1 INFO-VAX	Fri, 14 Oct 2005	Volume 2005 : Issue 573       Contents: Alpha last order date  RE: Alpha last order date  Re: Alpha last order date  Re: Alpha last order date * Re: Anyone have a firefox (6520) keyboard?* Re: Anyone have a firefox (6520) keyboard?0 Re: DECnet phase IV node drops itself. Why? How? File marked for delete Re: File marked for delete Re: File marked for delete Re: File marked for delete Re: File marked for delete Re: File marked for delete Re: File marked for delete Re: File marked for delete Re: File marked for deleteF Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?)F Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?)I Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) I Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) I Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) O Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) ?) ?) P Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) ?) ?) N Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) ?)?) Re: HP : Moving forward - HP Tech Forum - Shameless plug for my session 6 Re: Layoffs at HP : MH needs your suggestions or ideas Re: Quotation marks  Re: Samba-3 for OpenVMS  Re: Samba-3 for OpenVMS P Re: Usage of 3X-DEGXA-TB within AlphaServer 1000 5/400, AlphaStation DS20, Alpha Re: VMS logo feedback needed [Fwd: File marked for delete] ! Re: [Fwd: File marked for delete]   F ----------------------------------------------------------------------  % Date: Thu, 13 Oct 2005 15:26:29 -0400 ' From: Dave Froble <davef@tsoft-inc.com>  Subject: Alpha last order date0 Message-ID: <11ktcreolik8nfc@corp.supernews.com>  5 Just got some stuff from HP, e-mail newsletter stuff.   G VMS was rather prominent, mentioned in multiple locations.  Some stuff  C on Alpha also.  One item was October 27, 2006 as a last order date.   I Some were worried that 2006 meant Jan 1.  Not so.  Got a year to see how  F things work out.  If the itanic survives, that's good for VMS.  If it H has problems, well, they're accepting orders for another year, and last G ship date is probably some reasonable time after that.  Still possible  * at that time to extend Alpha if necessary.  D I'm feeling that in another year, the 'writing on the wall' for the G itanic will be much clearer.  Not saying that it will be bad, could go  - either way.  I'm hoping it finds some future.   I If the newsletter is any indication, HP is not hiding VMS.  That's good.  D   Then again, I may be a targeted reader, getting a special version.    Black helicopters and such.  :-)   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Thu, 13 Oct 2005 18:16:35 -0400 ' From: "Main, Kerry" <Kerry.Main@hp.com> " Subject: RE: Alpha last order dateR Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB70C3B9@tayexc19.americas.cpqcorp.net>   > -----Original Message-----3 > From: Dave Froble [mailto:davef@tsoft-inc.com]=20   > Sent: October 13, 2005 3:26 PM > To: Info-VAX@Mvb.Saic.Com   > Subject: Alpha last order date >=207 > Just got some stuff from HP, e-mail newsletter stuff.  >=20@ > VMS was rather prominent, mentioned in multiple locations. =20 > Some stuff=20 E > on Alpha also.  One item was October 27, 2006 as a last order date.  >=20B > Some were worried that 2006 meant Jan 1.  Not so.  Got a year=20 > to see how=20 J > things work out.  If the itanic survives, that's good for VMS.  If it=20= > has problems, well, they're accepting orders for another=20  > year, and last=20 B > ship date is probably some reasonable time after that.  Still=20
 > possible=20 , > at that time to extend Alpha if necessary. >=20H > I'm feeling that in another year, the 'writing on the wall' for the=20B > itanic will be much clearer.  Not saying that it will be bad,=20
 > could go=20 / > either way.  I'm hoping it finds some future.  >=20@ > If the newsletter is any indication, HP is not hiding VMS. =20 > That's good.=20 F >   Then again, I may be a targeted reader, getting a special version. >=20" > Black helicopters and such.  :-) >=20   Dave,   % Is the newsletter the Customer Times?   - If so, the online version can be accessed at: > http://www.hp.com/products1/evolution/customertimes/index.html   Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  # Date: Fri, 14 Oct 2005 00:49:46 GMT ( From: Alan Greig <greigaln@netscape.net>" Subject: Re: Alpha last order date< Message-ID: <KgD3f.23975$U9.12055@fe3.news.blueyonder.co.uk>   Main, Kerry wrote:  ' > Is the newsletter the Customer Times?  > / > If so, the online version can be accessed at: @ > http://www.hp.com/products1/evolution/customertimes/index.html  I Can this be downloaded as a complete version like the old OpenVMS Times?  G   If you download it all you get is the index pages which all link off  + to other files which need to be downloaded.    --  
