1 INFO-VAX	Sat, 25 Mar 2006	Volume 2006 : Issue 168       Contents:0 Re: "Threads are crap in inner-mode!" - Discuss.0 Re: "Threads are crap in inner-mode!" - Discuss.) Re: Converting SYSDEVICE  from ODS-2 to 5  Re: EFN$C_ENF Chalk or Cheese? Re: EFN$C_ENF Chalk or Cheese? Re: EFN$C_ENF Chalk or Cheese? Re: F$FAO (!%D)  Re: HP Support moving offshore?  Re: Intel to sell itanium? Re: Intel to sell itanium? Re: Intel to sell itanium? Re: Intel to sell itanium? Re: Intel to sell itanium?" Re: LWP::Simple crashes on VMSperl- Re: strange behaviour: toasted network cards? - Re: strange behaviour: toasted network cards? - Re: strange behaviour: toasted network cards? - Re: strange behaviour: toasted network cards? G Re: The Minimum You Need to Know to Be an OpenVMS Application Developer G Re: The Minimum You Need to Know to Be an OpenVMS Application Developer G Re: The Minimum You Need to Know to Be an OpenVMS Application Developer 2 VMS and responsible computing. Just Ask The Vendor+ VMS Securing Delivery - Just Ask The Vendor / Re: VMS Securing Delivery - Just Ask The Vendor   Where do DSA devices come from ?$ Re: Where do DSA devices come from ?$ Re: Where do DSA devices come from ?$ Re: Where do DSA devices come from ?$ Re: Where do DSA devices come from ?$ Re: Where do DSA devices come from ?$ Re: Where do DSA devices come from ?  F ----------------------------------------------------------------------    Date: 25 Mar 2006 07:25:27 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) 9 Subject: Re: "Threads are crap in inner-mode!" - Discuss. 3 Message-ID: <W4t06EWLMwoM@eisner.encompasserve.org>   f In article <e02ini$8e$1@news-02.connect.com.au>, "Richard Maher" <maher_rj@hotspamnotmail.com> writes:  L > While perusing the source listing for the VMS System Service Dispatcher, IK > had a serendipitous bump into something called the "inner-mode semaphore" G > and I have to say that my initial reaction was one of revulsion.As if N > thread-coders don't introduce enough of their own mutex bottlenecks; lurkingK > down deep in the guts of an innocent looking RTL, 3.8usecs of (relentless 0 > :-) CPU spin could be unleased on your system.  L What you dislike about VMS is what a lot of us like about VMS - reliability.I The Inner Mode Semaphore is a major contributor to the reliability of the  VMS support for threads.  J > I was of the opinion that the main appeal of threads was that everythingG > could be inlined. Being blocked and waited was no big deal 'cos other F > threads could keep going. No ASTs no asynchronous versions of SystemN > Services calls, "just wait it" but if it's an exec mode lock then aren't youJ > holding the semaphore at exec mode? Does Wait_Callers_Mode come in here?  E The major appeal of threads compared to ASTs, is that the programming C model is simpler to understand.  The drawback is running up against C performance issues that are not obvious but are essential to retain B operating system consistency.  Threads also allow some to run codeD originally written for other operating systems, where consistency is" left more to the luck of the user.  E Even if one engages in counting process slots as a simplistic measure F of overhead, each kernel thread takes up a process slot, so efficiency< is not a characteristic of kernel threads, compared to ASTs.   ------------------------------  % Date: Sat, 25 Mar 2006 13:23:47 -0500 ( From: Bill Todd <billtodd@metrocast.net>9 Subject: Re: "Threads are crap in inner-mode!" - Discuss. G Message-ID: <dZednYuuve8rF7jZnZ2dnUVZ_sednZ2d@metrocastcablevision.com>    Larry Kilgallen wrote:   ...   G > The major appeal of threads compared to ASTs, is that the programming ! > model is simpler to understand.   A I suppose that might have been true prior to the introduction of  B multi-processor systems, but that was rather a long time ago.  Or G perhaps you're just not familiar with applications which could benefit  3 from the concurrent use of more than one processor.    - bill   ------------------------------  % Date: Sat, 25 Mar 2006 05:02:05 -0500 ' From: Dave Froble <davef@tsoft-inc.com> 2 Subject: Re: Converting SYSDEVICE  from ODS-2 to 5/ Message-ID: <hcydnZfniNIZibjZRVn-qw@libcom.com>    Richard B. Gilbert wrote:  > Hoff Hoffman wrote:  > : >> In article <op.s6xwth06zgicya@hyrrokkin>, "Tom Linden"  >> <tom@kednos.com> writes: D >> :Well, I can confirm that it [the on-line BACKUP] was successful. >>L >>   Congratulations on your successful data transfer and data verification.J >>   Me?  I use the documented and recommended means, and I do not use theL >>   backup/ignore=interlock lightly -- I've recovered system disks and dataJ >>   (application) disks where the system manager had used this qualifier, >>   and had gotten caught out.  >>G >>   It's easy to implement a "usually works" backup, and y'all expect  	 >> better @ >>   of OpenVMS.  That's why I regularly flag /ignore=interlock. >>4 >>  ---------------------------- #include <rtfaq.h>   >> -----------------------------5 >>     For additional, please see the OpenVMS FAQ --   >> www.hp.com/go/openvms/faq6 >>  --------------------------- pure personal opinion  >> ---------------------------J >>        Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[\0100]hp.com >> >  >  > E > What DO you do when you can't do a standalone backup and don't use  I > /IGNORE=INTERLOCK?  