1 INFO-VAX	Mon, 09 Apr 2001	Volume 2001 : Issue 199       Contents: Re: CD backups.  Re: CD backups.  CETS2001 Web Site + Re: CI based cluster for max I/O perf. How? + Re: CI based cluster for max I/O perf. How? + Re: CI based cluster for max I/O perf. How?  Corba on OpenVMS AXP Re: Corba on OpenVMS AXP Re: Corba on OpenVMS AXP  RE: DEC Server software question  Re: DEC Server software question2 Re: Flipping from big-endian to little endian in C2 Re: Flipping from big-endian to little endian in C2 Re: Flipping from big-endian to little endian in C2 Re: Flipping from big-endian to little endian in C2 Re: Flipping from big-endian to little endian in C	 FMS files 
 Re: FMS files  RE: FTP hijacking of VMS sites6 Re: Galaxy Network Interfaces/cards and TCP/IP Sockets* Help needed with RMS Character Field Index* Help needed with RMS Character Field Index9 Re: Interesting news from those nice folk at The Register ' Re: LINK-I-DATMISMCH since some patch ?  More VMS ads spotted in the UK" Re: More VMS ads spotted in the UK( NSLOOKUP: *** Can't initialize resolver., Re: NSLOOKUP: *** Can't initialize resolver., Re: NSLOOKUP: *** Can't initialize resolver., Re: NSLOOKUP: *** Can't initialize resolver.- OpenVMS Job seeking in Hong Kong, SAR, China. " OpenVMS Times T4 tool availability Re: OT: "The Puzzle Palace"  Re: OT: "The Puzzle Palace"  Re: OT: "The Puzzle Palace"  Re: OT: "The Puzzle Palace"  Re: OT: "The Puzzle Palace"  Re: OT: "The Puzzle Palace"  Re: OT: "The Puzzle Palace"  RE: OT: "The Puzzle Palace" * Re: Problems with threads on VMS AXP 7.1-2 Re: Q: Unmapping global section & Running a Diagnostic on a Network Card Re: Shadowing Weirdness  Re: Shadowing Weirdness  Re: sys$io_performw  Re: UCX 5.1 BG device problem  Re: UCX 5.1 BG device problem & Re: VMS equivalent of Unix wc command? Re: VMS-Related: Affordable  Re: VMS-Related: Affordable   F ----------------------------------------------------------------------   Date: 9 Apr 2001 09:22:47 CDT = From: wayne@tachysoft.xxx.065234.killspam.015d (Wayne Sewell)  Subject: Re: CD backups.. Message-ID: <nRxCS0UIjnzg@tachxxsoftxxconsult>  m In article <9aqi1r$id2$1@newsg1.svr.pol.co.uk>, "Leigh G. Bowden" <LGBowden@bowdenfamily.fsnet.co.uk> writes: / > The way I read it I have to do the following:  > " > Attach a CD writer to a machine.G > Put what ever I want to preserve onto a HDD which is less than 650MB.  > Mount this HDD as /FOREIGN. ; > Then do a COPY disk: another_disk:[000000.cds]cd_disk.rom 8 > This generates a few whinges but these can be ignored.I > Use CDRECORD freeware utility to burn the CD with the file cd_disk.rom. I > Ensure that the machine doing the burning is doing nothing else to stop ! > interrupting the write process.     M Yes, that will work.  You may need to supply a /allocate=<block-count> on the N copy, where block-count is the next larger multiple of 100.  I have found thatM cdrecord does not like block counts that do not fall on the correct boundary.   K If you only want to preserve certain files, you can skip the step where you O copy the files to a regular disk and mount it as foreign.  If you get a virtual O disk program, such as the LDdriver program found on the freeware cdrom, you can G copy the files to the virtual disk container file and burn from that.     J The advantage to this approach is that you can burn a cdrom of exactly theL right size.  You can create the virtual disk initially with the full size ofM 650 meg.  Then after you have copied your files to it, you can check how many N blocks are actually used, reinitialize the virtual disk to that size, and copyO the files again.  If you don't have many files, this can drastically reduce the ? burn time, since you aren't burning thousands of unused blocks.   F I use the following set of command procedures (see description after):  : ========================== CREATE_CONTAINER_FROM_FILES.COMK $ create_verify = f$verify (f$trnlnm ("create_container_files_verify") + 0)  $! $! p1 = label to use on cdrom > $! p2 = top level directory name for cdrom; if blank, [000000] $! p3 = format of file input8 $!	FILE  --  single file name, with or without wildcardsG $!	LIST  --  list of file names, with or without wildcards and dest dir ; $!	TAPE  --  tape device; copy p4 directly to top directory 0 $!	SAVE  --  save set; restore saveset(s) in p4 7 $!	TSAVE  --  tape save set; restore saveset(s) on tape  $! p4 = files  $! p5 = set ownership  $! $!
 $ set noon $ say :== write sys$output $ label = p1 $ top_lev_direct = p2  $ file_type = p3 $ file_param = p4  $ file_owner = p5  $  $! $!0 $ if f$trnlnm("container_scratch_disk") .eqs. "" $	then3 $		say "container_scratch_disk logical not defined"   $ 		v = f$verify (create_verify)	 $		exit 4  $	endif  $! $ if label .eqs. ""  $	then$ $		say "label parameter is required"  $ 		v = f$verify (create_verify)	 $		exit 4  $	endif  $! $ if file_type .eqs. ""  $	then( $		say "file_type parameter is required"  $ 		v = f$verify (create_verify)	 $		exit 4  $	endif  $! $! $ if file_param .eqs. "" $	then) $		say "file_param parameter is required"   $ 		v = f$verify (create_verify)	 $		exit 4  $	endif  $! $! $! Startup LD if needed  $!B $ if .not. f$getdvi("lda0:","exists") then @sys$startup:ld$startup $! $! $!# $ container_name = label + "_image" W $ container_file = "container_scratch_disk:[cdrom_container_files]" + container_name +-  		 ".container;1" 7 $ credir container_scratch_disk:[cdrom_container_files] 6 $ say "searching for container file ''container_file'"$ $ file = f$search(container_file,42) $ if file .nes. "" $	then: $		say "container file ",container_file," already present" $ 		del 'container_file' $	endif  $! $! $!1 $! Use the full container file for the first pass  $!5 $ ld create /noback/log/size=1327100 'container_file'  $ status = $status $ if .not. status  $	then  A $		say "unable to create container file the full size of a cdrom"  $ 		delete/log 'container_file'   $ 		v = f$verify (create_verify) $		exit status $	endif  $! $! Populate the container file $! $ call setup_container_file  $ status = $status $ if .not. status  $	then   $ 		delete/log 'container_file'   $ 		v = f$verify (create_verify) $		exit status $	endif  $!G $! If there is a lot of free space, then make the container file a few  L $! thousand blocks larger than the actual data and repopulate it.  This willE $! speed up the cdrom burn immensely if there is a lot of free space.  $! $! $ if free_blocks .gt. 20000  $	then* $		say "container file has ",free_blocks,-7 			" blocks free---squeezing disk to ",blocks," blocks"  $ 		delete/log 'container_file' 8 $ 		ld create /noback/log/size='blocks' 'container_file' $ 		call setup_container_file  $ 		status = $status $ 		if .not. status 	 $			then  ! $ 				delete/log 'container_file' " $ 				v = f$verify (create_verify) $				exit status	 $			endif  $	endif  $!1 $ say "container file ",container_file," created"  $ exit $! $! $!! $setup_container_file: subroutine % $ ld conn/log/symbol 'container_file' 
 $ ownopt = "" = $ if file_owner .nes. "" then ownopt = "/own=[''file_owner']" , $ init/nohigh'ownopt' lda'ld_unit': 'label'	% $ mount lda'ld_unit': 'label' 'label' ; $ if top_lev_direct .eqs. "" then top_lev_direct = "000000" # $ credir 'label':['top_lev_direct'] 3 $ define cdrom_top_level 'label':['top_lev_direct']  $ dir/secur 'label':[000000] $ sho dev/ful 'label'  $ call copy_the_files  $ status = $status $ if .not. status  $	then: $ 		say "Unable to fit backup of ''full_filename' on CD-R" $ 		del cdrom_top_level:*.*;*  $	else2 $ 		freeblocks = f$getdvi("''label'","freeblocks") $ 		free_blocks == freeblocks  $ 		if freeblocks .gt. 20000 $			then1 $ 				maxblocks = f$getdvi("''label'","maxblock") . $ 				blocks == maxblocks - freeblocks + 20000 $ 				blocks == blocks / 100 $ 				blocks == blocks * 100/ $				if blocks .lt. 20000 then blocks == 20000  	 $			endif  $	endif  $ if status  $	then   $ 		if 0 ! file_owner .nes. ""   $			then" $				set file/own=['file_owner'] -' 					'label':['top_lev_direct'...]*.*;* 	 $			endif = $ 		dir/ful/nohe/notr/exclu=(*.sys,*.dir) 'label':[000000...]   $		dir/secur 'label':[000000...] $ 		sho dev/ful 'label'  $	endif  $ dismount 'label' $ ld discon/log lda'ld_unit': 
 $ exit status  $endsubroutine $! $! $copy_the_files: subroutine  $ call copy_'file_type $ status = $status $ if .not. status then exit 4  $endsubroutine $! $! $! $! $! $!D $copy_file: subroutine	! single file name, with or without wildcards $!			! p1 = target directory $!			! p2 = target file  $ targdir = p1 $ targfile = p2 : $ if targdir .eqs. "" then targdir = "[''top_lev_direct']"1 $ if targfile .eqs. "" then targfile = file_param ; $ if targdir .nes. "[000000]" then credir 'label':'targdir' $ $ copy 'targfile'  'label':'targdir' $ status = $status $ if .not. status then exit 4  $endsubroutine $! $!G $copy_list: subroutine	! list of file names, with or without wildcards   $			!	and dest dir. $ open/read/err=bad_file infile 'file_param'   $ readloop: $ $ read/end=endreadloop infile inline) $ inline = f$edit(inline,"compress,trim") $ $ the_file = f$element(0," ",inline)" $ if the_file .eqs. "" then exit 4# $ the_dir = f$element(1," ",inline) ( $ call copy_file "''the_dir'" 'the_file' $ status = $status $ if .not. status then exit 4  $ goto readloop  $ endreadloop: $ close infile   $ exit
 $bad_file: $ exit 4 $endsubroutine $! $!G $copy_tape: subroutine	! tape device; copy p4 directly to top directory  $ mount/over=id 'file_param'! $ targdir = "[''top_lev_direct']"  $ credir 'label':'targdir'* $ copy 'file_param':*.*  'label':'targdir' $ status = $status  $ dismount/nounload 'file_param' $ if .not. status then exit 4  $endsubroutine $! $!< $copy_save: subroutine	! save set; restore saveset(s) in p4  $ save_loop:  $ saveset = f$search(file_param) $ if saveset .eqs. "" then exit ( $ backup/log 'saveset'/save  'label':[*] $ status = $status $ if .not. status then exit 4  $ goto save_loop $endsubroutine $! $!C $copy_tsave: subroutine	! tape save set; restore saveset(s) on tape  $endsubroutine  6 ========================== COPY_CDROM_TO_CONTAINER.COM  J $ copy_verify = f$verify (f$trnlnm ("copy_cdrom_to_container_verify") + 0) $! $! p1 = cdrom device name  $!