 Alan Greig   ------------------------------  % Date: Fri, 14 Oct 2005 03:10:58 +0200 & From: Paul Sture <paul.sture@decus.ch>" Subject: Re: Alpha last order date+ Message-ID: <3r8et3FibtsdU1@individual.net>    Alan Greig wrote:  >  >  > Main, Kerry wrote: > ( >> Is the newsletter the Customer Times? >>0 >> If so, the online version can be accessed at:A >> http://www.hp.com/products1/evolution/customertimes/index.html  >  > K > Can this be downloaded as a complete version like the old OpenVMS Times?  K >  If you download it all you get is the index pages which all link off to  * > other files which need to be downloaded. > C Ouch - see what you mean. 10 urls on the first 2.5 pages, with the  E remaining 6.5 pages full of 1 line subjects with a url (e.g. 29 urls  0 referenced on page 4, to pick a page at random).   ------------------------------  # Date: Fri, 14 Oct 2005 03:23:26 GMT % From: Roger Ivie <rivie@ridgenet.net> 3 Subject: Re: Anyone have a firefox (6520) keyboard? 3 Message-ID: <slrndku95e.7g6.rivie@Stench.no.domain>   ; On 2005-10-12, Doug Phillips <dphill46@netscape.net> wrote:  > Roger Ivie wrote: D >> The 2000 has a DB-25 for the printer, a DB-9 for the console, andF >> a DB-15 for video, keyboard, and mouse. Well, except I think "DB-9"I >> should actually be "DE-9" and I'm not sure what "DB-15" should be (the E >> B refers to the size of the connector, so a DB-9 would be the size % >> of a DB-25, but only have 9 pins).  > I > The 9 pin RS232 connector is a DB-9. The 15 pin (3 rows of 5) connector  > is an HD-15.  H I'm pretty certain you're wrong about the HD-15. The HD-15 appears to beK the VGA connector, which is the same size as an alleged DB-9 but has three  F rows of pins. The 15-pin connector used in the VAXstations is slightly larger with two rows of pins.   E I'm also seeing web pages claiming I was right about the DB-9 being a @ DE-9, but I can't cut and paste between the web session and this; terminal session (different machines) to include a pointer.   A The Amigas used an even slightly larger connector with 23 pins. I @ suspect the Amiga connector would be DC-23, the 15-pin connector@ used in the VAXstation would be DD-15, and the 9-pin serial port= is DE-9. But I've been wrong before, and no doubt will again.  --  
 Roger Ivie rivie@ridgenet.net http://anachronda.webhop.org/  -----BEGIN GEEK CODE BLOCK----- 
 Version: 3.12 H GCS/P d- s:+++ a+ C++ UB--(++++) !P L- !E W++ N++ o-- K w O- M+ V+++ PS+? PE++ Y+ PGP t+ 5+ X-- R tv++ b++ DI+++ D+ G e++ h--- r+++ z+++   ------END GEEK CODE BLOCK------    ------------------------------  # Date: Fri, 14 Oct 2005 03:30:24 GMT % From: Roger Ivie <rivie@ridgenet.net> 3 Subject: Re: Anyone have a firefox (6520) keyboard? 3 Message-ID: <slrndku9ig.7g6.rivie@Stench.no.domain>   5 On 2005-10-14, Roger Ivie <rivie@ridgenet.net> wrote: C > The Amigas used an even slightly larger connector with 23 pins. I B > suspect the Amiga connector would be DC-23, the 15-pin connectorB > used in the VAXstation would be DD-15, and the 9-pin serial port? > is DE-9. But I've been wrong before, and no doubt will again.   @ Ah. I am wrong. The 15 would be a DA-15. C and D are LARGER thanA the DB-25; I think the connector used for external storage in the < VAXstation 2000 (the Leprechaun box) is a D-sized connector.  B I'm not sure what size the 23-pin connector used in the Amiga was. --  
 Roger Ivie rivie@ridgenet.net http://anachronda.webhop.org/  -----BEGIN GEEK CODE BLOCK----- 
 Version: 3.12 H GCS/P d- s:+++ a+ C++ UB--(++++) !P L- !E W++ N++ o-- K w O- M+ V+++ PS+? PE++ Y+ PGP t+ 5+ X-- R tv++ b++ DI+++ D+ G e++ h--- r+++ z+++   ------END GEEK CODE BLOCK------    ------------------------------  # Date: Fri, 30 Sep 2005 08:33:05 GMT A From: "Colin Butcher" <colin_DOT.butcher_AT@xdelta_DOT.co_DOT.uk> 9 Subject: Re: DECnet phase IV node drops itself. Why? How? > Message-ID: <5L6%e.118415$G8.58310@text.news.blueyonder.co.uk>  G http://ftp.digital.com/pub/DEC/DECnet/PhaseIV/ would be a good place to E start learning about how DECnet Phase IV really works under the hood.    --     Hope this helps, Colin. ) colin DOT butcher AT xdelta DOT co DOT uk E It's not mine, but I like this definition: Legacy = stuff that works.    ------------------------------  # Date: Thu, 13 Oct 2005 23:08:29 GMT ( From: Alan Greig <greigaln@netscape.net> Subject: File marked for delete = Message-ID: <NNB3f.112603$RW.45815@fe2.news.blueyonder.co.uk>   G A disk is mounted on 3 cluster nodes - 2 Vax, 1 Alpha. Versions of VMS  @ are VAX 6.2, Alpha 7.2-1. Disk is at zero free blocks due to an E application error. I delete a file around 300000 blocks which is not  F open (or even likely to have been open for 5 years) according to SHOW C DEV/FILE on all nodes. After the DELETE free disk space is still 0  E blocks. I do an ANAL/DISK and it finds the file as marked for delete  G (only warning it finds on the disk). I now do ANAL/DISK/REPAIR/CONFIRM  G and respond Y to fix this error. Afterwards disk is still at zero free  G blocks and anal/disk still shows file as marked for delete. Try this a  G couple of times with same result. Ten minutes later disk free space is  C around 300000 blocks and anal/disk no longer finds this file. What   happened here?   --  
 Alan Greig   ------------------------------  # Date: Thu, 13 Oct 2005 23:55:19 GMT   From: John Santos <john@egh.com># Subject: Re: File marked for delete + Message-ID: <HtC3f.24160$wm3.7946@trnddc01>    Alan Greig wrote:  > I > A disk is mounted on 3 cluster nodes - 2 Vax, 1 Alpha. Versions of VMS  B > are VAX 6.2, Alpha 7.2-1. Disk is at zero free blocks due to an G > application error. I delete a file around 300000 blocks which is not  H > open (or even likely to have been open for 5 years) according to SHOW E > DEV/FILE on all nodes. After the DELETE free disk space is still 0  G > blocks. I do an ANAL/DISK and it finds the file as marked for delete  I > (only warning it finds on the disk). I now do ANAL/DISK/REPAIR/CONFIRM  I > and respond Y to fix this error. Afterwards disk is still at zero free  I > blocks and anal/disk still shows file as marked for delete. Try this a  I > couple of times with same result. Ten minutes later disk free space is  E > around 300000 blocks and anal/disk no longer finds this file. What   > happened here?  0 Was the file or volume marked "Erase on Delete?"  