There are files on the system disk that are always  F > open and some are files you would not want to lose.  I thought that K > /IGNORE=INTERLOCK was a "best effort" that usually worked OK.  I did all  H > my backups that way for years and never had a problem the few times I  > needed to restore.  H Regardless of what Steve is saying, if you have an idle system, nothing H running, no pending updates and such, then you can get a good backup of F a system disk.  Taking down the networking and such helps to get that I idle system.  Stopping the queue manager also helps.  Having just done a  G minimum boot is even better.  Doing a verify pass is your indicator of   success.  G > There are systems that are supposed to run 24x7.  How should they be  K > backed up?  Solaris has something called "fssnap" that supposedly solves   > this problem. . . .   F Here's the gotcha.  If the system is supposed to run 24 x 7, then you E cannot guarantee that nothing is running, and that there are not any   pending updates, and such.  G Even doing a VERIFY pass will not guarantee you a good backup if there   are pending updates.  G The times I've used it is when I want to clone a system disk.  Nothing  C is running except backup.  If the queue file and such would have a  H problem, well it's not a problem because I'm going to change the system ) name and start a new queue file 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: 25 Mar 2006 05:44:42 -0800< From: "Hein RMS van den Heuvel" <heinvandenheuvel@gmail.com>' Subject: Re: EFN$C_ENF Chalk or Cheese? B Message-ID: <1143294282.207242.51150@i39g2000cwa.googlegroups.com>  5 [Dislaimer.... I have not had my morning coffee yet ]    >> How does EFN$C_ENF work?   D By not doing anything, not clearing it, not setting it, not checking8 whether the event of it being set will trigger a wakeup.  D It is there for folks who firmly believe that a chain of ASTs nicelyF keeps your process going. (like you, as indicated in the recent kernel thread topic?)  E So when those applications do a QIO, or ENQ, why bother set an EFN on D completion as the event is not supposed to wake up anything, the AST( will fire and make the next step happen.  < I requested a special EFN value while I was maintaining RMS.E We worked on (application) problems which were due to RMS using event G flags which were never really used by RMS itself. It made me think that E first and foremost it would be safer if RMS (and others) would simply C not use an EFN as they did not need one, and secondly some overhead  could be avoided.   G I never considered or speculated about specifying EFN$C_ENF with $SYNC. C That seems to make no sense at all to me. (Trying to use relatively 4 nice words here in case I have to eat them later :-)   Regards, Hein.    ------------------------------    Date: 25 Mar 2006 06:16:47 -0800< From: "Hein RMS van den Heuvel" <heinvandenheuvel@gmail.com>' Subject: Re: EFN$C_ENF Chalk or Cheese? C Message-ID: <1143296207.718613.284210@t31g2000cwb.googlegroups.com>   E Well, the documenation actually does strongly suggest using EFN$C_ENF F for the wait form of services. So I'm glad I used nice words. I'm justA coming from a different angle alltogether, not worrying about the A sync/wait (output) side but more concerned about the QIO, ENQ,... * (input) side. I guess I just hate waiting.  G http://h71000.www7.hp.com/doc/82FINAL/5841/5841pro_020.html#index_x_604   ( " 6.8.4 Using EFN$C_ENF Local Event FlagG The combination of EFN$C_ENF and a status block should be used with the ? wait form of system services, or with SYS$SYNCH system service. A EFN$C_ENF does not need to be initialized, nor does it need to be E reserved or freed. Multiple threads of execution may concurrently use B EFN$C_ENF without interference as long as they use a unique statusF block for each concurrent asynchronous service. When EFN$C_ENF is usedG with explicit event flag system services, it performs as if always set. ? You should use EFN$C_ENF to eliminate the chance for event flag 
 overlap. "  G http://h71000.www7.hp.com/doc/82FINAL/5841/5841pro_020.html#index_x_607      Hein.    ------------------------------    Date: 25 Mar 2006 08:28:12 -0800< From: "Hein RMS van den Heuvel" <heinvandenheuvel@gmail.com>' Subject: Re: EFN$C_ENF Chalk or Cheese? B Message-ID: <1143304092.649678.14070@v46g2000cwv.googlegroups.com>   [Ok, had my coffees]  " >> I requested a special EFN value   We did not 'invent it'. < The RMS team was just an other voice seeing the need for the no-efn-efn. B There is lots of history there, and I just got a nice writeup from0 someone who was more closely involved back then:  @ " It was all triggered by the move to Alpha.  The original Alpha> releases removed a piece of behind the scene magic VAX VMS did? with EFN's.  The result was that on Alpha lots of code that ran = correctly by LUCK on VAX failed.  Never nicely customers were = seeing lots of applications blow through $QIOW wihout the I/O  being done.   A        There were lots of cases but basically they were using EFN > 0 without realizing it.  Something in the program would set it@ and we would complete the I/O.  This was bad and chewing up lots@ of engineering time.  I looked at more than a few programs doingA I/O to terminals that were falling over.  About the same time Xyz 