 $ set noon $ say :== write sys$output $ cdrom_device = p1 - ":"  $  $ if cdrom_device .eqs. "" $	then# $		say "cdrom device not specified"  $ 		v = f$verify (copy_verify)	 $		exit 4  $	endif  $! $! $!0 $ if f$trnlnm("container_scratch_disk") .eqs. "" $	then3 $		say "container_scratch_disk logical not defined"  $ 		v = f$verify (copy_verify)	 $		exit 4  $	endif  $! $! $! Startup LD if needed  $!B $ if .not. f$getdvi("lda0:","exists") then @sys$startup:ld$startup $! $! $! $! $! $ mount/over=id 'cdrom_device'* $ label = f$getdvi(cdrom_device, "volnam"); $ create/dir container_scratch_disk:[cdrom_container_files] N $ container_file = "container_scratch_disk:[cdrom_container_files]" + label +- 		 "_image.container;1" 6 $ say "searching for container file ''container_file'"$ $ file = f$search(container_file,42)- $ if file .eqs. "" then call create_container  $ dism 'cdrom_device'  $ mount/for 'cdrom_device'1 $ copy/overwrite 'cdrom_device': 'container_file'  $ dism 'cdrom_device'  $ exit $! $! $! $! $ create_container: subroutine7 $ freeblocks = f$getdvi("''cdrom_device'","freeblocks") 4 $ maxblocks = f$getdvi("''cdrom_device'","maxblock")) $ blocks = maxblocks - freeblocks + 20000  $ blocks = blocks / 100  $ blocks = blocks * 100 + $ if blocks .lt. 20000 then blocks = 20000  6 $ ld create /noback/log/size='blocks' 'container_file' $ endsubroutine   , ========================== BACKUP_TO_CDR.COMB $ backup_verify = f$verify (f$trnlnm ("backup_to_cdr_verify") + 0)
 $ set noon. $ diag = f$trnlnm ("backup_to_cdr_verify") + 0	 $ if diag  $	then $		shosym = "sho sym"  $	else $		shosym = "!"  $	endif  $! $ say :== write sys$output $  $!K $ container_file = f$search("all_disks:[cdrom_container_files]*.container") + $ found_something = container_file .nes. ""  $!F $ if found_something .and. f$trnlnm("container_scratch_disk") .eqs. "" $	then3 $		disk = f$parse(container_file,,,"device")  - ":" ' $		define container_scratch_disk 'disk'  $	endif  $!0 $ if f$trnlnm("container_scratch_disk") .eqs. "" $	then3 $		say "container_scratch_disk logical not defined"   $ 		v = f$verify (backup_verify)	 $		exit 4  $	endif  $! $ call get_SCSI_bus  $ status = $status $ if .not. status  $	then    $ 		v = f$verify (backup_verify)	 $		exit 4  $	endif  $ call get_SCSI_unit $ status = $status $ if .not. status  $	then    $ 		v = f$verify (backup_verify)	 $		exit 4  $	endif  $ call get_real_or_dummy $ status = $status $ if .not. status  $	then    $ 		v = f$verify (backup_verify)	 $		exit 4  $	endif  $!N $ container_sear = "container_scratch_disk:[cdrom_container_files]*.container" $ cdrecord = "$util:cdrecord" = $ define backup_scratch container_scratch_disk:[scratch_area] 2 $ create/dir container_scratch_disk:[scratch_area]
 $get_file:+ $ container_file = f$search(container_sear)  $! $ if container_file .eqs. "" $	then8 $		say "no more files found in container file directory"  $ 		v = f$verify (backup_verify)	 $		exit 1  $	endif  $!: $ say "preparing to burn container file ''container_file'"# $ say "load blank CD-R into drive"  ( $ INQUIRE ready "hit return when ready"  $! $!< $! Now that the container file is set up, do the actual burn $!. $ if wrpar .eqs. "write" then write_param = ""8 $ if wrpar .eqs. "dummy" then write_param = """-dummy"""5 $ cdrecord_params = """-speed"" 2 " + write_param + - 9 		" ""dev=''scsi_bus_number',''scsi_unit_number',0"" " +-  		container_file $ shosym cdrecord_params  & $ say "starting the burn, please wait" $ set term/nobroad4 $ define/user sys$output backup_scratch:cdrecord.log $ define/user sys$error nl:  $ cdrecord 'cdrecord_params' $ status = $status $ if status  $	then' $		how_ended = "completed successfully"  $	else $		how_ended = "failed $	endif ? $ say "Backup of ''container_file' to CD-R has ''how_ended'..." 5 $ set file/attr=rfm:stmlf backup_scratch:cdrecord.log   $ ty backup_scratch:cdrecord.log $ set term/broad# $ del backup_scratch:cdrecord.log;*  $ if .not. status  $	then $		say "aborting procedure"   $ 		v = f$verify (backup_verify) $		exit status $	endif  $ if wrpar .eqs. "dummy"   $	then  F $		say "this was a dummy write, so container file will not be deleted" $		goto get_file $	endif 	 $inq_del: = $ inquire which "burn complete, delete container file? [y,n]"  $ if which .eqs. "Y" $	then $ 		del 'container_file'   $		goto get_file $	endif ' $ if which .eqs. "N" then goto get_file  $ goto inq_del $! $! $! $get_SCSI_bus: subroutine  $ scsi_bus_number == -1 
 $get_bus: 	 $ say " " b $ inquire SCSI_bus "On which SCSI bus does the CD writer reside (0 for dkaxxx, 1 for dkbxxx, etc "' $ if SCSI_bus  .eqs. "EXIT" then exit 4  $ scsi_bus_number == 'scsi_bus' = $ if (scsi_bus_number .lt. 0) .or. (scsi_bus_number .gt. 15)  	 $   then  - $    say " *********************************" - $    say " ***** Invalid SCSI bus **********" - $    say " *********************************"  $    say " " $    goto get_bus  $ endif  $endsubroutine $! $! $! $get_SCSI_unit: subroutine $ scsi_unit_number == -1 $get_unit:  	 $ say " " ; $ inquire SCSI_unit "What is the SCSI ID of the CD writer " ( $ if SCSI_unit  .eqs. "EXIT" then exit 4! $ scsi_unit_number == 'scsi_unit' > $ if (scsi_unit_number .lt. 0) .or. (scsi_unit_number .gt. 7) 	 $   then  - $    say " *********************************" . $    say " ***** Invalid SCSI unit **********"- $    say " *********************************"  $    say " " $    goto get_unit   $ endif  $endsubroutine $! $! $get_real_or_dummy: subroutine $get_real_or_dum: 	 $ say " " 6 $ inquire dummy_or_real "dummy (for testing) or real ", $ if dummy_or_real  .eqs. "EXIT" then exit 4H $ if (dummy_or_real  .nes. "DUMMY") .and. (dummy_or_real  .nes. "REAL") 	 $   then  < $    say " ************************************************"< $    say " ***** Invalid -- must be DUMMY or REAL *********"< $    say " ************************************************" $    say " " $    goto get_real_or_dumt $ endifD $ wrpar=="dummy"4 $ if dummy_or_real  .eqs. "REAL" then wrpar=="write" $! $! $endsubroutine $! $! ========================== . ========================== f ==========================    N This setup allows me to create multiple container files and burn them all backN to back.  Obviously, you must have enough disk space for several cdrom images, each up to 650 meg in size.   M The CREATE_CONTAINER_FROM_FILES command procedure accepts a list of files andiJ copies them to a virtual disk of exactly that size.  Well, except that the5 block count is padded to the next 100-block boundary.   N The COPY_CDROM_TO_CONTAINER copies an existing disk to a container file, whichN is similar to the method you describe in your post.  The procedure is intendedO for duplicating cdroms, but there is no reason why it can't work with any ods-2RL disk of cdrom size.  Again, the resulting container is just large enough for' the data, rather than the full 650 meg.   M The interactive BACKUP_TO_CDR procedure burns cd-rs for each of the containerlM files you created with the other two procedures.  Since a container is just atM container, the container files could have been created with either or both of N the other two command procedures.  The burn procedure just scans the container7 directory until there are no more container files left.e   WayneT -- eO =============================================================================== M Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachysoft.xxxm: http://www.tachysoft.xxx/www/tachyon.html and wayne.html  K change .xxx to .com in addresses above, assuming you are not a spambot  :-) O ===============================================================================tB Jed Clampett, checking into hotel: "This place got a cement pond?"+ Ellie May: "And do yuh let critters in it?"-   ------------------------------  $ Date: Mon, 9 Apr 2001 10:46:58 -0400% From: "John Vottero" <John@mvpsi.com>s Subject: Re: CD backups./ Message-ID: <td3ircj1bvki37@news.supernews.com>z  5 Good luck finding a hard disk that's less than 650MB..  L It seems easier to use one of the virtual disk drivers (VD or LD).  Then youG can create a virtual disk that's any size you need and you can skip therI mount/copy part.  Another advantage is that you can create a virtual disk 4 only the size you need so that the burns go quicker.  F "Leigh G. Bowden" <LGBowden@bowdenfamily.fsnet.co.uk> wrote in message) news:9aqi1r$id2$1@newsg1.svr.pol.co.uk... / > The way I read it I have to do the following:  >e" > Attach a CD writer to a machine.G > Put what ever I want to preserve onto a HDD which is less than 650MB.e > Mount this HDD as /FOREIGN. ; > Then do a COPY disk: another_disk:[000000.cds]cd_disk.romo8 > This generates a few whinges but these can be ignored.I > Use CDRECORD freeware utility to burn the CD with the file cd_disk.rom.oI > Ensure that the machine doing the burning is doing nothing else to stopl! > interrupting the write process.  >oI > Leigh G. Bowden wrote in message <9afu0b$ec4$1@newsg3.svr.pol.co.uk>...aK > >I suspect that this has been covered before but I can't find it. We haveoL > >quite a few remote sites with master/slave pairs of uVAX 3100/nn machinesL > >all running VMS6.2. Each has three SCSI 2.1GB disks and a CD. I'll not goJ > >into the history but there is no way of doing backups of these machinesL > >easily - there's certainly no tape drives and nobody to really put a tape > in > >reliably from day to day. > >aI > >Should a disk fail it is fairly easy to recreate it's data on the datat > disksyI > >from it's mirror but the system disk is harder to replace. There isn'tn? > >enough space on any of the other disks for a system saveset.a > >oK > >I was wondering therefore about using the CD. I know VMS doesn't supportEJ > >CD-W type devices. If I create a saveset on VMS machine and then FTP it inI > >binary mode to a PC and then blat it on a CD also in binary mode wouldt this > >work? > >fI > >The idea being to have standalone backup on all the other HDD and bootm fromK > >them. Mount the CD and restore the contents. It sounds to easy there hasc to > >be a catch. > >P > >  >p >e   ------------------------------  % Date: Mon, 09 Apr 2001 09:37:40 -0400v& From: Ken Robinson <ksrobin@erenj.com> Subject: CETS2001 Web Site7 Message-ID: <5.0.0.25.2.20010409093556.0406aeb0@clmail>e  G When I tried going to the CETS2001 website this morning, I got a "404"  
 error. :-(  = Is it down because it's going live? Or is it another problem?n   Thanks Ken Robinson   ------------------------------   Date: 9 Apr 2001 01:40:48 -0500f+ From: young_r@encompasserve.org (Rob Young)e4 Subject: Re: CI based cluster for max I/O perf. How?3 Message-ID: <Z9q9hXcFwNPE@eisner.encompasserve.org>(  _ In article <3AD05FDD.386729A8@pressenter.com>, Lyndon Bartels <lbartels@pressenter.com> writes:e > Hmmm.... To follow up. > J > I've been looking at the manual "Guidelines for cluster configurations,"I > Chapter 9. There are four configurations there that I'm interested in. a > G > Figure 9-1, is our current configuration. It has two CIPCAs per host,.D > and two HSJ per storage subsystem. All going through a single star