E What kind of disk?  On some at least, if the file is marked as having F a possible bad block, VMS will start up a bad-block scan on it when itG is deleted to try and isolate the bad block.  (I can't remember if this F applied to SDA disks or only to previous generations, such as RP's and@ RM's.  Or maybe it was an old technique that automatic bad-blockF replacement obviated the need for, but was resurrected for SCSI disks.   --   John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  % Date: Thu, 13 Oct 2005 20:38:50 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> # Subject: Re: File marked for delete , Message-ID: <434EFE07.887B12A2@teksavvy.com>  J re: file marker for delete blocks released much later than ana/disk/repair    F Which process would be busy performing the freeing up of blocks duringG those minute it takes for the operation to complete in the background ?   H I take it it would run on the node that first mounted the drive (or MSCP serves the drives). ?    ------------------------------    Date: 13 Oct 2005 18:08:17 -0700* From: "Alan Greig" <greigaln@netscape.net># Subject: Re: File marked for delete C Message-ID: <1129252097.462332.113040@g44g2000cwa.googlegroups.com>    John Santos wrote:   > 2 > Was the file or volume marked "Erase on Delete?"   No.    > G > What kind of disk?  On some at least, if the file is marked as having   G Served by dual HSJ40 conntrollers. No host based shadowing on this disk C but on some other volumes in the cluster. Other large files deleted E immediately. In fact it may have been deleting them that triggered an F immediate free up of that particular file. Unfortunately I needed freeC space fast and couldn't sit and try and trouble shoot in real time.   H > a possible bad block, VMS will start up a bad-block scan on it when itI > is deleted to try and isolate the bad block.  (I can't remember if this   F Bad block scan I thought of but I think I should have seen other signsC of that. So long since I last saw one now though! Didn't a DIR/FULL : report this and an obvious bad block scan process startup?  H > applied to SDA disks or only to previous generations, such as RP's andB > RM's.  Or maybe it was an old technique that automatic bad-blockH > replacement obviated the need for, but was resurrected for SCSI disks.   --  
 Alan Greig   ------------------------------    Date: 13 Oct 2005 18:27:57 -0700* From: "Alan Greig" <greigaln@netscape.net># Subject: Re: File marked for delete C Message-ID: <1129253277.646472.129490@f14g2000cwb.googlegroups.com>    Alan Greig wrote: H > A disk is mounted on 3 cluster nodes - 2 Vax, 1 Alpha. Versions of VMSA > are VAX 6.2, Alpha 7.2-1. Disk is at zero free blocks due to an F > application error. I delete a file around 300000 blocks which is notG > open (or even likely to have been open for 5 years) according to SHOW   E Actually not 5 years as I copied the file to another disk immediately 7 before the delete which might just be relevant I guess.  --  
 Alan Greig   ------------------------------  % Date: Thu, 13 Oct 2005 21:28:18 -0400 ( From: Bill Todd <billtodd@metrocast.net># Subject: Re: File marked for delete = Message-ID: <V6KdnXZF1P8ulNLeRVn-gw@metrocastcablevision.com>    Alan Greig wrote:  > I > A disk is mounted on 3 cluster nodes - 2 Vax, 1 Alpha. Versions of VMS  B > are VAX 6.2, Alpha 7.2-1. Disk is at zero free blocks due to an G > application error. I delete a file around 300000 blocks which is not  H > open (or even likely to have been open for 5 years) according to SHOW E > DEV/FILE on all nodes. After the DELETE free disk space is still 0  G > blocks. I do an ANAL/DISK and it finds the file as marked for delete  I > (only warning it finds on the disk). I now do ANAL/DISK/REPAIR/CONFIRM  " > and respond Y to fix this error.  G It has yet to be demonstrated that this was an 'error':  having a file  J marked for delete for *some* period of time is a common occurrence on VMS.  '   Afterwards disk is still at zero free I > blocks and anal/disk still shows file as marked for delete. Try this a  I > couple of times with same result. Ten minutes later disk free space is  E > around 300000 blocks and anal/disk no longer finds this file. What   > happened here?  F My only wild guess is that it could have been interaction between the I mechanism that caches released space privately to a node for some period  F of time such that every node will have some space it can grab quickly E without having to coordinate with other nodes.  If this space was so  H reserved, then the other nodes (and a free space query) *might* not see D it as being free for some period of time unless some other activity : triggered a global reevaluation of distributed free space.  A I'd suspect something related to aging data out of a cache as an  E alternate explanation, but you seem to have ruled that out by saying  - that the file hadn't been accessed for years.    - bill   ------------------------------    Date: 13 Oct 2005 18:41:09 -0700* From: "Alan Greig" <greigaln@netscape.net># Subject: Re: File marked for delete C Message-ID: <1129254069.555353.203300@g44g2000cwa.googlegroups.com>    Bill Todd wrote:    G > My only wild guess is that it could have been interaction between the J > mechanism that caches released space privately to a node for some periodG > of time such that every node will have some space it can grab quickly F > without having to coordinate with other nodes.  