove = >> CLOTHES  to reply <helbig@astro.multiCLOTHESvax.de> wrote:  >>  H >>> On the Storage Works building-block (SBB) disks (e.g. for use in theK >>> BA35x shelves), there are usually (always?) two indicator lights at the L >>> front.  However, what they mean seems to vary with the disk model.  Some> >>> have one light on all the time, some only during activity. >>   >>  K >> I had 4GB drives in the BA356 cannisters, took them out and replced with K >> 74GB Seagate drives, works like a charm.  of course, I only get 40MB not % >> the 160 the drives are capable of.  >  >Huh?  > , >Want to try that again, doesn't make sense. > G >Why wouldn't you be able to use the full size of the drive?  At least  H >with recent versions of VMS?  I can understand the CONSOLE of an older < >system not working with newer drives, but VMS usually will.  I Sure it makes sense.  But you have to read between the lines.  The 40 and G 160 are referring to the SCSI data transfer rate, not the storage size. H The SCSI bus bandwidth of the controller/backplane is 40Mbits/sec, whileH the drives would be capable of 160Mbit/sec if connected to a more modernH (Low-voltage differential) controller.  Of course, no existing SCSI diskH has a continuous data transfer rate from disk to head that is as fast as its bus interface can achieve.       carl --  B     carl lowenstein         marine physical lab     u.c. san diegoB                                                  clowenst@ucsd.edu   ------------------------------  % Date: Mon, 18 Apr 2005 11:20:09 -0700  From: Z <Z@no.spam>  Subject: Re: Login mystery+ Message-ID: <sTS8e.10029$c42.4377@fe07.lga>    Paul Dembry wrote:F >>If you are in a cluster, the SYSUAF may reside elsewhere in a sharedJ >>area, while your own node may have a very basic SYSUAF in case your node >>ever boots on its own. > 
 > No cluster.   - Are you in SYS$SYSTEM when you run authorize?    ------------------------------  + Date: Mon, 18 Apr 2005 19:03:29 +0000 (UTC) 7 From: moroney@world.std.spaamtrap.com (Michael Moroney)  Subject: Re: Login mystery( Message-ID: <d410a1$lm7$1@pcls4.std.com>  ' "PAUL DEMBRY" <pade@trifox.com> writes:   H >While trying to discover the compiler bug check problem, I thought thatL >perhaps my account (suddenly) doesn't have high enough quotas. However whenK >I run authorize, I cannot even find my user id! I did list * and looked at H >the listing and there is no entry for my user id. How is this possible?% >Again this is on OpenVMS 6.1 on VAX.   E If the logical SYSUAF is _not_ defined on your system, AUTHORIZE will E use the SYSUAF file in the current default directory, and will ask to H create it if it does not exist.  SYSUAF not being defined is the defaultJ configuration of VMS systems, you have to do manually add a define command8 in a startup command file if you don't want the default.  B To use the "real" SYSUAF.DAT if SYSUAF is not defined, either typeI $ SET DEFAULT SYS$SYSTEM or $ DEFINE SYSUAF SYS$SYSTEM:SYSUAF.DAT before   running AUTHORIZE.   --   -Mike    ------------------------------  % Date: Mon, 18 Apr 2005 13:37:20 -0700 , From: "Paul Dembry" <pade@nospam.trifox.com> Subject: Re: Login mystery1 Message-ID: <J8OdnWsnpa-Ph_nfRVn-qA@megapath.net>   B Dimwit here. There are two sysuaf.dat files on this system, one inI [sys0.syscommon.sysexe] and one in [sys0.sysexe]. I am so used to set def K sys0.syscommon.sysexe on our AXP systems that I just went there directly. I H have written "set def sys$system" 50 times on the blackboard as penance. Thanks,  Paul   ------------------------------  # Date: Mon, 18 Apr 2005 20:54:52 GMT # From: hoff@hp.nospam (Hoff Hoffman)  Subject: Re: Login mystery0 Message-ID: <w8V8e.4092$BU.359@news.cpqcorp.net>  ` In article <J8OdnWsnpa-Ph_nfRVn-qA@megapath.net>, "Paul Dembry" <pade@nospam.trifox.com> writes:9 :...There are two sysuaf.dat files on this system, one in 4 :[sys0.syscommon.sysexe] and one in [sys0.sysexe]...  