 > coupler. > H > Now according to this configuration, it looks like the second CIPCA is > not helping performance. >    	Not substantially:r  @ http://www.openvms.compaq.com:8000/72final/6318/6318pro_003.html  P "throughput does not increase substantially when you connect a second CI adapter to the same star coupler. "h  > 	In fact, if you are nearing saturation on a star you can rest@ 	assured adding additional CI adapters won't matter much.  I had> 	someone try to "sell" me on the additional CIPCA "idea" in a * 	single star and I didn't buy off on that.  ; 	However, if you add another star, throughput does go up asa 	that reference points out:c  M "You can increase throughput by connecting additional CI adapters to separatei star couplers;"   J > The second configuration looks a bit more promising. That introduces theE > a second star coupler. This one states that the second star coupleriH > provides increased I/O performance and availability. The disadvantagesF > here are: If a CIPCA fails disks won't fail-over to the other CIPCA. > Is this still true?     ; 	In the future, it helps to drop the URL out for us to see.h> 	All this stuff is available online.  The second configuration 	shows up here:s  Q http://www.openvms.compaq.com:8000/721final/6318/6318pro_012.html#config_assump_hs  E 	One of the notes for 9.4 Configuration 2, points out you still have   	connectivity:  N "Rationale: Either of a host's CI adapters can fail, and the host will retain G CI connectivity to the other host and to the HSJ storage controllers. "_  < 	Keep in mind what you are seeing is that HSJ1 and HSJ2 are ; 	a dual-redundant pair.  Does Config 2 have a single point e7         of failure? (1)  Trick question.  Read further.s  : 	Finally, on this you do find a disadvantage but it isn't  	a showstopper:w  N "Host CI adapter failure will not cause disks to automatically fail over to anM HSJ that still has full host connectivity.  If a host CI adapter fails and ifsM MSCP serving is enabled, then the other OpenVMS system will begin serving thee= unreachable HSJ's disks to the host with the failed adapter."l    C 	This means the nodes will still see each other via the other star.=B 	But drives in this dual-redundant config will be online to either> 	HSJ1 or HSJ2.  If Disk1 is online to HSJ1, it won't failover.  < 	The disks can still be reached as they are served back from? 	the other Host via MSCP serving.  It is VERY important to have B 	MSCP serving turned on for this to occur :-).  MSCP is additionalB 	overhead, now THAT is true overhead.  Can this be avoided?  Sure,D 	pop in two more CIPCAs in each host.  But now we are venturing into 	overkill territory. 	  > E > Configuration 4 (Figure 9-4) splits the disks and HSJs in half. I'mtJ > going to stay away from this one, because to get availability, I'll need> > to do host-based shadowing, and that'll take away some of my > performance. >   2 	Being vague... being vague as Evil Machine people 	may be listening... Shhhhh!  : 	I spent a couple dozen or so emails and many painful days? 	recently shooting down the nasty myth regarding "overhead" of = 	Volume Shadowing.    < 	The Wizard addresses the mythical volume shadowing overhead 	issue.u  & 	Search the title for "volume shadow":  / http://www.openvms.compaq.com/wizard/index.htmlp  @ 	Note topic 4407.  The Wizard talks about memory structures (who? 	cares, memory is cheap and how much can this be) and a "little"= 	bit of overhead" when determining which member to read from,nA 	additional write overhead, etc.  I went about using a noteworthy > 	performance tool to attempt to track down this mythical beast= 	and danged if I could find it!  So little CPU is consumed bywA 	the SHADOW_SERVER it *seldom* registers (Admission:  I only lookfH 	at daytime, nighttime when heavy shadow lifting - splits - is going on D 	, few are on the system and who cares what SHADOW_SERVER is doing).  B 	Now maybe somebody comes along and tries to rain on my parade andA 	point out that Shadow overhead is buried in the Kernel and shows$> 	up as Kernel usage.  I'm not seeing that either.  So if there@ 	is overhead, it is hiding so well I could give a rip if it ever 	comes out of hiding.a  C 	Secondly, with writeback cache turned on, the added "overhead" or n? 	delay on writes of writing to two or more physical disks is a dG 	non-issue.  Also, other vendors have write cache turned on by default sA 	in their products so this too is a straw dog. HA.  (You may noten@ 	CI bandwidth concerns because of double the writes, but what is3 	your write/read ratio?  How saturated is your CI?)   C 	Volume Shadowing is the way to go.  Split across two star couplers'3 	and use writeback caching at the controller level..   > ? > I'm thinking about replacing HSJ50s (considering they're nearpH > end-of-life) with HSJ80s. Would I then "Have to" upgrade to the 16 bit > I/O modules. >   > 	Why stick with CI?  Why not fibre?  It is the future and will> 	far outperform HSJs.  And from what I see it is stable enough	 	now ;-).    > B > Disk-drive layout: I need both I/O speed and availabilty. So I'mF > thinking RAID 0+1. Since there are six (6) buses on a given HSJ. I'mG > thinking of using 6 physical disk per storage unit. 3 members striped!H > and 3 more mirroring. Laying them out so that each physical is on it's
 > own bus. >   F 	That's good.  Now how much is in your disk budget?  Shadow to another= 	member just like it on another controller pair off the othergE 	star coupler and you are good to go.  In rare instances, controllers!A 	have been wedged.  A few folks have shared that experience here."F 	What that means is a contoller fails and the devices fail to failover= 	to the remaining dual-redundant controller.  They just hang.e  D 	Secondly, that 0+1 is online to one controller.  If that controllerB 	blows out its cache and you have writeback caching turned on, youA 	may be blowing the dust off your resume as those writes are lost C 	forever.  That is why many of us shadow across separate controllereA 	pairs.  I have had a cache module blow out.  Fortunately, it wasc: 	in the process of copying a database prior to going live.   > F > Considering the discussion regarding pagefiles/disk loss. Would I beJ > better off not having locally hosted SCSI disks for page files, and moveE > the page files onto CI based storage? I'm reluctant to do this. I'mt? > thinking I may loose paging performance.... But I'm not sure.  >   < 	How much do you page?  How congested (pending I/O) are yourE 	page disk(s)?  If page files are local, it is true when they go, youn= 	lose that host.  You do have more hosts, right?  But you are D 	correct in noting the higher availability of your pagefiles if theyF 	are on an HSJ WriteBack mirrorset.  An HSJ mirrorset could outperformC 	the locally attached disk drive *if* you are not swapping heavily, > 	which I suspect you aren't.  (Swapping, i.e. moving out large= 	working set based processes could be sending MBytes worth ofe 	data to disk).f  > 	Which brings up tuning Zen.  Tuning is not mysterious but hasG 	to be approached methodically. I do take time to know what everything lG 	is doing.  I monitor disks, controllers, CPUs and whatever else I can iD 	think of.  If you currently aren't saturated, then maybe you don't / 	have as big a problem as you first supposed.  i  ? 	Do you have bottlenecks?  If you are "not sure", you certainly.B 	don't want to spend money on the wrong things.  Find a bottleneck 	and THEN spend the money. :-)   				Rob=    O (1)  Yes.  Your single point of failure is the highly available dual-redundant  I      controller pair.  Disks can get wedged.  Rare event, but possible.  iM      Secondly, if you use writeback caching.. HSJ50s do not support mirrored iL      cache.  The cache board is a SPOF.  However, HSJ80s do support mirroredK      cache.  Finally, you will need to go to two stars with HSJ80s as they nN      will saturate a single star or so I have been told.  Also, they have TWO       sets of ports (A and B).p   ------------------------------  # Date: Mon, 09 Apr 2001 05:54:56 GMTl From: Dirk Munk <munk@home.nl>4 Subject: Re: CI based cluster for max I/O perf. How?' Message-ID: <3AD14EB0.48B83A2F@home.nl>o  
 Hi Lyndon,  C Two thinks strike me reading your message. You are interested in IOaI performance, and you're thinking of replacing your HSJ50's by HSJ80's ando using 16 bit shelves.   L The principle behind CI is beautiful, no doubt about it. It was way ahead ofK it's time when it was introduced. But these days it is not what you want iflG you're looking for speed. The maximum speed on a CI bus is appr. 8.5 MBdK /sec, so if you would buy a second star coupler you could get 17 MB/sec fort all hosts together.a  K Now look at fibrechannel. In theory you can get 100 MB/sec per host adapter"J !! Of course it depends how many HSG80's you would use, but anyway your IO= speed would be magnitudes more then what you can get with CI.   J So if you're considering investing, take a very good look at fibrechannel.J If I'm not mistaken there is even talk about fibrechannel adapters for Vax systems.   Regards,   Dirk   Lyndon Bartels wrote:    > Hmmm.... To follow up. >nJ > I've been looking at the manual "Guidelines for cluster configurations,"H > Chapter 9. There are four configurations there that I'm interested in. >eG > Figure 9-1, is our current configuration. It has two CIPCAs per host,rD > and two HSJ per storage subsystem. All going through a single star
 > coupler. >oH > Now according to this configuration, it looks like the second CIPCA is > not helping performance. >uJ > The second configuration looks a bit more promising. That introduces theE > a second star coupler. This one states that the second star couplernH > provides increased I/O performance and availability. The disadvantagesF > here are: If a CIPCA fails disks won't fail-over to the other CIPCA. > I > Is this still true? This manual was written in 1999. Is that still true*; > with the v7.2-1? Also, the manual states somewhere to useeH > decw$examples:prefer to set the paths of disks to load balance betweenH > HSJs. Bit *NOT* to use the HSJ set preferred command because it cannot? > be overwridden by a host "Prefer" or IO$_SETPRFPATH modifier.c >uD > Configuration 3 (Figure 9-3) has the A Paths going through on starG > coupler, and the B Paths going through the second coupler. This looks*H > like there is no improvement in I/O performance. But there seems to be > more availability. >* > E > Configuration 4 (Figure 9-4) splits the disks and HSJs in half. I'mgJ > going to stay away from this one, because to get availability, I'll need> > to do host-based shadowing, and that'll take away some of my > performance. >t@ > So it looks like configuration 2 or 3 is to be the one to use. >RG > Does anyone have experience with either of these configurations? Each*G > CIPCA and HSJ has it's own ID. Do the IDs still have to be unique? Or G > are the numbers duplicated? For example, the the CIPCAs in Host 1, In E > config 2, do they have the same ID? I'm sure they must be unique inl > config 3." >eI > I have questions: Is the fail-over still an issue? Is the documentation=I > regarding the HSJ SET_PREFER command still valid? There doesn't seem top< > be any I/O performance difference between the two configs. >a" > Next, as for the HSJs and disks: >eI > I'm using HSJ50s and 8 bit I/O personality modules. But "VW" drives. Am H > I shakling my I/O throughput by using the 8 bit I/O modules? Would the > 16 bit I/O module be better. > ? > I'm thinking about replacing HSJ50s (considering they're neareH > end-of-life) with HSJ80s. Would I then "Have to" upgrade to the 16 bit > I/O modules. >aB > Disk-drive layout: I need both I/O speed and availabilty. So I'mF > thinking RAID 0+1. Since there are six (6) buses on a given HSJ. I'mG > thinking of using 6 physical disk per storage unit. 3 members striped H > and 3 more mirroring. Laying them out so that each physical is on it's