 (Hein: -5 ;-) ! was chasing some in other places.   D        We decided on a couple of things.  One there really should beF an EFN for programmers who really did not need or want one.  Also thatE the rules for done in $SYNCH would change.  Both were good things the D $SYNCH change was done silently I don't remember if we did a releaseA note or not.  What rules now in synch is the low word in the IOSB ! changing from 0 to anything else.   A        The no EFN, EFN I believe is a good if seldom used thing."    And...  % VMS 6.2 comments:          4-AUG-1994 B                SYNCH has previously assumed that a state where theA                status word is non-zero and the EFN is clear could C                not occur, except with a defective application code.   @                The assumption is now reversed.  If a status wordG                (address) is supplied, a non-zero status value indicates "                service completion.   And...  D       " OK, I looked in V7.0 which is where efn$c_enf showed up.  ItF does not do much with it.  If it is selected it will just blow through the B wait.  Now what was speculated it does not care about the state ofF other EFN, also in wait for OR, or AND versions supplying that gets anB error for obvious reasons.  It means what it says no EVENT FLAG. "   Cheers,  Hein.    ------------------------------  % Date: Sat, 25 Mar 2006 12:07:09 -0600 @ From: "David J. Dachtera" <djesys.nospam@NeOaSrPtAhMlNiOnWk.net> Subject: Re: F$FAO (!%D)6 Message-ID: <442586CD.3821E1EA@NeOaSrPtAhMlNiOnWk.net>   Hein RMS van den Heuvel wrote: > ? > The !%D FAO directive makes DCL pick up a 64 bit value from a  > semi-random stack location. C > You can coerce DCL to get a usefull 64 bit value there through an : > undocumented, unsupported (for DCL) other directive !AD:! > F$CVUI(32,32,F$FAO(!AD,8,text)) = > This had been published several times before. For examples:  > d > http://groups.google.com/group/comp.os.vms/search?group=comp.os.vms&q=%22%21%25D%22+F%24FAO&qt_g=1 >  > One specific example: E > binary_time[ 0,32] = %xA4FE0000       ! low longword of binary date 7 > binary_time[32,32] = %x0099D8C4       ! high longword  > date_string = -  > L > f$cvtime(f$fao("!%D",f$cvui(32,32,f$fao("!AD",8,binary_time))),"ABSOLUTE") > F > This unsupported technique has survived all major VMS version on VAX > and Alpha alike.I > I have not had a chance to try it under V8, or Itanium, but I assume it  > will continue to work.   Seems to still be functional:    TD183::DJESYS$ vers * IA64 V8.2-1   (HP rx2600  (1.40GHz/1.5MB)) TD183::DJESYS$ @dcl_date(   DATE_STRING = "7-FEB-1996 09:10:53.00" TD183::DJESYS$ ty dcl_date.comE $ binary_time[ 0,32] = %xA4FE0000       ! low longword of binary date 7 $ binary_time[32,32] = %x0099D8C4       ! high longword  $ date_string = - J f$cvtime(f$fao("!%D",f$cvui(32,32,f$fao("!AD",8,binary_time))),"ABSOLUTE") $ sh sym date_string $ exit TD183::DJESYS$     --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  + Date: Sat, 25 Mar 2006 11:38:17 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)( Subject: Re: HP Support moving offshore?$ Message-ID: <e03a38$afp$2@online.de>  3 In article <rI1dF6r6Vnmn@eisner.encompasserve.org>, 6 scopelliti@encompasserve.org (Pat Scopelliti) writes:   K > I hear stories that HP Support is being moved to India and Costa Rica (an O > interesting combination).  True?  Do others feel this is a good thing or bad?   I HP is a company owned by shareholders.  Most, if not all, are interested  I in making money ("earn" seems out of place here) before they are too old  H or ill to spend it.  Why should they care about the long-term future of  the company?  H (Of course, this assumes that offshore support is bad for the long-term I future of the company; this might not be the case.  Of course, it is bad  I for people previously employed who get fired because someone else can do  G it cheaper, but, hey, that's capitalism.  As long as the government is  G not subsidising the company (I don't know if any governments subsidise  2 HP, there is little they can do about it, either.)   ------------------------------  % Date: Sat, 25 Mar 2006 09:23:37 +0100 + From: Karsten Nyblad <nospam@nospam.nospam> # Subject: Re: Intel to sell itanium? = Message-ID: <4424fdff$0$67260$157c6196@dreader2.cybercity.dk>    JF Mezei wrote: E > A buyer of IA64 woudln't be buying it to maintain IA64, it would be F > buying it to gain access to patents. And Intel could claim that theyF > hace recuperated some of its investment into a big heavy slow movingG > monster. You can bet that Intel would state something to the order of 4 > "tersm of the transations were not made public".   > E > Intel could also do something akin to what PSION did to offload its C > EPOC32 OS: create a consortium to own IA64 with Intel still a big J > shareholder, and Intel could slowly sell shares to another member of theJ > consortium until Intel was no longer a shareholder at all.  The recentlyO > formed IA64 solutions alliance might be the foundation for such a consortium.  > H > Intel could then bill that alliance for engineering work that needs toG > be done to IA64 and then the alliance could find a FABbing contractor  > (likely intel again).   