If this space was so  E I did a SET VOLUME/REBUILD which seemed to free up no space but might D also have been a trigger.  Also did an ANAL/DISK/REPAIR from the oneD Alpha node just before I noticed the space actually free up - may be significant.  I > reserved, then the other nodes (and a free space query) *might* not see E > it as being free for some period of time unless some other activity < > triggered a global reevaluation of distributed free space. > B > I'd suspect something related to aging data out of a cache as anF > alternate explanation, but you seem to have ruled that out by saying/ > that the file hadn't been accessed for years.   D I did copy it to another disk immediately before the delete - shouldF have mentioned that. Later I copied and deleted a file of similar sizeG and it went instantly as I would expect. Files hanging about in "marked ? for delete" status for 10 minutes when they are not open is not * something I can recall ever seeing before. --  
 Alan Greig   ------------------------------    Date: 13 Oct 2005 19:02:33 -0700* From: "Alan Greig" <greigaln@netscape.net># Subject: Re: File marked for delete B Message-ID: <1129255353.547344.86870@g49g2000cwa.googlegroups.com>   JF Mezei wrote: L > re: file marker for delete blocks released much later than ana/disk/repair  B I'm starting to think along the lines of cache incoherency betweenA nodes. I do seem to recall a bug in mixed clusters back about VMS E versions VAX 6.m/Alpha 7.n where you could do a DIR after a DELETE on C different nodes (or something like that) and see different results. ! Anyone remember the details here?  --  
 Alan Greig   ------------------------------  % Date: Thu, 13 Oct 2005 22:18:09 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> # Subject: Re: File marked for delete , Message-ID: <434F1548.C59602F5@teksavvy.com>   Bill Todd wrote:B > I'd suspect something related to aging data out of a cache as anF > alternate explanation, but you seem to have ruled that out by saying/ > that the file hadn't been accessed for years.     C How much caching of bitmap.sys and other .sys files at the root  is 
 involved ????   D Could it be a question of ana/disk/repair acting at such a low levelG that caches for the control files not being invalitated, so the systems E would still us their cached data instead of getting updated data from  the .sys files ?   ------------------------------    Date: 13 Oct 2005 15:59:27 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) O Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) 3 Message-ID: <N0Bbv48AZ8Qt@eisner.encompasserve.org>   \ In article <434EC730.177DD67E@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes: > "Stanley F. Quayle" wrote:E >> Not true.  I had a set of processes communicating through a global D >> section.  Worked perfectly on VAX, but totally messed up on AlphaG >> until I put explicit memory barrier instructions.  Single processor.  >  > G > Pardon my ignorance here, but what does "memory barrier instructions"  > mean ? > F > DO you mean semaphores/flags/locks to signal that one portion of theE > application wants exclusive access to the memory and tell the other 3 > portions to wait an instant before accessing it ?  >  > D > Or is this at a much lower level, declaring a page of memory to be- > non-shareable at the virtual memory level ?   F   MB tells the Alpha CPU to stall until all in-process memory accessesE   are completed to the actual chip that holds the data.  Normally the E   Alpha does not wait in between instructions for writes or modifies  9   to complete thier journey through the memory subsystem.   A   This is necessary before handing off work to another processor, 9   whether that processor is another CPU or an I/O device.   G   Normally only the OS kernel and inter-process communications routines    need to worry about MB.    ------------------------------  % Date: Thu, 13 Oct 2005 17:43:59 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> O Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) , Message-ID: <434ED517.4BFEF65B@teksavvy.com>   Bob Koehler wrote:H >   MB tells the Alpha CPU to stall until all in-process memory accessesG >   are completed to the actual chip that holds the data.  Normally the F >   Alpha does not wait in between instructions for writes or modifies; >   to complete thier journey through the memory subsystem.   H Is this something that came about with Wildwire class machines where theG memory isn't a single block of RAM but rather a collection of blocks of E RAM ? Are there signals back to the CPU originating the write request H that the RAM was in fact updated ? Or does the "stall" just hold the CPUD for a pre-determined amount of time , an amount calculated to ensureA that memory updates to the furthest memory have time to be done ?     H Is this something that is needed even on simpler multi CPU machines like DS/ES class machines ?    H If you have a single RAM controlled by a single memory controller, wouldH such instructions still be needed ? Would one CPU not have direct accessG to all of the RAM and not have to worry about some writes taking longer 