 > own bus. > F > Considering the discussion regarding pagefiles/disk loss. Would I beJ > better off not having locally hosted SCSI disks for page files, and moveE > the page files onto CI based storage? I'm reluctant to do this. I'mw? > thinking I may loose paging performance.... But I'm not sure.r >r0 > Any advice, thoughts, experiences are welcome. >r > Thanks in advance, >i > Lyndon >  > --I > My opinions are mine and mine alone. They seldom align with those of my= > employer.    ------------------------------    Date: 09 Apr 2001 16:52:39 +0800, From: Paul Repacholi <prep@prep.synonet.com>4 Subject: Re: CI based cluster for max I/O perf. How?- Message-ID: <87g0fia208.fsf@prep.synonet.com>   - young_r@encompasserve.org (Rob Young) writes:=  D >      cache.  Finally, you will need to go to two stars with HSJ80sB >      as they will saturate a single star or so I have been told.3 >      Also, they have TWO sets of ports (A and B).91                        ^^^^^^^^^^^^^^^^^^^^^^^^^^n  A For clarification Rob, By the above I assume you mean two sets ofeA dual-ports, ie 4 CI cables. 2 A cables and 2 B cables, to use then traditional A-B names.   -- u< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.n@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  $ Date: Mon, 9 Apr 2001 09:41:23 +0200$ From: "Jakob Erber" <erberj@post.ch> Subject: Corba on OpenVMS AXP  Message-ID: <3ad167a4$1@hcwe67>o  K Despite all signs that OpenVMS will be there for a long time to come, theretG is still support missing for software, I find vital for an important OSF
 plattform.  I For example, as far as I can see, there is no Corba product available fora1 VMS, which offers proffessionell support as well.O  J There are some very good freeware Orbs, but can you use them for a mission critical system?   best regards   Jakobi   ------------------------------  % Date: Mon, 09 Apr 2001 11:13:21 +0200 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>s! Subject: Re: Corba on OpenVMS AXPe) Message-ID: <3AD17D31.30FFCAE5@gtech.com>e   Jakob Erber wrote:M > Despite all signs that OpenVMS will be there for a long time to come, there I > is still support missing for software, I find vital for an important OS  > plattform. > K > For example, as far as I can see, there is no Corba product available forn3 > VMS, which offers proffessionell support as well.y > L > There are some very good freeware Orbs, but can you use them for a mission > critical system?  " Depends on the specific freeware !  B Some freewre products like Apache and GCC are used in many mission critical	 contexts.m   Arne   ------------------------------  # Date: Mon, 09 Apr 2001 14:34:42 GMTs2 From: seibel_r@localhost.localdomain (Rich Seibel)! Subject: Re: Corba on OpenVMS AXPt; Message-ID: <slrn9d3i3q.fqt.seibel_r@localhost.localdomain>   F On Mon, 9 Apr 2001 09:41:23 +0200, Jakob Erber <erberj@post.ch> wrote:L >Despite all signs that OpenVMS will be there for a long time to come, thereH >is still support missing for software, I find vital for an important OS >plattform.e > J >For example, as far as I can see, there is no Corba product available for2 >VMS, which offers proffessionell support as well. > K >There are some very good freeware Orbs, but can you use them for a missiont >critical system?E >0
 >best regards  >n >Jakob >a >0     -- oD --------------------------------------------------------------------D Rich Seibel, Software Engineer                 (314)579-0066 ext 220D Object Computing, Inc.                           seibel_r@ociweb.comD Need ACE training?                      See http://www.theaceorb.comD --------------------------------------------------------------------   ------------------------------  $ Date: Mon, 9 Apr 2001 04:54:22 -04001 From: Mal Matheson <mal_matheson@MailAndNews.com>t) Subject: RE: DEC Server software questiont& Message-ID: <3B12D466@MailAndNews.com>  M Try Digital Networks at http://www.dnpg.com - this used to be the networking iA division of DEC, and are now an independent company.  They still u7 produce/enhance terminal servers including the DS900TM.    Regards, Mal Matheson  K >===== Original Message From "Patrick Massey" <pmassey@lakecountyohio.org>   =====s >Hello,lI >I've recently installed a DS900TM and found that only the first 16 of 32 M >ports work.  The software it's loading WWENG2 ver 1.1a makes it think it's arK >DS700.  I really need to get the other 16 ports working.  Does anyone knowaL >where I can find an updated loadfile?  I've contacted Compaq and the vendor9 >I ordered the part from, neither have been very helpful.i >Thanks in advance.t >a >-Pat Massey >f   ------------------------------  " Date: Mon, 9 Apr 2001 16:34:03 GMT( From: Terry Kennedy <terry@gate.tmk.com>) Subject: Re: DEC Server software questiono' Message-ID: <GBJA0r.AKs@spcuna.spc.edu>r  3 Mal Matheson <mal_matheson@mailandnews.com> writes:mO > Try Digital Networks at http://www.dnpg.com - this used to be the networking oC > division of DEC, and are now an independent company.  They still t9 > produce/enhance terminal servers including the DS900TM.n  L   I just looked at that site and they don't offer downloadable firmware, andI their SPD's list version 2.2 as the latest, despite Compaq shipping V2.3AlL quite a long time ago (like a year). And some of the links point to .dec.com$ addresses which are no longer valid.  H   So, if they are providing software, etc. they need to fix up their web site.l  4         Terry Kennedy             http://www.tmk.com5         terry@tmk.com             Jersey City, NJ USAc   ------------------------------    Date: 09 Apr 2001 10:15:24 +0200G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>e; Subject: Re: Flipping from big-endian to little endian in CfH Message-ID: <y4elv27alf.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  + rankin@eql.caltech.edu (Pat Rankin) writes:p   >     outbuf[i+1] = inbuf[i];wD >                           000C          movb    (r1)[r0],1(r2)[r0] >     outbuf[i] = inbuf[i+1];aD >                           0012          movb    1(r1)[r0],(r2)[r0]  H I'm pretty sure the VAX FORTRAN compiler would do even better...it wouldC do the equivalent of running the pointer across the arrays, and useeE autoincrement mode for the second expression. The variable i wouldn'tn" exist at all, just the loop count.   	Jan   ------------------------------  % Date: Mon, 09 Apr 2001 14:58:35 -0000 - From: wspencer@ap.nospam.org (Warren Spencer)r; Subject: Re: Flipping from big-endian to little endian in C / Message-ID: <td3jgrfqodki38@news.supernews.com>   C jfmezei.spamnot@videotron.ca (JF Mezei) wrote in <3AD11F76.8D02E7C1  @videotron.ca>:w   >Hoff Hoffman wrote:J >>   Me?  I'd probably use the C htons/htonl and ntohs/ntohl calls, and/or? >>   the available XDR (eXternal Data Representation) library. o > K >help cc run htons  doesn't work. Where do I find info about those run timeo& >library calls ? Are they alpha only ? >o) >And wheere is the on-line help for XDR ?e  F In the C manual, it's in the appendices with other socket-related api  calls.  Also, it's here:  D http://www.openvms.compaq.com/commercial/c/5763p063.htm#htons_socket   ws   -- r1 << Marriage is Grand.  Divorce is Fifty Grand. >>r   Warren Spencer Senior Software Engineer The Associated Press  ? ** My employer does not necessarily agree with my statements **a   ------------------------------  # Date: Mon, 09 Apr 2001 15:36:34 GMTu2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman); Subject: Re: Flipping from big-endian to little endian in Ci2 Message-ID: <6GkA6.672$fB6.18032@news.cpqcorp.net>  _ In article <td3jgrfqodki38@news.supernews.com>, wspencer@ap.nospam.org (Warren Spencer) writes:bD :jfmezei.spamnot@videotron.ca (JF Mezei) wrote in <3AD11F76.8D02E7C1 :@videotron.ca>: :  :>Hoff Hoffman wrote:iK :>>   Me?  I'd probably use the C htons/htonl and ntohs/ntohl calls, and/ort@ :>>   the available XDR (eXternal Data Representation) library.  :># :>help cc run htons  doesn't work. d     Try:       help cc socket htons    < :>Where do I find info about those run time library calls ?  :>Are they alpha only ?e  F   Nope.  The cvt$ call is weird, in that it is in the RTL LIB$ manual.F   The IP-related documentation is split across the C documentation setI   and the IP documentation set -- the core socket APIs are documented in oE   the C manuals, the $qio and most of the other IP programming and IPAH   library (and IP utility) documentation is in the IP documentation set.  * :>And wheere is the on-line help for XDR ? :tG :In the C manual, it's in the appendices with other socket-related api u	 :calls...d  E   Ayup.  XDR and the network API stuff is all kept (for better or foro=   worse) separately from the bog-standard non-socket C stuff.a  N  ---------------------------- #include <rtfaq.h> -----------------------------N       For additional, please see the OpenVMS FAQ -- www.openvms.compaq.com    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  $ Date: Mon, 9 Apr 2001 12:20:15 -04002 From: "John Gemignani, Jr." <john@ossc.DELETE.net>; Subject: Re: Flipping from big-endian to little endian in Cc+ Message-ID: <3ad1e207$1@newsfeed.vitts.com>    > >     outbuf[i+1] = inbuf[i]; F > >                           000C          movb    (r1)[r0],1(r2)[r0] > >     outbuf[i] = inbuf[i+1];tF > >                           0012          movb    1(r1)[r0],(r2)[r0]       Bug.  K     The result would be two bytes of inbuf[i], as you didn't preserve [i+1]      before overwriting it.  	     -Johnt   ------------------------------  % Date: Mon, 09 Apr 2001 10:31:58 -0700a! From: Shane.F.Smith@Healthnet.comu; Subject: Re: Flipping from big-endian to little endian in C D Message-ID: <OFC185C916.6E0371C9-ON88256A29.005FE61D@foundation.com>  I A word of warning. If you're mixing languages, and the descriptor was setoF up by something like BASIC, do NOT change the length or address fieldsK yourself. Call the appropriate STR$ routines instead. C's memory managementoG isn't quite the same as the other 3gl's on VMS, and dicking around with 8 shared descriptors directly can cause a lot of problems.   Shanel          ; jmcmahon38@home.com (Jim McMahon) on 04/07/2001 07:12:35 AMi  3 Please respond to jmcmahon38@home.com (Jim McMahon)b   To:   Info-VAX@Mvb.Saic.Comd cc:s  < Subject:  Re: Flipping from big-endian to little endian in C    . JF Mezei <jfmezei.spamnot@videotron.ca> wrote:  C >I have to read a rather large amount of binary data (2 byte signeds	 integers)t/ >in C. The data is stored as big-endian shorts.h >sG >If I have a entire row of 4800 big-endian signed shorts in consecutiver memory,sG >what would be the most efficient to flip all of them to litte-endian ?p (VAX >architecture, DEC-C). >k >i1 >And while I am at it, a question on descriptors:s >t( >I used to declare descriptors such as :5 >$DESCRIPTOR(my_time_desc,"dd-mmm-yyyy hh:mm:ss.mm");t >wF >and in the program, my_time_desc.dsc$a_pointer was filled with actual data.rA >However, since DEC-C, the default ahs been to made the string inw $DESCRIPTORiH >in non-writable memory, causing access violations when you try to write it.  >i( >I have since change my programs to have >    char my_time_char[25]; - >    $DESCRIPTOP(my_time_desc,my_time_char) ;D >o >Is there a better way ?    F I'm newly back to VAX after a 12 year absence so I don't know if thereE is a better way, but I've recently just had to fix a bug in some code D where the exact same mechanism was used to fill in descriptor values at run-time.  ? The problem:  strings were either truncated or padded with nullb characters.S  A The reason:  original coder didn't update the length field of thedD descriptor structure to account for it's actual resultant length, if4 that was different that the size initially declared.  D Just a heads up if you hadn't thought of it, since I just ran across