E It is very difficult to be a CPU vendor without having your own chip  > plants to produce them, and it is very expensive to develop a > semiconductor process suitable for producing high performance F microprocessors.  Thus, the only likely buyers are those that already > have facilities for producing high performing microprocessors.  H I just don't believe in IA64 being owned by somebody without facilities A for producing microprocessors.  There are only so many with such  : facilities, and who of them would be interested in Itanic?  H The original story on The Inquirer sounds more like a wild rumor than a # story backed by creditable sources.    ------------------------------  % Date: Sat, 25 Mar 2006 04:03:41 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> # Subject: Re: Intel to sell itanium? , Message-ID: <4425074C.423584A9@teksavvy.com>   Karsten Nyblad wrote: F > It is very difficult to be a CPU vendor without having your own chip? > plants to produce them, and it is very expensive to develop a ? > semiconductor process suitable for producing high performance  > microprocessors.    C Is that really the case ? Can't chip design be done independantly ? H After all, Alpha was like that for a number of years and was still quite2 viable at the time Compaq/HP decided to murder it.  G Also remember that IA64 is developped and already has one generation in F the pipeline and I suspect a buyer would be obligated to just completeG that design before they can declare IA64 "mature". A bit like finishing 3 EV7 after Alpha's death was declared June 25 2001.    6 >  Thus, the only likely buyers are those that already@ > have facilities for producing high performing microprocessors.  B Those interested would be those who would benefit from the patentsE associated with that IA64 thing. And remember that Intel may actually 7 pay the company to complete the last iteration of IA64.   I > The original story on The Inquirer sounds more like a wild rumor than a % > story backed by creditable sources.     G Of course it is a wild rumour. But this rumour brings into play another B possible exit strategty for Intel with regards to that IA64 thing.   ------------------------------  % Date: Sat, 25 Mar 2006 12:06:17 +0100 + From: Karsten Nyblad <nospam@nospam.nospam> # Subject: Re: Intel to sell itanium? = Message-ID: <44252420$0$60778$157c6196@dreader1.cybercity.dk>    JF Mezei wrote:  > Karsten Nyblad wrote:  > F >>It is very difficult to be a CPU vendor without having your own chip? >>plants to produce them, and it is very expensive to develop a ? >>semiconductor process suitable for producing high performance  >>microprocessors. >  >  > E > Is that really the case ? Can't chip design be done independantly ? J > After all, Alpha was like that for a number of years and was still quite4 > viable at the time Compaq/HP decided to murder it.  E Sure, but it is becoming more and more difficult.  You do today, and  D that means that you would have to work closely together with a chip I producer for the four years period it takes to develop a microprocessor.  G   The chip producer would have to tell the processor design team about  F their new process.  How many chip producers would be willing to share C information about technology were they might not have had time for   applying for patents yet?   B Sun and Mips have always been fabless.  When Sun and Mips started G developing Sparc and Mips, you did not use cutting edge semi conductor  D processes for microprocessors.  Mips is dead and Sun, well the only @ impressive processor they have produced for years is the new T1.  D Please note that from Digital sold it chip production facilities to A Alpha was killed, Alpha slided from an undisputed first place in  % performance to be one of the leaders.   I > Also remember that IA64 is developped and already has one generation in H > the pipeline and I suspect a buyer would be obligated to just completeI > that design before they can declare IA64 "mature". A bit like finishing 5 > EV7 after Alpha's death was declared June 25 2001.    H Intel can keep Itanic alive for a long time by putting more and more of H the cores currently being developed on each single chip.  That does not  take a third partner.   6 >> Thus, the only likely buyers are those that already@ >>have facilities for producing high performing microprocessors.  ' And why would Intel sell those patents?   D > Those interested would be those who would benefit from the patentsG > associated with that IA64 thing. And remember that Intel may actually 9 > pay the company to complete the last iteration of IA64.   E As far as I can see, your assertion is that the Intel management has  G realized that Itanic is a failure, and that they admit it when talking  G to each other.  (Please note that it is not enough that each person in  F Intel's management admit it to himself.  The Intel management have to B admit it to each other.)  Further you seem to assume that Intel's C management is willing to pay a lot of money not to admit it to the  4 public.  I find your assumptions highly speculative.  I > Of course it is a wild rumour. But this rumour brings into play another D > possible exit strategty for Intel with regards to that IA64 thing.  D An easier exit strategy would be to give Itanic too few development G resources, such that it will continue to be not quite as fast as Power  I and having problem competing to x86.  