 than others ?    ------------------------------    Date: 13 Oct 2005 15:53:03 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) R Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?)3 Message-ID: <GRXqrCN59cGo@eisner.encompasserve.org>   c In article <434E5CE6.32015.D870169@localhost>, "Stanley F. Quayle" <squayle@insight.rr.com> writes: , > On 13 Oct 2005 at 9:44, Bob Koehler wrote:@ >>    But mostly people used single processor systems, where all# >>    processes see the same thing.  > E > Not true.  I had a set of processes communicating through a global  D > section.  Worked perfectly on VAX, but totally messed up on Alpha F > until I put explicit memory barrier instructions.  Single processor. > 7 > Hoff told me I'd have to do that, and he was right...   H    I've seen the same behaviour change on VAXen moving between compilers.    versions.  Caused by an optimizer change.    E    My copy of the Alpha Architecture Reference goes into a very long  F    discussion about how a single processor must see a consistent view G    of memory.  For Alpha, I/O devices are processors in this respect.   C    Multiple processes on a single CPU are using a single processor.    ------------------------------  # Date: Fri, 14 Oct 2005 01:08:44 GMT 9 From: Bob Harris <nospam.News.Bob@remove.Smith-Harris.us> R Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?)D Message-ID: <nospam.News.Bob-D2CCEC.21084613102005@news.verizon.net>  . In article <434E5CE6.32015.D870169@localhost>,4  "Stanley F. Quayle" <squayle@insight.rr.com> wrote:  , > On 13 Oct 2005 at 9:44, Bob Koehler wrote:A > >    But mostly people used single processor systems, where all $ > >    processes see the same thing. > E > Not true.  I had a set of processes communicating through a global  D > section.  Worked perfectly on VAX, but totally messed up on Alpha F > until I put explicit memory barrier instructions.  Single processor. > 7 > Hoff told me I'd have to do that, and he was right...  >  > --Stan Quayle  > Quayle Consulting Inc. >  > ----------/ > Stanley F. Quayle, P.E. N8SQ  +1 614-868-1363 5 > 8572 North Spring Ct., Pickerington, OH  43147  USA 2 > stan-at-stanq-dot-com       http://www.stanq.com+ > "OpenVMS, when downtime is not an option"   @ I'm guessing that this is because the Alpha cache and TLB works C off of virtual addresses, and if process A's changes have not been  @ forced back to memory before process B attempts to access them, C then it might (I say might) be possible that a global section with  ? 2 different virtual addresses in process A and B might not get  - in-flight A changes unless an MB is executed.    This is wild A$$ Guessing.  @ But I would have expected that an MB would be executed during a A process switch, so I would have expected consistency on an Alpha  ' without the user doing any explicit MB.   = Then again, this might be good training for single processor  @ systems as training for applications that might then get run on 
 SMP boxes.  2                                         Bob Harris   ------------------------------  % Date: Thu, 13 Oct 2005 23:36:24 -0400 2 From: "Stanley F. Quayle" <squayle@insight.rr.com>R Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?)- Message-ID: <434EEF78.7009.FC37B3A@localhost>   ) On 14 Oct 2005 at 1:08, Bob Harris wrote: B > But I would have expected that an MB would be executed during a B > process switch, so I would have expected consistency on an Alpha) > without the user doing any explicit MB.   E It doesn't do that.  It would nuke performance on a system with lots  A of context switches.  From what I've experienced (not having the  D Alpha VMS source listings), it's clear there's never a MB in the OS  unless it's needed...   ? > Then again, this might be good training for single processor  B > systems as training for applications that might then get run on  > SMP boxes.  B The problem I had to fix was first discovered on a 4-way ES box.  C Took about a year for me to reproduce on a DPWS (1 processor), but   it's there, too.  
 --Stan Quayle  Quayle Consulting Inc.  
 ----------- Stanley F. Quayle, P.E. N8SQ  +1 614-868-1363 3 8572 North Spring Ct., Pickerington, OH  43147  USA 0 stan-at-stanq-dot-com       http://www.stanq.com) "OpenVMS, when downtime is not an option"    ------------------------------  % Date: Thu, 13 Oct 2005 17:57:37 -0400 2 From: "Stanley F. Quayle" <squayle@insight.rr.com>X Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) ?) ?). Message-ID: <434EA011.11054.E8D57EE@localhost>  ( On 13 Oct 2005 at 16:44, JF Mezei wrote:G > Pardon my ignorance here, but what does "memory barrier instructions"  > mean ?  F Not having a copy of the Alpha Architecture Reference handy, but with A multiple bullet wounds from dealing with global memory access on  - single-processor Alphas, my understanding is:   =   A Memory Barrier instruction forces the memory subsystem to :   flush all pending writes out to memory.  Other processes;   accessing the same data cell will then be able to see the 9   changes.  If no Memory Barrier instruction is executed, 4   when the changes are visible is non-deterministic.  F In C, there's a built-in _MB() function on Alpha to make this happen. D The same built-in on Itanium does a Memory Fence instruction, which  is similar.   F > DO you mean semaphores/flags/locks to signal that one portion of theE > application wants exclusive access to the memory and tell the other 4 > portions to wait an instant before accessing it ?   F This is (except for the "wait") is basically standard ENQ/DEQ stuff.  D Fortunately, the global section (about 75% of the time) was covered F with a lock, so I added the _MB() to anywhere there's a conversion to 
 null mode.  5 Finding the other 25% was the real not-so-fun part...     