 it myself.    6 Being ordinary and nothing special is a full-time job.. jmcmahon38@home.com (Jim McMahon in real life)   ------------------------------   Date: 09 Apr 2001 16:00:37 GMT From: gtirb@aol.com (GTIRB)a Subject: FMS files: Message-ID: <20010409120037.08454.00003453@ng-fe1.aol.com>  K Hi, I have a microvaxII and had a system drive crash. I have since replacedaK this drive and went to restore my backups and to no avail my backups are noeO good. I am in need of the FMS(forms package) Any ideas of where or who I shoulda$ contact to replace these. Thank you.   ------------------------------  % Date: Mon, 09 Apr 2001 17:13:08 +0100 - From: Tim Llewellyn <tim.llewellyn@bbc.co.uk>c Subject: Re: FMS files) Message-ID: <3AD1DF94.4318346C@bbc.co.uk>    GTIRB wrote:  M > Hi, I have a microvaxII and had a system drive crash. I have since replaced M > this drive and went to restore my backups and to no avail my backups are nouQ > good. I am in need of the FMS(forms package) Any ideas of where or who I shoulde& > contact to replace these. Thank you.  R FMS is available on the Condist CD's still. It is no longer supportable, so if you haveR the original licence I doubt there will be any upgrade licence issues if it hasn't beenQ on support. Of course, you don't say what vintage of VMS you are /were running so 6 it might be a problem tracking down condists that old.  P If your backups are totally useless I'd be more worried about where the originalP FMS licence (paper copy) is so I could load it on the rebuilt system. If you can gete: the licence database off the backups that would save time.       --6 Tim Llewellyn, OpenVMS Infrastructure, Remarcs Project0 MedAS at the BBC, Whiteladies Road, Bristol, UK.A Email tim.llewellyn@bbc.co.uk. Home tim.llewellyn@cableinet.co.uko  A I speak for myself only and my views in no way represent those oft MedAS or the BBC.s   ------------------------------  % Date: Mon, 09 Apr 2001 09:31:17 +0100 8 From: John Macallister <J.Macallister1@physics.ox.ac.uk>' Subject: RE: FTP hijacking of VMS sitessN Message-ID: <35666012DF4CD411BE940090279FA240010BEF28@ppnt41.physics.ox.ac.uk>  K If you allow anonymous writing you need to have procedures in place to dealdI with the situation where the disk becomes full. You could report the freeRL space available or a maxmimum permitted file size on login. Bona fide peopleH will work within your restrictions but there will always be the ones who	 won't ...*  J I would be inclined to automatically delete all files or the largest filesJ and/or oldest files when disk space runs out i.e. assuming you really need+ to have an anonymous FTP writing facility. f  $ It's not particularly a VMS problem.   John  B Name: John B. Macallister  E-mail: j.macallister1@physics.ox.ac.ukH Post: Nuclear and Astrophysics Laboratory, Keble Road, Oxford OX1 3RH,UKA Phone: +44-1865-273388 (direct)  273333 (reception)  273418 (Fax)c   ------------------------------  $ Date: Mon, 9 Apr 2001 15:26:40 +1000B From: "Matt Muggeridge" <Matt.Muggeridge@compaq.ssppaammffree.com>? Subject: Re: Galaxy Network Interfaces/cards and TCP/IP Sockets 2 Message-ID: <0SbA6.664$fB6.17565@news.cpqcorp.net>   >a$ > 1) NODE1 (in a cluster with NODE2) >sF > On NODE1 with Alpha VMS 7.2 with two network interfaces 1.2.3.10 andJ > 1.2.3.20, my listening socket program can choose to listen on INADDR_ANY orF > either of the above two address. If NODE1 goes down and a copy of myK > listener program comes up on NODE2 then is there no way for my program to  beL > able to send or receive TCP/IP traffic on 1.2.3.10 or .20? (assuming NODE2  > has one network card 1.2.3.30)  D A couple of solutions come to mind, and since I don't know what yourK application is trying to achieve I can only make general statements.  Firste a quick comment...  G Your client-server application should not be dependent on IP addresses.xH Rather it should use names.  If it did, then you could rely on DNS, LoadK broker and Metric server, to detect the outage of particular interfaces andaK maintain high availability by using a DNS cluster alias (not to be confusedr with an IP cluster alias).  I OK, if you don't like that method (it is my preference) then you could donF some more coding.  Essentially, before the listener comes up on NODE2,D simply create aliases for those interfaces.  E.g. (assuming you have executed TCPIP$DEFINE_COMMANDS)x       ifconfig we0 alias 1.2.3.10s     ifconfig we1 alias 1.2.3.20   # (To delete the alias use "-alias").n   >eE > 2) GALAXY/WILDFIRE VMS 7.3 One box two virtual nodes (ie a two node- cluster  > in a box)r >0L > Is it correct that the cards are stuck in the box and available to all hot  > swapable CPU to node thingies? >?  H You will have to read the manual describing GALAXY in the Wildfire.  I'mL familiar with Galaxy in an ES40, and in that case the top 4 PCI slots belongJ to instance0 and the bottom 6 slots belong to instance1.  I don't know how# the Wildfire splits up the PCI bus.s  K > Or put another way, are the Network Cards served/on a bus/controllered so H > that I can have a copy of my server program listening on both internal nodeseI > at either address. For example, if NODE1 whent down when my program was.H > listening at 1.2.3.10 and it was rerun automatically on NODE2 could it+ > listen and tranceive at the same address?l  L No.  The network card is assigned to a particular instance.  There is no way= of sharing it.  So you will need a network card per instance.-  L It sounds like your trying to achieve high-availability.  If so, take a lookL at Load Broker and Metric Server.  You will need a DNS server which supports dynamic updates too.   Matt.S   ------------------------------   Date: 9 Apr 2001 12:58:10 GMTs From: Gabriel_hogan@hotmail.como3 Subject: Help needed with RMS Character Field Index * Message-ID: <9asbl2$2uc$1@news.netmar.com>   Hi,oN 	I have a sequential RMS file of fixed record length. I want to index the fileL on a character field (third field, length 30, contains surnames). To do thisM I need to extract the character field, reverse it and insert the new field at1: the beginning of each record before indexing it with FDL.   M Has anyone had to do this before and if so, is there a VMS tool to do this orF how did you go about it? r   thx, Gabrielp  O  -----  Posted via NewsOne.Net: Free (anonymous) Usenet News via the Web  -----lM   http://newsone.net/ -- Free reading and anonymous posting to 60,000+ groups I    NewsOne.Net prohibits users from posting spam.  If this or other posts L made through NewsOne.Net violate posting guidelines, email abuse@newsone.net   ------------------------------  % Date: Mon, 09 Apr 2001 16:12:28 +0200s$ From: Michael Unger <unger@decus.de>3 Subject: Help needed with RMS Character Field Indexr* Message-ID: <009FA498.5669FB3A.7@decus.de>  $ > Date: Mon, 9 Apr 2001 12:58:10 GMT! > From: Gabriel_hogan@HOTMAIL.COM45 > Subject: Help needed with RMS Character Field Indexe >    > Hi,RW >         I have a sequential RMS file of fixed record length. I want to index the file N > on a character field (third field, length 30, contains surnames). To do thisO > I need to extract the character field, reverse it and insert the new field atu=                                          ^^^^^^^^^^ ... why ?p; > the beginning of each record before indexing it with FDL.o  C What about specifying "KEY n DATA TYPE ... DSTRING" in the FDL file L ("descending character string") ? I didn't use it myself so I can only guessE but from the HELP included in the FDL editor it seems to fulfill your0
 requirements.-  I (Use the "invoke" option of the FDL editor and you will be asked a lot of,< questions but the result will be syntactically [?] correct.)   >   O > Has anyone had to do this before and if so, is there a VMS tool to do this or  > how did you go about it?   Or did I miss something ?-   Michael-   ------------------------------  % Date: Mon, 09 Apr 2001 11:45:21 +0100-0 From: andrew harrison <andrew.nospam@uk.sun.com>B Subject: Re: Interesting news from those nice folk at The Register* Message-ID: <3AD192C1.1C75AB1D@uk.sun.com>   "Richard D. Piccard" wrote:  > M > In part, the article says:  "To get around the issue Sun is issuing a patch,L > that turns off the prefetch pipeline, a feature of the UltraSPARC III chipL > which tries to retrieve [the] instruction it is most likely to need next." > H > The only conclusion that seems tenable is that ALL publically releasedL > benchmarks must be re-done, and that the results will be worse.  Does SPECK > have regulations that require the withdrawal of benchmarks that cannot be $ > reproduced with shipping products? >   C Not all publically released benchmarks. The 6800 benchmark results lA for SPECJBB, Oracle Apps etc do not need to be rerun because they # the prefetch patch was in for them.a  C I assume that we will be re-running the origional Blade SPECint and-E SPECfp benchmarks though it is unlikely to have an impact in SPECint..     Regardsj Andrew Harrisone Enterprise IT Architects   ------------------------------  $ Date: Mon, 9 Apr 2001 09:36:21 +0200> From: "Jean-Francois Marchal" <jean-francois.marchal@x9000.fr>0 Subject: Re: LINK-I-DATMISMCH since some patch ?. Message-ID: <9arojs$itr$1@reader1.imaginet.fr>   thanks a lot to all ...   D "Peter LANGSTOEGER" <eplan@kapsch.net> a crit dans le message news: 3ad0ce97$1@news.kapsch.co.at...eH > In article <9ajpeu$4dg$1@reader1.imaginet.fr>, "Jean-Francois Marchal"( <jean-francois.marchal@x9000.fr> writes:. > >I'm using 7.2-1 with the latest patches ... > > 0 > >Hadn't have to link new apps for a long time. > >Now for a RDB installation,# > >I get the following messages ...l > > 8 > >%LINK-I-DATMISMCH, creation date of 31-AUG-2000 14:457 > >in shareable image SYS$COMMON:[SYSLIB]DECC$SHR.EXE;1-) > >differs from date of 29-DEC-1999 03:560= > >in shareable image library VMI$ROOT:[SYSLIB]IMAGELIB.OLB;1r8 > >%LINK-I-DATMISMCH, creation date of 26-OCT-2000 00:495 > >in shareable image SYS$COMMON:[SYSLIB]LIBRTL.EXE;1.) > >differs from date of 29-DEC-1999 03:53m= > >in shareable image library VMI$ROOT:[SYSLIB]IMAGELIB.OLB;1u > >i7 > >Same messages occur for all languages I tested with. - > >I updated ALL languages from the MARCH CD.  > >[8 > >As these are information, I shouldn't be worried, BUT< > >RDB has declared my installation failed, probably because2 > >the IVP output does not match the expected one, > >due to embedded messages. > >r8 > >Would anybody know what (patch ?) has caused the date. > >mismatch, and what should I do to fix this? >eI > It is the ACRT ECO and the problem is listed in the readme of this ECO.  >1> > $ LIBRARY /REPLACE SYS$SHARE:IMAGELIB SYS$SHARE:DECC$SHR.EXE >uG > should fix it (though the readme mentions a SYS$SHARE:DECCSHR.EXE ;-)t >g > --> > Peter "EPLAN" LANGSTOEGER           Tel.    +43 1 81111-2651= > Network and OpenVMS system manager  Fax.    +43 1 81111-888f> > <<< KAPSCH AG  Wagenseilgasse 1     E-mail  eplan@kapsch.netJ > A-1121 VIENNA  AUSTRIA              "I'm not a pessimist, I'm a realist"   ------------------------------  % Date: Mon, 09 Apr 2001 10:54:11 +0100o% From: Alan Greig <a.greig@virgin.net>e' Subject: More VMS ads spotted in the UKe8 Message-ID: <jg03dtg4f0k97r2okgu49p48um8kpcmunv@4ax.com>   They are still coming...  D In the latest issue of Computing dated 5th April the VMS AlphaserverD ad appears in one of the most prominent possible positions taking upE the whole of Page 2. Page 1 is a newspaper style front page so Page 2n4 is the inside front cover. You can't really miss it.  D So that's it now appeared in all three of the biggest circulation UKE professional computing weekly magazines. Now would be a good time fordE Compaq to take advantage by sending out some updates and links to VMSlF future presentations. Although there isn't supposed to be a connectionC there always is one. Magazines are more likely to regurgitate press(? releases as news if the company is also buying ad space for the)F product. Both the upcoming VMS technical update days in London and the< DII COE project should be worth of inclusion by any of these