Windows and Linux will continue to  H get better and better at big servers.  At least Linux will get features I for high availability.  That will force the producers of non Windows and  C Linux machines to consentrate on very large machines and very high  F availability, but that market is only so big.  There will not be room I for all of the vendors, that are currently trying to get a slice of that  C market.  If Itanic does not get competitive in the high end of the   market soon, then it will die.   ------------------------------    Date: 25 Mar 2006 04:19:56 -0800 From: shoppa@trailing-edge.com# Subject: Re: Intel to sell itanium? C Message-ID: <1143289196.929676.289340@u72g2000cwu.googlegroups.com>    Dave Froble wrote:! > shoppa@trailing-edge.com wrote: J > > That said, this is not the first time in the industry (for Intel it isF > > at least the fourth) that they swore up and down that a particularA > > chip/architecture was going to be the future of all computing  > H > Lets see, how does it go?  Fool me once, shame on you.  Fool me twice,5 > shame on me.  But who to blame for the fourth time?   G Maybe this particular newsgroup is dominated by old-timers, but most IT F industry types have only been around for a generation of technology atE most. And fewer and fewer have been exposed to more than one brand of  CPU/OS.   @ Itanium is admittedly a much bigger "fool" thing than any of theF previous Intel iterations. Even the second-biggest incident (iAPX 432)F didn't extinguish a whole generation of competing architectures (maybeG it influenced a couple negatively in terms of "we need more complexity" 4 but there were other influences that said that too.)  E Perhaps the only historical incident of similar magnitude is IBM with B "Future Systems" (FS). And I doubt many outside IBM corporate ever: drank that Kool-Aid (but internally it was made a-plenty!)  : But even the FS effort only lasted 4 years (1971 to 1975).  > Interestingly a quick google turned up a 1999 comp.arch thread comparing Itanic with FS.   C The Japanese attempt at "Fifth Generation computing" in the 80's so E completely missed the boat... but a lot of US companies worried about D it enough to think about building their own boats. (Sort of like howB mineral/oil companies started worrying that they would miss out onG mining Manganese Nodules from the Ocean Floor after the Glomar Explorer  hit the press!)   E Hindsight is 20/20, but looking at that 1999 comp.arch thread I don't F think anyone could've argued that Itanic *had* to flop the way it did.C Most were arguing that 80x86 CPU's really would run out of steam in C very early 2000's, and their arguments really seem sound. They just  weren't right!  F IF everyone at Intel had truly drunk the Kool-Aid AND as a result theyC stopped all IA32 development to focus on IA64, THEN IA64 would have D actually succeeded. But the competition from AMD et al actually kept Intel in the IA32 batting cage.    Tim.   ------------------------------  % Date: Sat, 25 Mar 2006 07:27:13 -0800 # From: "Tom Linden" <tom@kednos.com> # Subject: Re: Intel to sell itanium? ) Message-ID: <op.s6y6jnmfzgicya@hyrrokkin>   K On Sat, 25 Mar 2006 03:06:17 -0800, Karsten Nyblad <nospam@nospam.nospam>    wrote:  H > As far as I can see, your assertion is that the Intel management has  J > realized that Itanic is a failure, and that they admit it when talking  J > to each other.  (Please note that it is not enough that each person in  I > Intel's management admit it to himself.  The Intel management have to   E > admit it to each other.)  Further you seem to assume that Intel's   F > management is willing to pay a lot of money not to admit it to the  6 > public.  I find your assumptions highly speculative. > J Well, since we are speculating I would guess that the (projected) ROI on   Itanium + can only be a fraction of that for the X86.    ------------------------------  % Date: Sat, 25 Mar 2006 08:49:17 -0600 6 From: "Craig A. Berry" <craigberry@mac.com.spamfooler>+ Subject: Re: LWP::Simple crashes on VMSperl @ Message-ID: <craigberry-3C9D36.08491725032006@free.teranews.com>  B In article <1143237161.931669.48370@t31g2000cwb.googlegroups.com>,-  "juna" <ggl.20.jjpon@spamgourmet.com> wrote:   I > I have a perl script called getprint_test.pl to fetch and display a web  > page:  >  > use LWP::Simple;# > getprint "http://www.google.com";  > 6 > It crashes on my VMSperl with the following message: > E > Fatal VMS error (status=316) at P_ROOT:[000000]VMS.C;2, line 662 at 1 > /perl_root/lib/site_perl/LWP/Simple.pm line 20. ; > Compilation failed in require at getprint_test.pl line 1. : > BEGIN failed--compilation aborted at getprint.pl line 1.' > %SYSTEM-F-IVCHAN, invalid I/O channel  > 8 > I'm running OpenVMS v8.2 on an Alpha DS10. Perl v5.8.6F > I tried re-installing libwww-perl versions 5.805, 5.800, 5.69, 5.51, > 5.10 with the same result. >   B I can't reproduce this with Perl 5.8.4, LWP 5.79, which is what I B happen to have installed.  The error you are getting is  where it E spawns a subprocess running SHOW LOGICAL * and then reads and parses  F the output in order to populate the %ENV hash.  