 --Stan Quayle  Quayle Consulting Inc.  
 ----------- Stanley F. Quayle, P.E. N8SQ  +1 614-868-1363 3 8572 North Spring Ct., Pickerington, OH  43147  USA 0 stan-at-stanq-dot-com       http://www.stanq.com) "OpenVMS, when downtime is not an option"    ------------------------------  % Date: Thu, 13 Oct 2005 19:52:14 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> Y Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) ?) ?)  , Message-ID: <434EF31E.DFFE1496@teksavvy.com>   "Stanley F. Quayle" wrote:? >   A Memory Barrier instruction forces the memory subsystem to < >   flush all pending writes out to memory.  Other processes= >   accessing the same data cell will then be able to see the ; >   changes.  If no Memory Barrier instruction is executed, 6 >   when the changes are visible is non-deterministic.  F I take it this has much more to do with memory caches than complex RAM; architectures in a system (aka: Wildfire class machines). ?       F Will the Intel implememntation of dual core (with each core having itsF independant cache) ensure that the memory barrier instruction not onlyD flushes a new value to RAM but also to the caches of other cores and chips ?    ------------------------------  % Date: Thu, 13 Oct 2005 16:44:37 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> W Subject: Re: fork() (Was: Re: Will "COM for OpenVMS" be ported to Itanium ?) ?) ?) ?)?) , Message-ID: <434EC730.177DD67E@teksavvy.com>   "Stanley F. Quayle" wrote:D > Not true.  I had a set of processes communicating through a globalC > section.  Worked perfectly on VAX, but totally messed up on Alpha F > until I put explicit memory barrier instructions.  Single processor.    E Pardon my ignorance here, but what does "memory barrier instructions"  mean ?  D DO you mean semaphores/flags/locks to signal that one portion of theC application wants exclusive access to the memory and tell the other 1 portions to wait an instant before accessing it ?     B Or is this at a much lower level, declaring a page of memory to be+ non-shareable at the virtual memory level ?    ------------------------------  # Date: Fri, 30 Sep 2005 07:51:59 GMT  From: kashe@sonic.net   Subject: Re: HP : Moving forward8 Message-ID: <jcrpj1t309ul4ep9cukqlqr02ur4pr5ds2@4ax.com>  4 On Thu, 29 Sep 2005 21:11:13 -0500, David J Dachtera" <djesys.nospam@comcast.net> wrote:   >... (is that HIPAA compliant?)   3 	It's my understanding that Windows was known to be D non-compliant, but the basic MS argument was, "Do you really want toF try to get the majority of hospitals and doctors change to another OS?A You know they will slay HIPAA if that's a condition." So the govt : stamped Windows as compliant and just grandfathered it in.   ------------------------------  % Date: Thu, 13 Oct 2005 16:40:38 -0500 & From: Thomas Wirt <twnews@kittles.com>6 Subject: HP Tech Forum - Shameless plug for my sessionE Message-ID: <ef1a0$434ed456$4367aba2$24938@msgid.meganewsservers.com>   C For all of you that are coming to the HPTF in Orlando next week, I  C encourage you to attend my Windows/Citrix session # 1185 "Managing  I Multi-user Windows (Citrix) from a System Manager's View".  This talk is  I aimed at the VMS Systems Managers that would rather not use Windows, but  H see the hand writing on the wall that it is an inevitable skill to have F at least a little.  For me it was a situation where, what can I do to G avoid a PC on every desk in the company.  Terminal Services and Citrix  D give me central control and management like I am used to in the VMS B world.  When configured correctly, it even gives you a redundancy I similar (though of course no where near as good as) to the redundancy of   a VMS cluster.  G I hope that some of you will check it out.  I think that you will find  J that it will give you a really strong understanding of multi-user Windows. --     Thomas Wirt  Operations Manager, IS Dept. Kittle's Home Furnishings  Indianapolis, IN   ------------------------------  % Date: Fri, 30 Sep 2005 17:31:21 +0100 1 From: Russell Wallace <russell.no.spam@gmail.com> ? Subject: Re: Layoffs at HP : MH needs your suggestions or ideas 0 Message-ID: <sMd%e.16465$R5.1090@news.indigo.ie>   Lyck wrote: @ > As an European living in US, probably for the rest of my life,= > I have a point of view based on experience on both sides of 9 > the "Great Lake." I sure see many magnificent traits in B > Americans and their society, however, the labor law is a horror.= > Let me summarize how I see it, in the following list, which ! > might be shocking to Europeans:  > : > 5) Nobody protects American jobs for Americans. I see an3 >    unspeakable flood of Indians pouring into this 9 >    country, and many American veterans of the High-Tech ; >    revolution getting out of their jobs and are forced to  >    move out of California.  H So taking these two paragraphs together, what you're saying is that the # Americans ought to deport you ASAP?    --  ) "Always look on the bright side of life." 5 To reply by email, replace no.spam with my last name.    ------------------------------  % Date: Fri, 30 Sep 2005 13:05:04 +0200 , From: Albrecht Schlosser <ajs856@tiscali.de> Subject: Re: Quotation marks, Message-ID: <956jhd.8er.ln@news.hus-soft.de>  # nathanbloomfield@hotmail.com wrote:   = > I'm tring to include quotation marks in the RFILTER string.  >  > $ RFILTER = "23-002-1"0 > $ DEFINE/TABLE=XEN$SYMBOLS TESTFILTER 'RFILTER" > $ SHOW LOGICAL/TABLE=XEN$SYMBOLS    C use one (") to delimit the string and two ("") to represent one (") @ within the string (and ''RFILTER' instead of 'RFILTER within the string):   $ RFILTER = "23-002-1"( $ DEFINE/JOB TESTFILTER """''RFILTER'""" $ SHOW LOGICAL TESTFILTER 2     "TESTFILTER" = ""23-002-1"" (LNM$JOB_8176F2C0)  ( or simply (without a symbol definition):  & $ DEFINE/JOB TESTFILTER """23-002-1""" $ SHOW LOGICAL TESTFILTER 2     "TESTFILTER" = ""23-002-1"" (LNM$JOB_8176F2C0)   Albrecht   ------------------------------  % Date: Thu, 13 Oct 2005 20:42:22 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>  Subject: Re: Samba-3 for OpenVMS+ Message-ID: <434F0CFD.FD831199@comcast.net>    Nigel Barker wrote:  > R > On Wed, 12 Oct 2005 20:17:06 -0500, David J Dachtera <djesys.nospam@comcast.net> > wrote: > % > >Tapani.Rundgren@treville.se wrote:  > >> > >> Hi David! > >>I > >> Distributed Netbeans does not use Windows I/O. It could, but in case = > >> of file access on OpenVMS server, FTP will do just fine. B > >> And You can "mount" a local drive and then copy files from/to5 > >> OpenVMS servers, if that is what You need to do. A > >> There are other FTP based products that are intergrated with C > >> Windows I/O but use FTP for transport, WebDrive comes to mind.  > > H > >Unfortunately, as was dicussed here in another thread some time back,F > >SMB provides more than upload/download capability. Remote files areE > >opened by applications as if they were locally resident, providing 4 > >record-level access which is beyond scope of FTP. > R > Record level access is not required for Distributed NetBeans so FTP will do just > fine.   D Hhmmm... Sounds like high-overhead, and very demanding of bandwidth.   Limited potential usage.   --   David J Dachtera dba DJE Systems  http://www.djesys.com/  ) Unofficial OpenVMS Hobbyist Support Page: " http://www.djesys.com/vms/support/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/   " Unofficial OpenVMS-IA32 Home Page: http://www.djesys.com/vms/ia32/    Coming soon:& Unofficial OpenVMS Marketing Home Page   ------------------------------  # Date: Fri, 14 Oct 2005 02:04:08 GMT   From: John Santos <john@egh.com>  Subject: Re: Samba-3 for OpenVMS, Message-ID: <smE3f.24180$wm3.21923@trnddc01>   David J Dachtera wrote:  > Nigel Barker wrote:  > R >>On Wed, 12 Oct 2005 20:17:06 -0500, David J Dachtera <djesys.nospam@comcast.net> >>wrote: >> >>% >>>Tapani.Rundgren@treville.se wrote:  >>> 
 >>>>Hi David!  >>>>H >>>>Distributed Netbeans does not use Windows I/O. It could, but in case< >>>>of file access on OpenVMS server, FTP will do just fine.A >>>>And You can "mount" a local drive and then copy files from/to 4 >>>>OpenVMS servers, if that is what You need to do.@ >>>>There are other FTP based products that are intergrated withB >>>>Windows I/O but use FTP for transport, WebDrive comes to mind. >>> H >>>Unfortunately, as was dicussed here in another thread some time back,F >>>SMB provides more than upload/download capability. Remote files areE >>>opened by applications as if they were locally resident, providing 4 >>>record-level access which is beyond scope of FTP. >>R >>Record level access is not required for Distributed NetBeans so FTP will do just >>fine.  >  > F > Hhmmm... Sounds like high-overhead, and very demanding of bandwidth. >  > Limited potential usage.  D Not limited all if it is exactly what you need.  The Original PosterD was asking specifically about Samba-3 in the context of using it forB NetBeans.  If NetBeans is happy with FTP access then it solves the problem.  C BTW, NetBeans is a program development environment written in Java. ? I imagine the issue here is checking out modules (files) from a C library for local editing and testing, then checking them back into = the library, and then uploading them to the target system for A compilation, testing and deployment.  A file-based solution seems  entirely appropriate for this.       --   John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------    Date: 13 Oct 2005 14:31:02 -0700' From: "syslost" <wm.reynolds@gmail.com> Y Subject: Re: Usage of 3X-DEGXA-TB within AlphaServer 1000 5/400, AlphaStation DS20, Alpha B Message-ID: <1129239062.581434.13700@g43g2000cwa.googlegroups.com>   Hummm.   The short version:  C You may have to go into the AlphaBIOS console to configure pci card   E I don't what you're trying to do.  