 publications.o -- Alan   ------------------------------  % Date: Mon, 09 Apr 2001 14:14:15 +0100u- From: Tim Llewellyn <tim.llewellyn@bbc.co.uk>c+ Subject: Re: More VMS ads spotted in the UKo( Message-ID: <3AD1B5A7.5C87800@bbc.co.uk>  ? The new "Alpha Powered Special Offers" glossy in the UK is alsos? a breath of fresh air, equal(ish) parity givien to VMS and T64, F plus not a menhtion of the M$ work anywhere. I like it. Can't quote it all but:  C "In an industry full of short term technology, OpenVMS is a durablee solutionI that has done more than stand the test of time, it's grown in stature ande performance.   Alan Greig wrote:    > They are still coming... >h     --6 Tim Llewellyn, OpenVMS Infrastructure, Remarcs Project0 MedAS at the BBC, Whiteladies Road, Bristol, UK.A Email tim.llewellyn@bbc.co.uk. Home tim.llewellyn@cableinet.co.uke  A I speak for myself only and my views in no way represent those ofi MedAS or the BBC.-   ------------------------------  % Date: Mon, 09 Apr 2001 12:38:52 -0300Q) From: fabio_compaq@ep-bc.petrobras.com.br-1 Subject: NSLOOKUP: *** Can't initialize resolver.-L Message-ID: <OF0B05895F.D0D2519D-ON03256A29.005522AD@ep-bc.petrobras.com.br>   People  H Any idea why it is appering this message. My NAME SERVICE is Started and Enable   TCPIP SERVICE V5.0A    RegardsD   FC   ------------------------------  % Date: Mon, 09 Apr 2001 17:08:21 +0100-  From: steven.reece@quintiles.com5 Subject: Re: NSLOOKUP: *** Can't initialize resolver. H Message-ID: <OFC10FFA48.43B1C9B2-ON80256A29.00586DBB@qedi.quintiles.com>  J Is your DNS still there and still accessible?  Can you telnet/ping it from your VMS system?K I'd initially suspect (without any better information) that your VMS systemiF can't get through to the DNS to do a lookup.  This could be because ofC configuration problems, transport problems, DNS crashed due to bitst* flipping in cache or cosmic radiation etc.  1 Has it ever worked?  If yes, then what's changed?8 Steve.   Fabio asked:K >>>Any idea why it is appering this message. My NAME SERVICE is Started anda Enable   TCPIP SERVICE V5.0Ab <<<e   ------------------------------  % Date: Mon, 09 Apr 2001 17:08:30 +0100l- From: Tim Llewellyn <tim.llewellyn@bbc.co.uk>o5 Subject: Re: NSLOOKUP: *** Can't initialize resolver. ) Message-ID: <3AD1DE7E.FE6CA373@bbc.co.uk>s  * fabio_compaq@ep-bc.petrobras.com.br wrote:   > People > J > Any idea why it is appering this message. My NAME SERVICE is Started and > Enable >m  J Is is defined and started /SYSTEM? Did you have all privs enabled when youI declared it? If not, then it will appear to be declared properly but WILLe	 NOT WORK.@  6 If that doesn't help, post output from TCPIP SHOW NAME     >b > TCPIP SERVICE V5.0Ae > 	 > Regards  >  > FC   --6 Tim Llewellyn, OpenVMS Infrastructure, Remarcs Project0 MedAS at the BBC, Whiteladies Road, Bristol, UK.A Email tim.llewellyn@bbc.co.uk. Home tim.llewellyn@cableinet.co.uke  A I speak for myself only and my views in no way represent those ofe MedAS or the BBC.t   ------------------------------  % Date: Mon, 09 Apr 2001 13:25:38 -0300I) From: fabio_compaq@ep-bc.petrobras.com.br)5 Subject: Re: NSLOOKUP: *** Can't initialize resolver. L Message-ID: <OF841D704A.ACCD58A5-ON03256A29.005989F2@ep-bc.petrobras.com.br>  K I never tested the nslookup before in this machine.  The ping works but thel NSLOOKUP is not .....w  K Now I STOPPED the NAME SERVICE and I am having troubles reconfiguring it  .r  .  I The message of Local Domain *Mismatch* is displayed when I check the NAMEy SERVICE.    ; And how do I restart the BIND Resolver after shutdown it !?e   Regardst   FC        1 steven.reece@quintiles.com em 09/04/2001 13:08:21h  , Favor responder a steven.reece@quintiles.com             Info-VAX@Mvb.Saic.Com       5 Assunto: Re: NSLOOKUP: *** Can't initialize resolver.l        J Is your DNS still there and still accessible?  Can you telnet/ping it from your VMS system?K I'd initially suspect (without any better information) that your VMS systemaF can't get through to the DNS to do a lookup.  This could be because ofC configuration problems, transport problems, DNS crashed due to bitst* flipping in cache or cosmic radiation etc.  1 Has it ever worked?  If yes, then what's changed?- Steve.   Fabio asked:K >>>Any idea why it is appering this message. My NAME SERVICE is Started and  Enable   TCPIP SERVICE V5.0A5 <<<0   ------------------------------  $ Date: Mon, 9 Apr 2001 15:02:06 +0800! From: "BB" <Trishgoo@hotmail.com>i6 Subject: OpenVMS Job seeking in Hong Kong, SAR, China.1 Message-ID: <9armpl$e21$1@mail.hk.iasiaworks.com>a  	 Dear All,e  I I am an OpenVMS administrator for about 4 years and managed a vendor core-K application. Due to the application has been replaced by other that runninglK on IBM platform. So I would like to find job here because there is not muchDL recruitment agency has job offer related to this retiring O/S and I think to< post an ad. in this special group might draw your attention.  H I am willingness to give you further info. about my work history and....5 just to say thanks here if you can offer me a chance.r   Regards, Mr. L.   ------------------------------   Date: 9 Apr 2001 14:33:25 GMTs% From: andrew.rycroft@intrinsitech.coml+ Subject: OpenVMS Times T4 tool availabilityu* Message-ID: <9ash7l$bi0$1@news.netmar.com>   Hi,a  M The latest OpenVMS Times refers to a T4 tool ( Tabular Timeline Tracking TooleM ). It sounds great, but it does not say if it is available, or how you obtain  it. Anyone know ?    Thanks Andrew    O  -----  Posted via NewsOne.Net: Free (anonymous) Usenet News via the Web  ----- M   http://newsone.net/ -- Free reading and anonymous posting to 60,000+ groups I    NewsOne.Net prohibits users from posting spam.  If this or other postsPL made through NewsOne.Net violate posting guidelines, email abuse@newsone.net   ------------------------------   Date: 8 Apr 2001 23:40:50 -0500o+ From: young_r@encompasserve.org (Rob Young)t$ Subject: Re: OT: "The Puzzle Palace"3 Message-ID: <icrDsHGPCIBX@eisner.encompasserve.org>0  P In article <3AD0B25B.4E63E758@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes: > All: > F > After a few folks in c.o.v. raved about the book by James Bamford, I	 > orderedc > a copy from Amazon.o > I > It's not quite what I expected.  I was looking more for the techie siden > ofC > what the NSA can do as opposed to the biographical stories of ther	 > various % > folks who held the DIRNSA position.  >    [snip]   > F > Overall, I was pretty disappointed by the book.  I guess ex-military > people who areG > used to seeing acronyms in every sentance probably got into it but its > was a littleJ > too dry for me.  However, I did enjoy the spy stories at the very end of > the book.. > F > Is there anything more current or more technically relevant that has > been written > about the NSA? >     	His follow-up ships April 24th:  e http://www.amazon.com/exec/obidos/ASIN/0385499078/qid=986784522/sr=1-1/ref=sc_b_2/002-3230952-8372858   # 	Buy.com is 3 or 4 dollars cheaper.-  E 	I thought the Puzzle Palace was excellent.  I read it about 10 yearsuG 	ago so little is available for recall.. but I do remember particularlysA 	the Harvester that was so advanced it was in use for many years p> 	(massive tape loader kind of thing).  Backgrounder on how theF 	agency got started, its mission etc.  He obviously isn't able to giveB 	away technique or sizes so much of the techno-geek stuff will not@ 	be in there, but what do you expect?  60 minutes Tuesday a few  	weeks ago had a nicerC 	20 minute or so segment with Dan Rather.  A "first" inside peek ateB 	what goes on at the No Such Agency.  An admission by the directorE 	that they are falling behind technologically.  The director admittedo? 	that a Ben Laden (others like him) with cellphone and readily nD 	available crypto technology (PGP, etc.) are difficult to track and  	determine what they are up to.i    I > On a related note:  This month's issue of Popular Mechanics has a coverI
 > story aboutyI > how "Our government spies on you".  I think I learned as much about theo > technicaliG > abilities of the NSA from the Popular Mechanics article as I did fromn > reading the 500+ pages > of Bamford's book. >   A 	Hmmmm.  Much of the technical abilities of the NSA are shrouded d? 	in mystery.  Things of national security and whatnot are stillo= 	very crucial to trying to get a leg up on others set againsts
 	our country.    				RobP   ------------------------------   Date: 8 Apr 2001 23:40:50 -0500g+ From: young_r@encompasserve.org (Rob Young)u$ Subject: Re: OT: "The Puzzle Palace"3 Message-ID: <icrDsHGPCIBX@eisner.encompasserve.org>-  P In article <3AD0B25B.4E63E758@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes: > All: > F > After a few folks in c.o.v. raved about the book by James Bamford, I	 > orderedl > a copy from Amazon.h > I > It's not quite what I expected.  I was looking more for the techie sideC > ofC > what the NSA can do as opposed to the biographical stories of the 	 > varioush% > folks who held the DIRNSA position.s >    [snip]   > F > Overall, I was pretty disappointed by the book.  I guess ex-military > people who areG > used to seeing acronyms in every sentance probably got into it but it. > was a littleJ > too dry for me.  However, I did enjoy the spy stories at the very end of > the book.c > F > Is there anything more current or more technically relevant that has > been written > about the NSA? >     	His follow-up ships April 24th:  e http://www.amazon.com/exec/obidos/ASIN/0385499078/qid=986784522/sr=1-1/ref=sc_b_2/002-3230952-8372858i  # 	Buy.com is 3 or 4 dollars cheaper.t  E 	I thought the Puzzle Palace was excellent.  I read it about 10 yearssG 	ago so little is available for recall.. but I do remember particularlyeA 	the Harvester that was so advanced it was in use for many years c> 	(massive tape loader kind of thing).  Backgrounder on how theF 	agency got started, its mission etc.  He obviously isn't able to giveB 	away technique or sizes so much of the techno-geek stuff will not@ 	be in there, but what do you expect?  60 minutes Tuesday a few  	weeks ago had a nicelC 	20 minute or so segment with Dan Rather.  A "first" inside peek atoB 	what goes on at the No Such Agency.  An admission by the directorE 	that they are falling behind technologically.  The director admitted?? 	that a Ben Laden (others like him) with cellphone and readily nD 	available crypto technology (PGP, etc.) are difficult to track and  	determine what they are up to.6    I > On a related note:  This month's issue of Popular Mechanics has a covert
 > story abouteI > how "Our government spies on you".  I think I learned as much about the  > technical-G > abilities of the NSA from the Popular Mechanics article as I did from  > reading the 500+ pages > of Bamford's book. >   A 	Hmmmm.  Much of the technical abilities of the NSA are shrouded  ? 	in mystery.  Things of national security and whatnot are stillM= 	very crucial to trying to get a leg up on others set againste
 	our country.    				Rob    ------------------------------   Date: 8 Apr 2001 23:40:50 -0500e+ From: young_r@encompasserve.org (Rob Young)l$ Subject: Re: OT: "The Puzzle Palace"3 Message-ID: <icrDsHGPCIBX@eisner.encompasserve.org>i  P In article <3AD0B25B.4E63E758@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes: > All: > F > After a few folks in c.o.v. raved about the book by James Bamford, I	 > orderedn > a copy from Amazon.D > I > It's not quite what I expected.  I was looking more for the techie sidee > ofC > what the NSA can do as opposed to the biographical stories of them	 > variousl% > folks who held the DIRNSA position.o >    [snip]   > F > Overall, I was pretty disappointed by the book.  I guess ex-military > people who areG > used to seeing acronyms in every sentance probably got into it but ite > was a littleJ > too dry for me.  However, I did enjoy the spy stories at the very end of > the book.- > F > Is there anything more current or more technically relevant that has > been written > about the NSA? >     	His follow-up ships April 24th:  e http://www.amazon.com/exec/obidos/ASIN/0385499078/qid=986784522/sr=1-1/ref=sc_b_2/002-3230952-8372858w  # 	Buy.com is 3 or 4 dollars cheaper.:  E 	I thought the Puzzle Palace was excellent.  I read it about 10 yearspG 	ago so little is available for recall.. but I do remember particularlyrA 	the Harvester that was so advanced it was in use for many years h> 	(massive tape loader kind of thing).  Backgrounder on how theF 	agency got started, its mission etc.  He obviously isn't able to giveB 	away technique or sizes so much of the techno-geek stuff will not@ 	be in there, but what do you expect?  60 minutes Tuesday a few  	weeks ago had a niceaC 	20 minute or so segment with Dan Rather.  A "first" inside peek atsB 	what goes on at the No Such Agency.  An admission by the directorE 	that they are falling behind technologically.  The director admitted ? 	that a Ben Laden (others like him) with cellphone and readily iD 	available crypto technology (PGP, etc.) are difficult to track and  	determine what they are up to.-    I > On a related note:  This month's issue of Popular Mechanics has a coverj