You should check your < PRCLM quota to make sure you aren't exceeding the number of B subprocesses you're allowed to create and also make sure you have E NETMBX and TMPMBX privileges since the IPC is done through a mailbox.   G You might want to debug this further by seeing if you can run anything   that accesses %ENV, such as:  ) $ perl -e "print join qq/\n/, keys %ENV;" . *** Free account sponsored by SecureIX.com ***X *** Encrypt your Internet usage with a free VPN account from http://www.SecureIX.com ***   ------------------------------  % Date: Sat, 25 Mar 2006 13:01:41 +0100  From: rejoc <rejoc@FREEfree.fr> 6 Subject: Re: strange behaviour: toasted network cards?4 Message-ID: <44253127$0$27895$626a54ce@news.free.fr>  2 Phillip Helbig---remove CLOTHES to reply a crit :E > About a week ago, there was an unexpected power outage.  After the  I > cluster came back up, some strange behaviour was noticed.  Maybe there  I > is no relation, but I suspect there is, due to the coincidence in time.    > B > Could the power outage have damaged network cards or something, I > resulting in this problem?  (There have been power outages before, and  ) > they never seemed to have bad effects.)  > G could be bad negotiation (half/full duplex) between the NIC and switch  * on one (or more) systems in the cluster...E It can cause nodes disappearing/coming back in a cluster under heavy  
 network load.    ------------------------------  % Date: Sat, 25 Mar 2006 14:08:02 +0100  From: rejoc <rejoc@FREEfree.fr> 6 Subject: Re: strange behaviour: toasted network cards?3 Message-ID: <442540b4$0$9638$636a55ce@news.free.fr>   2 Phillip Helbig---remove CLOTHES to reply a crit :< > In article <44253127$0$27895$626a54ce@news.free.fr>, rejoc > <rejoc@FREEfree.fr> writes:  > J >> could be bad negotiation (half/full duplex) between the NIC and switch - >> on one (or more) systems in the cluster... H >> It can cause nodes disappearing/coming back in a cluster under heavy  >> network load. > K > OK, but is this something one would expect to have changed after a power   > cycle? > ; If the connections are autonegocatied, yes, I've seen such  % mis-negociation after a power off/on.   C That's why I always force the configuration of NIC and switch port.   @ You can check the configuration with LANCP on the system side...   ------------------------------  % Date: Sat, 25 Mar 2006 08:05:24 -0500 ) From: "Neil Rieck" <n.rieck@sympatico.ca> 6 Subject: Re: strange behaviour: toasted network cards?9 Message-ID: <gebVf.7784$qX6.287656@news20.bellglobal.com>   M "Phillip Helbig---remove CLOTHES to reply" <helbig@astro.multiCLOTHESvax.de>  / wrote in message news:e038fm$afp$1@online.de... D > About a week ago, there was an unexpected power outage.  After theH > cluster came back up, some strange behaviour was noticed.  Maybe thereI > is no relation, but I suspect there is, due to the coincidence in time.  > G > The strange behaviour is the following.  While working interactively, H > everything is fine, then things freeze up for a few seconds, then workH > fine again.  It is ALMOST like a slow network, but not quite; when theH > network is slow since it is overloaded, it is SLOW---what is happeningE > now is that throughput is in fits and starts---it's either there at 4 > apparently normal speed or it is not there at all. > F > I'm accessing the cluster remotely, but I'm sure that the problem isF > within the cluster since other machines on the LAN, both locally and$ > remotely, don't show this problem. > G > Here is the error count on PEA0: for the three nodes currently in the E > cluster: 1229, 819 and 20472.  In the first case, this is the error I > count since 18 March.  For the other two, there was an unplanned reboot I > (i.e. it happened automatically; no power outage) a day and a half ago, J > so the error count was reset then.  Before this reboot of the two nodes,E > the ratio of errors was about the same.  Of course, 20472 is a huge , > error count on PEA0: for a day and a half! > . > Could a too-high error count cause a reboot? > A > Could the power outage have damaged network cards or something, H > resulting in this problem?  (There have been power outages before, and) > they never seemed to have bad effects.)  > J > The error count doesn't grow gradually, but also in fits and starts, butI > it is NOT the case that it increases when there is a freeze up; the two , > problems seem to be unrelated temporaally. > G > Is there a way, after an unplanned reboot, to find out WHY the reboot * > occurred?  I didn't find any crash dump. >   J The following network problem just happened to me and was caused by cheap H network hardware. After a power surge, I started to notice poor network L performance (but no OpenVMS crash). My problem was traced to an old LinkSys L Router (BEFSR41 ver-1) which we were using as an inexpensive alternative to F the CISCO PIX firewall. A Google search revealed a known power supply J problem in the 16-LED version of this product. See the following link for 	 pictures. # http://www.dondoucette.com/befsr41/ I I opened my unit and noticed the primary filter capacitor was bulged out  K more than the pictures in the link above. Note that "Ver-1" (16-LED) units  C have a 16v filter capacitor while "Ver-3" (6-LED) units have a 25v  D capacitor. We've only got one bad unit out of many so don't get the - impression that I'm against LinkSys products.   