J   Unless the two SYSUAF.DAT files are closely synchronized in terms of theL   UICs involved -- please see my posting over in the thread on node-specificK   rightslist identifiers earlier today for more details -- this dual-SYSUAF J   configuration will tend to lead to run-time problems, and potentially to>   unexpectedly denied or unexpectedly allowed object accesses.  L   Of course, [sys0.syscommon.sysexe] is the common directory and more commonM   home of SYSUAF; it's also known as the [VMS$COMMON.SYSEXE] directory, while M   [sys0.sysexe] is the node-specific system root for whatever node happens to    be using the [SYS0] root.   K   I tend to use a [COMMON] or other similarly-named directory for my shared K   files, and I tend to use the SYSUAF logical name to aim the system at the J   [COMMON]SYSUAF_node.DAT or [COMMON]SYSUAF_usage.DAT file when I must useL   parallel SYSUAF files.  I normally define this and the other related filesH   in SYLOGICALS.COM.  I tend to avoid having the file SYSUAF.DAT in bothH   SYS$COMMON:[SYSEXE] and SYS$SPECIFIC:[SYSEXE], as having the same nameH   looks as much like a configuration error as a deliberate configurationG   decision.  If I have a couple of big, tuned, AlphaServer systems and  I   various smaller VAX workstations, I'll use more generic file names such I   as [COMMON]SYSUAF_PRODUCTION.DAT, [COMMON]SYSUAF_VAX.DAT or other such, A   and share the SYSUAF file across a subset of the cluster nodes.   J   For the list of files expected to be common and shared within a cluster,E   please see the SYLOGICALS.TEMPLATE file on OpenVMS V7.2 and later.        N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com    ------------------------------    Date: 18 Apr 2005 21:58:25 -0700 From: dooleys@snowy.net.au Subject: Re: Login mysteryC Message-ID: <1113886705.087662.285220@z14g2000cwz.googlegroups.com>   D > To use the "real" SYSUAF.DAT if SYSUAF is not defined, either typeC > $ SET DEFAULT SYS$SYSTEM or $ DEFINE SYSUAF SYS$SYSTEM:SYSUAF.DAT  before > running AUTHORIZE.@ I have always used $mcr authorize, believing it to be equivalentE to a $set def sys$system followed by $run authorize. Is this correct?  Phil   ------------------------------  % Date: Tue, 19 Apr 2005 01:40:14 -0400 ' From: Dave Froble <davef@tsoft-inc.com>  Subject: Re: Login mystery0 Message-ID: <11696e721q0r886@corp.supernews.com>   dooleys@snowy.net.au wrote: D >>To use the "real" SYSUAF.DAT if SYSUAF is not defined, either typeC >>$ SET DEFAULT SYS$SYSTEM or $ DEFINE SYSUAF SYS$SYSTEM:SYSUAF.DAT  >  > before >  >>running AUTHORIZE. > B > I have always used $mcr authorize, believing it to be equivalentG > to a $set def sys$system followed by $run authorize. Is this correct?  > Phil >    Most definitely NOT correct.  7 MCR <program> is equavalent to RUN SYS$SYSTEM:<program>   > This will run AUTHORIZE, but if your default directory is not 9 SYS$SYSTEM:, you won't open the system's SYSUAF.DAT file.   B As you write above, set the default directory, then run AUTHORIZE.   ------------------------------  + Date: Mon, 18 Apr 2005 20:45:13 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)) Subject: Re: two questions about MINICOPY $ Message-ID: <d4168p$4kh$2@online.de>  - In article <POUKWEv$MGt1@cuebid.zko.dec.com>, 2 brooks@cuebid.zko.dec.nospam (Rob Brooks) writes:   M > > This makes me wonder what determines which node in a cluster handles the  L > > shadow-copy operation.  Is this documented somewhere?  Can one override H > > the default behaviour (other than by setting SHADOW_MAX_COPY to all 7 > > nodes other than the one which should do the copy)?  > Q > The supported way to influence which node will deal with either a copy or merge J > for any given shadow set is explained in excruciating detail in the V8.2L > New Features manual (well, I suppose it may be the release notes; the doc M > web site is not available now).  As part of the Host-Based Minimerge (HBMM) M > work, we also added a mechanism to allow complete control over the order in < > which nodes make an attempt to handle recovery operations.  I I'm not at 8.2 yet.  I'm still at 7.3-1/7.3.  I had hoped to go to 7.3-2  G last weekend, but got distracted.  I now plan it for the middle of May.   