 > story aboutrI > how "Our government spies on you".  I think I learned as much about theD > technical$G > abilities of the NSA from the Popular Mechanics article as I did from  > reading the 500+ pages > of Bamford's book. >   A 	Hmmmm.  Much of the technical abilities of the NSA are shrouded a? 	in mystery.  Things of national security and whatnot are stillf= 	very crucial to trying to get a leg up on others set against 
 	our country.e   				Robw   ------------------------------   Date: 8 Apr 2001 23:40:50 -0500f+ From: young_r@encompasserve.org (Rob Young)I$ Subject: Re: OT: "The Puzzle Palace"3 Message-ID: <icrDsHGPCIBX@eisner.encompasserve.org>.  P In article <3AD0B25B.4E63E758@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes: > All: > F > After a few folks in c.o.v. raved about the book by James Bamford, I	 > ordered  > a copy from Amazon.M > I > It's not quite what I expected.  I was looking more for the techie sider > ofC > what the NSA can do as opposed to the biographical stories of thes	 > variousa% > folks who held the DIRNSA position.t >    [snip]   > F > Overall, I was pretty disappointed by the book.  I guess ex-military > people who areG > used to seeing acronyms in every sentance probably got into it but ito > was a littleJ > too dry for me.  However, I did enjoy the spy stories at the very end of > the book.h > F > Is there anything more current or more technically relevant that has > been written > about the NSA? >     	His follow-up ships April 24th:  e http://www.amazon.com/exec/obidos/ASIN/0385499078/qid=986784522/sr=1-1/ref=sc_b_2/002-3230952-8372858m  # 	Buy.com is 3 or 4 dollars cheaper.g  E 	I thought the Puzzle Palace was excellent.  I read it about 10 years G 	ago so little is available for recall.. but I do remember particularlyHA 	the Harvester that was so advanced it was in use for many years  > 	(massive tape loader kind of thing).  Backgrounder on how theF 	agency got started, its mission etc.  He obviously isn't able to giveB 	away technique or sizes so much of the techno-geek stuff will not@ 	be in there, but what do you expect?  60 minutes Tuesday a few  	weeks ago had a nicecC 	20 minute or so segment with Dan Rather.  A "first" inside peek atAB 	what goes on at the No Such Agency.  An admission by the directorE 	that they are falling behind technologically.  The director admittedg? 	that a Ben Laden (others like him) with cellphone and readily .D 	available crypto technology (PGP, etc.) are difficult to track and  	determine what they are up to.     I > On a related note:  This month's issue of Popular Mechanics has a coveru
 > story aboutaI > how "Our government spies on you".  I think I learned as much about the  > technical G > abilities of the NSA from the Popular Mechanics article as I did frome > reading the 500+ pages > of Bamford's book. >   A 	Hmmmm.  Much of the technical abilities of the NSA are shrouded d? 	in mystery.  Things of national security and whatnot are stillf= 	very crucial to trying to get a leg up on others set against-
 	our country.,   				Rob1   ------------------------------   Date: 8 Apr 2001 23:40:50 -0500j+ From: young_r@encompasserve.org (Rob Young)s$ Subject: Re: OT: "The Puzzle Palace"3 Message-ID: <icrDsHGPCIBX@eisner.encompasserve.org>e  P In article <3AD0B25B.4E63E758@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes: > All: > F > After a few folks in c.o.v. raved about the book by James Bamford, I	 > ordered  > a copy from Amazon.n > I > It's not quite what I expected.  I was looking more for the techie sideo > ofC > what the NSA can do as opposed to the biographical stories of the 	 > variousi% > folks who held the DIRNSA position., >    [snip]   > F > Overall, I was pretty disappointed by the book.  I guess ex-military > people who areG > used to seeing acronyms in every sentance probably got into it but itl > was a littleJ > too dry for me.  However, I did enjoy the spy stories at the very end of > the book.e > F > Is there anything more current or more technically relevant that has > been written > about the NSA? >     	His follow-up ships April 24th:  e http://www.amazon.com/exec/obidos/ASIN/0385499078/qid=986784522/sr=1-1/ref=sc_b_2/002-3230952-8372858m  # 	Buy.com is 3 or 4 dollars cheaper.   E 	I thought the Puzzle Palace was excellent.  I read it about 10 years-G 	ago so little is available for recall.. but I do remember particularlyiA 	the Harvester that was so advanced it was in use for many years  > 	(massive tape loader kind of thing).  Backgrounder on how theF 	agency got started, its mission etc.  He obviously isn't able to giveB 	away technique or sizes so much of the techno-geek stuff will not@ 	be in there, but what do you expect?  60 minutes Tuesday a few  	weeks ago had a nice C 	20 minute or so segment with Dan Rather.  A "first" inside peek at B 	what goes on at the No Such Agency.  An admission by the directorE 	that they are falling behind technologically.  The director admitted ? 	that a Ben Laden (others like him) with cellphone and readily ID 	available crypto technology (PGP, etc.) are difficult to track and  	determine what they are up to.,    I > On a related note:  This month's issue of Popular Mechanics has a coveru
 > story aboutOI > how "Our government spies on you".  I think I learned as much about the. > technical_G > abilities of the NSA from the Popular Mechanics article as I did froms > reading the 500+ pages > of Bamford's book. >   A 	Hmmmm.  Much of the technical abilities of the NSA are shrouded l? 	in mystery.  Things of national security and whatnot are stillr= 	very crucial to trying to get a leg up on others set against'
 	our country.t   				RobP   ------------------------------  % Date: Mon, 09 Apr 2001 11:15:13 +0200-= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> $ Subject: Re: OT: "The Puzzle Palace") Message-ID: <3AD17DA1.8CF8A892@gtech.com>i   Scott Vieth wrote:F > After a few folks in c.o.v. raved about the book by James Bamford, I	 > ordered  > a copy from Amazon.- > I > It's not quite what I expected.  I was looking more for the techie sideo > ofC > what the NSA can do as opposed to the biographical stories of thee	 > variouse% > folks who held the DIRNSA position.2  6 I think too detailed and too uptodate info about NSA's? technical side will reward the author with 50 years in prison !P   :-)    Arne   ------------------------------   Date: 9 Apr 2001 13:26:34 GMTr1 From: bill@triangle.cs.uofs.edu (Bill Gunshannon)a$ Subject: Re: OT: "The Puzzle Palace", Message-ID: <9asdaa$2r6f$1@info.cs.uofs.edu>  ) In article <3AD0B25B.4E63E758@wi.rr.com>, '  Scott Vieth <svieth@wi.rr.com> writes:e |> 0J |> I thought it was interesting to find out that civilians were subject to |> a more rigorous( |> background check than military folks  |> i  8 Well, considering that my personal experience shows this9 to be patently false, I would then have to wonder if this?9 book is any more than a pipe dream that really belongs ina7 the fiction department.   If he doesn't even know aboutl9 simple administrative procedures, why would anyone expectr# any accuracy in technical details??b   bill6 [I have held clearances as both civilian and military.3  Not only is the exact same DD Form used for eitheri5  application, the background investigation for eitherM5  is done by the same office and in at least two cases 6  of mine, by the same agent who went out and asked the!  same people the same questions.]m   -- <J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   t   ------------------------------  % Date: Mon, 09 Apr 2001 09:43:41 -0500.+ From: Christopher Smith <csmith@amdocs.com>r$ Subject: RE: OT: "The Puzzle Palace"L Message-ID: <3B55D7F383B0D31197D9009027541CBF0D9D1D2E@cmiexch1.cmi.itds.com>   > -----Original Message-----D > From: young_r@encompasserve.org [mailto:young_r@encompasserve.org]  A 	Hmmmm.  Much of the technical abilities of the NSA are shrouded  ? 	in mystery.  Things of national security and whatnot are stillp= 	very crucial to trying to get a leg up on others set againstu
 	our country.T  I Well said, but somewhat incorrect, IMO.  If it were really "our country,"aL we'd know what was going on with it, don't you think?  It's "their country,"! and they're not willing to share.e   Regards,   Chrisa  ! Christopher Smith, Perl Developere Amdocs - Champaign, IL   /usr/bin/perl -e '? print((~"\x95\xc4\xe3"^"Just Another Perl Hacker.")."\x08!\n");  '    ------------------------------  * Date: Mon, 9 Apr 2001 17:39:08 +0000 (UTC); From: Olivier.Hurez-Martin@ca.com ("Hurez-Martin, Olivier")t3 Subject: Re: Problems with threads on VMS AXP 7.1-2gF Message-ID: <21AAC260C0BCD411AE9B009027AA4DE702C0F48E@usilms03.ca.com>  , > Has anybody experienced something similar?   Yes...  G Prohibited: STARLET.SETAST(FALSE); (even if you think you reenable ASTsrI soon enough...) as that would interfere with the Ada task scheduler which.E uses ASTs itself, probably in conjunction with timer queue entries...   1 Prohibited: STARLET.HIBER(); for obvious reasons.,  K Recommended: Use the TASKING_SERVICES versions of synchronous system calls,IJ to avoid suspending the whole application instead of only the calling task until completion....   with TASKING_SERVICES; use TASKING_SERVICES;e ...SD sts := TASK_QIOW( ...);    -- Instead of   sts := STARLET.QIOW(...); sts := TASK_ENQW( ...);- sts := TASK_GETDVIW( ...);     	-- Olivier.     -- n+ Posted from mail1.cai.com [141.202.248.38]  1 via Mailgate.ORG Server - http://www.Mailgate.ORGn   ------------------------------  # Date: Mon, 09 Apr 2001 13:02:29 GMTh$ From: mark@NOSPAMtechop.co.uk (Mark)( Subject: Re: Q: Unmapping global section/ Message-ID: <3ad1b2b6.19981491@news.force9.net>   @ On Fri, 06 Apr 2001 15:56:25 GMT, mark@NOSPAMtechop.co.uk (Mark) wrote:   >Hi, >-D >I am mapping a global section with sys$mgblsc.  How can I unmap it? >e  7 Thanks for all the replies. sys$deltva is the solution.a   Cheers,e Mark   ------------------------------  # Date: Mon, 09 Apr 2001 14:37:05 GMT  From: sfm1115@bjc.orgi/ Subject: Running a Diagnostic on a Network Cardd1 Message-ID: <3ad1c86e.512320277@news.starnet.net>t  3 I am running an AlphaServer 2100 with OpenVMS 7.2-1e  D Is there an inherited tool inside of the OpenVMS software that I canC use to diagnose the netword card in this server.  I am not seeing asF device errors on the system but I need to make sure.  The server keeps dropping users on and off.   Thanks   Shawne   ------------------------------  % Date: Mon, 09 Apr 2001 00:35:58 -0400i2 From: rdeininger@mindspring.com (Robert Deininger)  Subject: Re: Shadowing WeirdnessL Message-ID: <rdeininger-0904010035580001@user-2ive76n.dialup.mindspring.com>  P In article <sad09e0d.048@aaas.org>, John Eisenschmidt <jeisensc@aaas.org> wrote:    I Please try to turn off the "quoted printable" stuff in your postings.  Wen* already have all the equals signs we need.    