 Neil Rieck Kitchener/Waterloo/Cambridge,  Ontario, Canada.8 http://www3.sympatico.ca/n.rieck/links/cool_openvms.html      ------------------------------  % Date: Sat, 25 Mar 2006 14:33:38 +0100  From: rejoc <rejoc@FREEfree.fr> 6 Subject: Re: strange behaviour: toasted network cards?3 Message-ID: <442546b6$0$7934$636a55ce@news.free.fr>   2 Phillip Helbig---remove CLOTHES to reply a crit :; > In article <442540b4$0$9638$636a55ce@news.free.fr>, rejoc  > <rejoc@FREEfree.fr> writes:  > 5 >> Phillip Helbig---remove CLOTHES to reply a crit : > >>> In article <44253127$0$27895$626a54ce@news.free.fr>, rejoc  >>> <rejoc@FREEfree.fr> writes:  >>> L >>>> could be bad negotiation (half/full duplex) between the NIC and switch / >>>> on one (or more) systems in the cluster... J >>>> It can cause nodes disappearing/coming back in a cluster under heavy  >>>> network load.M >>> OK, but is this something one would expect to have changed after a power  
 >>> cycle? >>> > >> If the connections are autonegocatied, yes, I've seen such ( >> mis-negociation after a power off/on. >>F >> That's why I always force the configuration of NIC and switch port. >>C >> You can check the configuration with LANCP on the system side...  > ' > Sounds interesting; where do I start?  >  $ mc lancp show conf> will show you the speed and duplex currently setup on the NIC.  D For the switch, if it is manageable you should be able to check the  configuration.  - Check this on all the systems in the cluster.    ------------------------------    Date: 25 Mar 2006 01:30:34 -0800' From: "toby" <toby@telegraphics.com.au> P Subject: Re: The Minimum You Need to Know to Be an OpenVMS Application DeveloperB Message-ID: <1143279034.166621.93930@t31g2000cwb.googlegroups.com>   Christopher Browne wrote: O > Clinging to sanity, "toby" <toby@telegraphics.com.au> mumbled into her beard:  > > davidc@montagar.com wrote:G > >> >From the description, this looks like an excellent book for a new M > >> OpenVMS Programmer.  Thanks for the effort in putting this bok together.  > > J > > It does look interesting. As a hobbyist I wish I could justify the $90? > > - but coincidentally my day job is currently building MySQL H > > applications on other operating systems. It's fascinating that MySQLI > > has made it to VMS and is considered a useful component there (too) - , > > whether one approves of MySQL or not ;-) > B > I have a hard time fathoming how MySQL would be of interest on aD > platform that has a full-fledged ISAM system, RMS, *built into theC > operating system*, but I suppose there are all types out there...   G MySQL is a rather different animal. For one thing, it offers SQL out of G the box. Its popularity elsewhere means that skills can be transferred; G MySQL is easy to configure and maintain. Web applications built for the @ LAMP stack become portable to VMS. MySQL offers transactions andA replication (and v5 includes triggers, stored procedures, and the ? kitchen sink). Just a few things that come immediately to mind.    > --' > output = ("cbbrowne" "@" "ntlug.org") , > http://linuxdatabases.info/info/rdbms.htmlH > "I heard that if you play  the Windows CD  backward, you get a satanicH > message. But that's  nothing compared to  when you play it forward: It) > installs Windows...." -- G. R. Gaudreau    ------------------------------  % Date: Sat, 25 Mar 2006 15:50:04 +0100 6 From: Alexander Schreiber <als@usenet.thangorodrim.de>P Subject: Re: The Minimum You Need to Know to Be an OpenVMS Application Developer? Message-ID: <slrne2am4s.2js.als@mordor.angband.thangorodrim.de>   & toby <toby@telegraphics.com.au> wrote: >  > Christopher Browne wrote: P >> Clinging to sanity, "toby" <toby@telegraphics.com.au> mumbled into her beard: >> > davidc@montagar.com wrote: H >> >> >From the description, this looks like an excellent book for a newN >> >> OpenVMS Programmer.  Thanks for the effort in putting this bok together. >> >K >> > It does look interesting. As a hobbyist I wish I could justify the $90 @ >> > - but coincidentally my day job is currently building MySQLI >> > applications on other operating systems. It's fascinating that MySQL J >> > has made it to VMS and is considered a useful component there (too) -- >> > whether one approves of MySQL or not ;-)  >>C >> I have a hard time fathoming how MySQL would be of interest on a E >> platform that has a full-fledged ISAM system, RMS, *built into the D >> operating system*, but I suppose there are all types out there... > I > MySQL is a rather different animal. For one thing, it offers SQL out of I > the box. Its popularity elsewhere means that skills can be transferred; I > MySQL is easy to configure and maintain. Web applications built for the B > LAMP stack become portable to VMS. MySQL offers transactions andC > replication (and v5 includes triggers, stored procedures, and the A > kitchen sink). Just a few things that come immediately to mind.   D It also doesn't believe in data integrity and has very dubious ideasD about SQL and what a RDBMS is supposed to do. Just google for "MySQL	 gotchas".    Regards,        Alex. --  M What they (MySQL) lose in usability, they gain back in benchmarks, and that's N all that matters: getting the wrong answer really fast.  -- Randal L. Schwartz   ------------------------------  % Date: Sat, 25 Mar 2006 10:22:04 -0500 / From: "Steven N. Hirsch" <shirsch@adelphia.net> P Subject: Re: The Minimum You Need to Know to Be an OpenVMS Application Developer; Message-ID: <OMqdnbLsv5yA_bjZnZ2dnUVZ_vmdnZ2d@adelphia.com>    David J. Dachtera wrote:  F >>> David Turner at Island is probably single handedly responsible forF >>> putting more cheap Alpha boxes in the hands of home users than anyI >>> other vendor. He's got a bunch of DS10ls going for $89 dollars at the  >>> moment.   I Really? If that's the case they are well-hidden on Island's web site.  I  A just spent 5 minutes poking around and found nothing of the sort.    ------------------------------    Date: 25 Mar 2006 08:48:53 -08009 From: "d_gillbilly@hotmail.com" <d_gillbilly@hotmail.com> ; Subject: VMS and responsible computing. Just Ask The Vendor B Message-ID: <1143305333.584421.24380@z34g2000cwc.googlegroups.com>  4 When you ask the questions, what answers do you get?  