If your trying to bring the Alphas C into a cluster via the gb ethernet cards, the cards will have to be G recognized at the console.  Do a show config at the SRM >>> prompt.  If 9 the cards are shown, and sysgen params vaxcluster =2, and D niscs_load_pea0 = 1, you should be able to cluster the machines.  IfG the cards show as ???? from the show config, you could put another card C that is recognized by the system, then in syconfig.com, install the G drivers for the gig cards, disable the other cards and you'll have your  cluster.  G As far as using the cards for the network (decnet, tcpip, ...) load the ' drivers for the cards from syconfig.com   / Again, what are you trying to do with the card?    ------------------------------  # Date: Thu, 13 Oct 2005 21:30:12 GMT 6 From: "Kenneth Farmer" <kfarmer@NOSPAM.spyderbyte.com>% Subject: Re: VMS logo feedback needed < Message-ID: <ElA3f.4025$pP1.222263@twister.southeast.rr.com>  ; "JF Mezei" <jfmezei.spamnot@teksavvy.com> wrote in message  & news:434D78B2.486D7650@teksavvy.com... > Kenneth Farmer wrote:  >>F >> Looks good JF. Would the Great White look bigger with smaller eyes? > F > Vernon's eyes were done last year... you should have complained back > then :-) :-) :-) > J > However, I do plan a revision this fall to fix one line.  Should the eye > really be reduced ?  >  > 1 >> Check out Cafe Press, http://www.cafepress.com  > G > Blank page on netscape. But Was able to get something out via Mosaic. C > (validator.w3.org found about 200 html errors on the front page).  > F > So they basically want 300dpi for printing.  For an 8** inch image , > that would be 2400*2400. >  > H >> I can get some test tshirts/mugs/mouse pads made if the guy is still 
 >> around.K >> Let me know if interested.  I'll try to contact him today and follow-up   >> with  >> you.  >  > @ > How would you normally function ? Get a big batch done ? Or doH > print-on-demand whenever you have an order  you get thenm to produce a6 > mug or t-shirt or polo shirt and have them ship it ?    L I'll call the guy next week.  I'm off for a weekend of fishing then Orlando.       Ken   % _____________________________________  Kenneth Farmer <>< 336-736-7376 3 www.OpenVMS.org | dba.OpenVMS.org | dcl.OpenVMS.org  HP OpenVMS News and Info   ------------------------------  % Date: Thu, 13 Oct 2005 19:44:09 -0400 % From: BRAD <bradhamilton@comcast.net> & Subject: [Fwd: File marked for delete]* Message-ID: <434EF149.5000306@comcast.net>  " -------- Original Message -------- Subject: File marked for delete # Date: Thu, 13 Oct 2005 23:08:29 GMT ( From: Alan Greig <greigaln@netscape.net> To: Info-VAX@Mvb.Saic.Com     F A disk is mounted on 3 cluster nodes - 2 Vax, 1 Alpha. Versions of VMS? are VAX 6.2, Alpha 7.2-1. Disk is at zero free blocks due to an D application error. I delete a file around 300000 blocks which is notE open (or even likely to have been open for 5 years) according to SHOW B DEV/FILE on all nodes. After the DELETE free disk space is still 0D blocks. I do an ANAL/DISK and it finds the file as marked for deleteF (only warning it finds on the disk). I now do ANAL/DISK/REPAIR/CONFIRMF and respond Y to fix this error. Afterwards disk is still at zero freeF blocks and anal/disk still shows file as marked for delete. Try this aF couple of times with same result. Ten minutes later disk free space isB around 300000 blocks and anal/disk no longer finds this file. What happened here?   --  
 Alan Greig  I Since the file filled up the disk, I'm going to hazard a guess here, and  H say that the file may have been extremely fragmented.  Depending on the D disk I/O subsystem, I've seen files of that size (or larger) take a H while to "delete" for that reason (fragmentation).  Also, was "erase on = delete" set on the volume?  That might slow things down, too.    ------------------------------    Date: 13 Oct 2005 18:15:51 -0700* From: "Alan Greig" <greigaln@netscape.net>* Subject: Re: [Fwd: File marked for delete]C Message-ID: <1129252551.493364.216440@g14g2000cwa.googlegroups.com>    BRAD wrote:  >  > J > Since the file filled up the disk, I'm going to hazard a guess here, andI > say that the file may have been extremely fragmented.  Depending on the   F I don't think it was fragmented. It was one of the oldest files on theC disk (an OLB library that hadn't been used in years). Newer heavily = fragmented .LOG files of similar size later deleted virtually  immediately.  E > disk I/O subsystem, I've seen files of that size (or larger) take a I > while to "delete" for that reason (fragmentation).  Also, was "erase on   C I also checked disk activity. It was effectively zero (that I could G see) across the cluster on this disk until I managed to free some space  by deleting other files.  ? > delete" set on the volume?  That might slow things down, too.    Unfortunately not.   ------------------------------   End of INFO-VAX 2005.573 ************************