K > I'm setting up two new DS10s that are about to go into production. Both =dH > are using Volume Shadowing, and this is the first time anyone at our =& > company has traveled down that path. > I > One system has 14 disks configured as 7 shadow pairs. The other has 4 =o% > disks configured as 2 shadow pairs.u > M > Three or four days ago, on the system with 14 disks, I applied a bunch of =-L > 7.2-1 ECOs to the system, and after I rebooted the shadow pairs were all =M > out of sync, and ended up doing a full shadow copy. Now, the first time I =eI > did that, it took a couple hours. After applying the ECOs, a full day = " > later they were all around 30%.   . > I dismounted the pairs, init'ed the second =M > disk of each pair, mounted them again to do a full shadow merge, and like =i- > the first time it only took a couple hours.   - Shouldn't this have been a copy, not a merge?s  I Don't have any good ideas here.  Were there any hints in the installationSH notes for the patches?  Are these two systems clustered?  If so, did theG second system have the shadow sets mounted while you applied patches onm the first system?   J You may need to do some archaeology in the operator logs and error logs to4 reconstruct the sequence of operations on the disks.   M > On the system with 4 disks, which has been running fine, I ran autogen on =eL > the system because I had installed some layered products and it had been =K > up a couple days. After I rebooted, the sysgen parameters for shadowing =0" > and alloclass had been cleared.   J This sounds like a MODPARAMS.DAT error, or some non-default paramater fileC goings-on.  Look at the autogen report in detail.  Any clues there?   C Most likely, someone previously changed the shadowing and alloclasssJ parameters in the disk file via SYSMAN, but did not make the corresponding> changes in MODPARAMS.DAT.  Autogen follows the instructions inI MODPARAMS.DAT.  It's easy to forget to keep MODPARAMS in sync; that's whyiJ fiddling with parameters via SYSMAN is discouraged, except in very special cases.   > I reset them and rebooted. g  J You reset them _how_?  Your choice of words suggests that you used SYSMAN,F not MODPARAMS + AUTOGEN.  In that case the problem will recur the next time you AUTOGEN.   G Unless you are very sure you understand exactly what's going to change,rG you should run autogen so it tells you what it's changing, but does NOT J make the changes.  If you like what it tells you, run it again to actuallyB change the parameter file.  If you don't like it, no harm is done.  D MODPARAMS gets messy over time, as installations and cluster_configsI append more and more stuff.  Clean it out, get rid of duplicates, and tryd% to understand everything that's left.B   >I was finally =) > able to get the system pair to remount.o > G > $ MOUNT/CONFIRM/SYSTEM DSA0: /SHADOW=3D($56$DKB0,$56$DKC0) ALPHA_72-1o, > %MOUNT-F-SHDWCOPYREQ, shadow copy requiredN > Virtual Unit - _DSA0:                             Volume Label - ALPHA_72-1= > 7 >      Member                    Volume Label Owner UIC_: > %%%%%%%%%%%  OPCOM   8-APR-2001 15:41:51.92  %%%%%%%%%%%G > DSA0: shadow master changed.  Dump WILL be written if system crashes.t > : > %%%%%%%%%%%  OPCOM   8-APR-2001 15:41:51.92  %%%%%%%%%%%3 > Mount verification has completed for device DSA0:  > 3 >      _$56$DKC0: (BARIUM)       ALPHA_72-1   [1,1] 6 > Allow FULL shadow copy on the above member(s)? [N]:Y0 > %MOUNT-I-MOUNTED, ALPHA_72-1 mounted on _DSA0:D > %MOUNT-I-ISAMBR, _$56$DKB0: (BARIUM) is a member of the shadow setK > %MOUNT-I-SHDWMEMCOPY, _$56$DKC0: (BARIUM added to the shadow set with a =r > copy operation=20a > N > I have to say that all these problems worry me. The system with 4 disks is =J > scheduled to go into production this coming week. I have to think that =N > it's something I'm doing wrong, so if someone could offer some suggestions = > I'd be greatful.  E Before your systems go into production, spend some time doing lots of I tests with some of your shadow sets, so you become very familiar with theo whole business./   -- 8 Robert Deininger rdeininger@mindspring.com2   ------------------------------  % Date: Mon, 09 Apr 2001 11:47:01 +0100z" From: Nic P Clews <nclews@csc.com>  Subject: Re: Shadowing Weirdness@ Message-ID: <OF177F4001.39B20B6E-ON80256A29.003B3C96@eu.csc.com>  / You said you applied a number of ECO's to 7.2-1y  + Was one of these SHADOWING 5 by any chance?B  7 Did you specify you wanted the increased functionality?c  8 The release notes contain information that may just calm= your fears, but as already suggested to you, do make yourselfh  familiar with what you now have.  7 The Technical Update seminars being held have a sessiontH which will explain all, it would('ve?) be(en) worth your while attending  4 (I must also say I appreciate the efforts of the VMS: Engineering team to back port this and other features into* earlier versions of the operating system).   Regards, Nic Clews nclews at csc dot comy   ------------------------------    Date: 09 Apr 2001 10:22:00 +0200G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>a Subject: Re: sys$io_performwH Message-ID: <y4bsq67aaf.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  . Paul Repacholi <prep@prep.synonet.com> writes:  H > HAs anyone measured the performance diffs between QIO, OI_PERFORM, andA > paging? I assume the paging has inherited any speedups as well.   G Paging, at least under VMS at the moment, has the disadvantage of beingtI synchronous - that is, there is no euiquivalent to prefetching in paging.c  I As I understand it, the advantage of IO_PERFORM, compared to QIO, is that:J you promise to always use the same region of your process's virtual memoryE for I/O, so memory management can lock those pages into memory, whichgB simplifies I/O startup and completion significantly. Anyhing else?   	Jan   ------------------------------  $ Date: Mon, 9 Apr 2001 09:47:26 +0200- From: "Roland Hauk" <roland.hauk@indramat.de> & Subject: Re: UCX 5.1 BG device problem- Message-ID: <9arp3c$7ra1@sunny.mannesmann.de>r   Please try,t   $ pid=f$getjpi ("", "PID") $! $ sho proc /out='pid'_1.out . $ search 'pid'_1.out "Host:" /out= 'pid'_2.out $ open/read infile 'pid'_2.out $  read infile recordt $  write sys$output record' $  myhost =f$element (3, ":", record  )i' $  myport =f$element (4, ":", record  )t $ close infile $ myhost = myhost - "Port" $ myport = myport - ")"n $! $ define sys$output 'pid'_3.outf2 $       tcpip sh dev /host='myhost' /port='myport' $ deassign sys$outputl, $ search 'pid'_3.out TELNET /out='pid'_4.out $! $ open/read infile 'pid'_4.out $ read infile record( $       bg =f$element (2, " ", record  ) $ close infile $! $ sho sym bg $ delete 'pid'*.out*;* $ tcpip disc dev 'bg': $ logout     -- bye,5 Roland Hauk        ( Email: roland.hauk@indramat.de ) , Rexroth Indramat  ( http://www.indramat.de )   ------------------------------  $ Date: Mon, 9 Apr 2001 12:03:18 +0200- From: "Roland Hauk" <roland.hauk@indramat.de>i& Subject: Re: UCX 5.1 BG device problem- Message-ID: <9as1ln$7ri4@sunny.mannesmann.de>u   or try thist  ( $ RECORD=f$getdvi ("TT", "TT_ACCPORNAM"), $       myhost =f$element (1, ":", record  ), $       myport =f$element (2, ":", record  ) $ myhost = myhost - "Port" $ myport = myport - ")"t $! $ define sys$output 'pid'_1.oute2 $       tcpip sh dev /host='myhost' /port='myport' $ deassign sys$output , $ search 'pid'_1.out TELNET /out='pid'_2.out $! $ open/read infile 'pid'_2.out $ read infile record( $       bg =f$element (2, " ", record  ) $ close infile $!
 $! sho sym bge $ delete 'pid'*.out;*  $ tcpip disc dev 'bg': $ logout     -- ...r5 Roland Hauk        ( Email: roland.hauk@indramat.de )a, Rexroth Indramat  ( http://www.indramat.de )   ------------------------------  $ Date: Mon, 9 Apr 2001 10:24:24 -0400& From: "Syltrem" <syltrem@videotron.ca>/ Subject: Re: VMS equivalent of Unix wc command?+7 Message-ID: <MwjA6.5063$7A3.490678@weber.videotron.net>t  # You get less output with using /LOGd! $ SEARCH file blabla /NOUTPUT/LOGn --   Syltrem   http://pages.infinit.net/syltrem  : <jamese@beast.dtsw.army.mil> a crit dans le message news:% 01040614293904@beast.dtsw.army.mil...- > Greg,  >a5 > "Greg Zymbaluk" <greg.zymbaluk@compaq.com> wrote oneK > Fri, 6 Apr 2001 11:01:14 -0600 in <ADmz6.636$fB6.16968@news.cpqcorp.net>:s >C: > > Is there a VMS equivalent of the Unix wc -l command to > > count the lines in a file? >@F > There are several ways to do this. You can use SEARCH, specify /STAT. > and give it a nonsense string to search for: >y' > $ search login.com "^&&^^%%$$$" /stat G > Files searched:                 1       Buffered I/O count:        13AG > Records searched:             430       Direct I/O count:           9kG > Characters searched:        14804       Page faults:               16eJ > Records matched:                0       Elapsed CPU time:  0 00:00:00.00J > Lines printed:                  0       Elapsed time:      0 00:00:00.14) > %SEARCH-I-NOMATCHES, no strings matchedn > D > Or you can get wc from <ftp://alpha.nuceng.ufl.edu/wc> and use it.@ > There was one posted to this list in 1992 that I can send you. >n< > Ed James                           ed.james@telecomsys.com7 > TeleCommunications Systems, Inc.   voice 410-295-19197= > 2024 West Street, Suite 300              800-810-0827 x1919r7 > Annapolis, MD 21401-3556           fax   410-280-1094e   ------------------------------    Date: 09 Apr 2001 09:53:49 +0200G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>n$ Subject: Re: VMS-Related: AffordableH Message-ID: <y4hezy7ble.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  ) "Bill Todd" <billtodd@foo.mv.com> writes:   J > No matter what XFC does, there's likely to be some minor performance hitK > compared with the Unix approach simply because in making its enhancementsaM > transparent (which I agree is generally a Good Thing) it will result in twokN > data copies (once to/from a process's internal RMS buffer, and a second time> > to/from the XFC's lower-level cache) rather than Unix's one.  A I'm not sure I understand this completely. _Somebody_ must do theaB blocking/deblocking from the application's byte stream to the diskB blocks - I guess that somebody isn't the file system, so it is RMSD or the C RTL. Then these blocks must go to disk - traditionally, VMSD does direct I/O without a system buffer, instead of Unix's approach J to first copy the data to a system buffer and then, if required/requested,A to disk. So I'd say the usual VMS approach saves one memory copy,m+ at the cost of possibly unwanted disk I/Os.S  H Does any system offer double-mapping file blocks into both the process'sG buffer and the system filesytem cache? Accounting would be complicated,o but it should be possible...   	Jan   ------------------------------    Date: 09 Apr 2001 17:03:09 +0800, From: Paul Repacholi <prep@prep.synonet.com>$ Subject: Re: VMS-Related: Affordable- Message-ID: <87bsq6a1iq.fsf@prep.synonet.com>h  I Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de> writes:o  @ > Does any system offer double-mapping file blocks into both theF > process's buffer and the system filesytem cache? Accounting would be+ > complicated, but it should be possible...   : Some unix systems go this way I believe for mmap'ed files.  B One of the side effects of the unix kernel buffers, is that the IO? system has only ever had to deal with kernel addresses. On someiF machines, a context switch nukes the user section of the address space@ (sound like a machine or two we know :) ), so the file is double? mapped to some set of kernel addresses and the IO done to thoserD addresses, and the 'bottom' of the kernel wories about fixing up the user mode mappings.w   -- m< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.c@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------   End of INFO-VAX 2001.199 ************************