G For the record, a minicopy worked fine with a shadow set with one node  I on a VAX and one node on an ALPHA, doing the dismount and mount from the  C ALPHA.  I have a shadow set split across two VAX nodes, and had to  B reboot one.  I set SHADOW_MAX_COPY to 0 on the VAXes, and did the G DISMOUNT from an ALPHA.  Unfortunately, when the VAX booted it mounted  H the shadow set, and no minicopy was performed.  Since a full copy takes H several hours, I can't test this too often.  Apparently SHADOW_MAX_COPY C being 0 is not sufficient enough; one has to make sure that no VAX  I mounts the shadow set.  (Either that, or at least one member has to have  H a direct connection to an ALPHA, but I can't see why that should be the  case.)  B Has anyone successfully used minicopy on a shadow set with direct G connections to only VAX nodes (issuing the DISMOUNT and MOUNT commands   from an ALPHA, of course)?   ------------------------------  + Date: Mon, 18 Apr 2005 19:17:17 +0000 (UTC) 7 From: moroney@world.std.spaamtrap.com (Michael Moroney) Y Subject: Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot data ( Message-ID: <d4113t$lm7$2@pcls4.std.com>  = koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:   ` >In article <d3vtl8$jvg$1@blackmamba.itd.rl.ac.uk>, "Richard Brodie" <R.Brodie@rl.ac.uk> writes: >>  G >> I think too much has been made of it earlier on this thread. I mean, B >> my TCP/IP kernel responds to ARP packets (which is a completely8 >> different protocol to IP) but I don't worry about it. >>  F >   ARP is a separate protocol, but IP won't run over ethernet without@ >   it.  And there's no point running ARP if you're not doing IP >   (although you could).   H Yes.  IP uses ARP which is a separate protocol, but IP could have easilyD been designed to use a separate function of its own protocol had theF designers so chose.  It appears that ARP _may_ have been designed as aJ general purpose address resolver, not just IP, does anything else actuallyH use it?  The way most IP stacks are written, I don't think anything else can use it.9  I Not that unusual for multiple protocols used by an application.  Even MOPEH uses two, 60-01 for downline load/upline dump, 60-02 for the auto sysid/ remote console thing.a  B >   DECnet and MOP are separate protocols, each cabable of runningD >   without the other and both having full functionality without the
 >   other.  F That is what is important.  DECnet and MOP are really independent.  IPI over ethernet won't work without ARP, and I don't know if ARP is used for  anything other than IP.M -- k -Mikee   ------------------------------   Date: 18 Apr 2005 19:29:43 GMT( From: bill@cs.uofs.edu (Bill Gunshannon)Y Subject: Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot datac, Message-ID: <3cig57F6mck4pU1@individual.net>  ( In article <d4113t$lm7$2@pcls4.std.com>,: 	moroney@world.std.spaamtrap.com (Michael Moroney) writes:? > koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:L > a >>In article <d3vtl8$jvg$1@blackmamba.itd.rl.ac.uk>, "Richard Brodie" <R.Brodie@rl.ac.uk> writes:h >>> H >>> I think too much has been made of it earlier on this thread. I mean,C >>> my TCP/IP kernel responds to ARP packets (which is a completely49 >>> different protocol to IP) but I don't worry about it.i >>> G >>   ARP is a separate protocol, but IP won't run over ethernet without/A >>   it.  And there's no point running ARP if you're not doing IPd >>   (although you could). > J > Yes.  IP uses ARP which is a separate protocol, but IP could have easilyF > been designed to use a separate function of its own protocol had theH > designers so chose.  It appears that ARP _may_ have been designed as aL > general purpose address resolver, not just IP, does anything else actuallyJ > use it?  The way most IP stacks are written, I don't think anything else
 > can use it.t  E ARP is part of the IP Suite and not some "separate protocol".  It mayyD not be part of TCP, but it is definitely part of IP.  It is also not, the only IP Protocol that is outside of TCP.   bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   S   ------------------------------    Date: 18 Apr 2005 16:06:54 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)yY Subject: Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot datal3 Message-ID: <7eeBu5aN3EW5@eisner.encompasserve.org>e  W In article <3cig57F6mck4pU1@individual.net>, bill@cs.uofs.edu (Bill Gunshannon) writes:h > G > ARP is part of the IP Suite and not some "separate protocol".  It maytF > not be part of TCP, but it is definitely part of IP.  It is also not. > the only IP Protocol that is outside of TCP.  K    ARP is a separate protocol from the point of view of ethernet by virtue .-    of its different ethernet protocol number.d  E    IP did run over other transports prior to ethernet and did not uset    ARP for them.   ------------------------------  + Date: Mon, 18 Apr 2005 22:39:11 +0000 (UTC)s7 From: moroney@world.std.spaamtrap.com (Michael Moroney) Y Subject: Re: VMS FAQ: changing volume label of system disk: DECnet MOP or LANCP boot datae( Message-ID: <d41cuf$8l3$1@pcls4.std.com>  * bill@cs.uofs.edu (Bill Gunshannon) writes:  K >> Yes.  IP uses ARP which is a separate protocol, but IP could have easilyeG >> been designed to use a separate function of its own protocol had therI >> designers so chose.  It appears that ARP _may_ have been designed as ajM >> general purpose address resolver, not just IP, does anything else actuallySK >> use it?  The way most IP stacks are written, I don't think anything else  >> can use it.  > >ARP is part of the IP Suite and not some "separate protocol".  G OK.  It appeared to allow for variable lengths for both the mac address D and the field that the IP address uses so I figured there "might" be something else using it.  	 >  It mayy5 >not be part of TCP, but it is definitely part of IP.*  D Yup.  I figured if they wanted to, they could have added a differentH ICMP function and sent the request to the broadcast address using the IP	 protocol.n   >  It is also not - >the only IP Protocol that is outside of TCP.   G I think you mean the only LAN protocol used by IP outside of IP's main tG protocol itself and ARP, correct?  The different protocols on top of IPiG (TCP, UDP, ICMP and a bunch of others nobody seems to use) are the nexts, layer up and don't apply to this discussion.  G I know of reverse-ARP, which isn't used much I guess, are there others?  -- . -Mike'   ------------------------------  % Date: Mon, 18 Apr 2005 11:57:42 -0700 , From: Ken Fairfield <my.full.name@intel.com>: Subject: Re: What do you expect from DIR/SIZ=(ALLO,USED) ?+ Message-ID: <d40vv6$88q$1@news01.intel.com>,    Peter 'EPLAN' LANGSTOEGER wrote:\ > In article <d3phpq$kin$1@news01.intel.com>, Ken Fairfield <my.full.name@intel.com> writes: >  [...]eE >>    Given that no syntax error "should" be generated given the CLD,4E >>then we fall back to sequential processing, along the lines of (b),v >>and get the answer (3).  >  > 5 > It is _not_ sequential. USED always wins currently.o  C      Ah, that's interesting...  I can _imagine_ why this is so, butf) I won't guess.  Well, OK, I'll guess. :-)d  C      The CLD defines the syntax at DCL, but it doesn't define how aeA program processes the command line.  It _looks_ like DIRECTORY isfC reparsing the command line internally and looking for keywords in afF particular order.  In a similar way, adding the "ALL" keyword anywhere@ in the list results in both used and allocation being displayed.  C      I'm still of the opinion (at least prior to V7.3-2 or wheneverdC the UNITS keyword was added) that putting a list of ALL, ALLOCATIONaA and/or USED is a syntax violation, as defined by HELP and the DCLeD Dictionary, and DIRECTORY can (and does!) do anything it likes.  TheD CLD for DIRECTORY is substandard in that it doesn't catch the error.   > E >>    Anyone want to file an SPR against DIRECTORY for not generating  >>a syntax error?  :-) >  >  > Not me, I'm afraid.o  F       What's to be afraid of?  Being ignored by VMS Engineering??? :-)   	-Kenf --  6 I don't speak for Intel, Intel doesn't speak for me...  
 Ken Fairfieldy! D1C Automation VMS System Supportt" who:   kenneth dot h dot fairfield where: intel dot com   ------------------------------   End of INFO-VAX 2005.217 ************************