 Mr Vendor,  9 Who should take responsibility for securing the internet?   2 You for selling me defective/incomplete solutions?  5 Or me for buying your defective/incomplete solutions?   , If it is your responsibility, please fix it!  B If it is my responsibility, please provide me the tools to fix it!      VMS: Providing Alternatives,          D. Gillespie         In between games         Nova Scotia   , If you don't know who should be responsible,   Just Ask The Vendor   5 Ask them if VMS could be a responsible home solution.     Then ask them if you can buy it.   ------------------------------    Date: 25 Mar 2006 09:02:01 -08009 From: "d_gillbilly@hotmail.com" <d_gillbilly@hotmail.com> 4 Subject: VMS Securing Delivery - Just Ask The VendorB Message-ID: <1143306121.062542.79000@z34g2000cwc.googlegroups.com>   Just ask the questions!   
 Mr Vendor,  1 I know business is important, but so is my house.   E I see lots of new wonderful technology, but nothing seems to help me.   ) Can we change the way the internet works?   F Could you provide me with a secure, reliable, dependable appliance for my home?  = Do secure malware and email servers exist for this appliance?   F If the computer was secure, would it be possible to secure my network?  @ If I ran secure services in my home, would that help to make the internet more secure?   3 How long would it take you to build this appliance? 
    [ ] Now    [ ] 6 months 
    [ ] 1 year     [ ] 2 years    [ ] it is impossible   =    VMS: Yesterday's technology providing tomorrows solutions,             D. Gillespie 6            Inspired by youthful and passionate hockey,            Cape Breton  ; If you think it is time for more control over the internet,   - Just Ask The Vendor about VMS home solutions.   E More Ideas? Please give them to the vendor. They don't appear to have  any.  C With the wife driving, I still have time for a beer before the next  game.  I'm not mad anymore, but I'm still plenty angry!    ------------------------------  % Date: Sat, 25 Mar 2006 13:29:27 -0500 3 From: "Richard B. Gilbert" <rgilbert88@comcast.net> 8 Subject: Re: VMS Securing Delivery - Just Ask The Vendor0 Message-ID: <JOCdnZo4Ue6VEbjZRVn-jQ@comcast.com>   Bill Gunshannon wrote:  C > Is it just me or do others find this much more annoying than Dave  > announcing the new VMS book? >  > bill > D > In article <1143306121.062542.79000@z34g2000cwc.googlegroups.com>,> > 	"d_gillbilly@hotmail.com" <d_gillbilly@hotmail.com> writes: >  >>Just ask the questions!  >> >>Mr Vendor, >>3 >>I know business is important, but so is my house.  >>
 <big snip>  F It's MORE annoying because it's not clear, to me anyway, what all the  drum beating is about.   ------------------------------    Date: 25 Mar 2006 09:59:11 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) ) Subject: Where do DSA devices come from ? 3 Message-ID: <jCBL572c4pOQ@eisner.encompasserve.org>   > On an Alpha system with the latest copy of VMS I have verifyedF that the VOLSHAD license is present and the SHADOWING system parameter has a value of 2 after reboot.  H But I see no DSA device, and attempting a mount does not see one either.   ------------------------------    Date: 25 Mar 2006 08:06:30 -08003 From: "Ralf Gaertner" <ralf.gaertner@t-systems.com> - Subject: Re: Where do DSA devices come from ? B Message-ID: <1143302790.196510.96330@t31g2000cwb.googlegroups.com>   Larry Kilgallen schrieb:  @ > On an Alpha system with the latest copy of VMS I have verifyedH > that the VOLSHAD license is present and the SHADOWING system parameter  > has a value of 2 after reboot. > J > But I see no DSA device, and attempting a mount does not see one either.  B Did you set a nonzero ALLOCLASS parameter? Which MOUNT command didE issue? What is the error message if you perform the comand manually ?    Ralf   ------------------------------  # Date: Sat, 25 Mar 2006 16:29:29 GMT " From:   VAXman-  @SendSpamHere.ORG- Subject: Re: Where do DSA devices come from ? 0 Message-ID: <00A53387.D7CE90BC@SendSpamHere.ORG>  x In article <1143302790.196510.96330@t31g2000cwb.googlegroups.com>, "Ralf Gaertner" <ralf.gaertner@t-systems.com> writes: >  >  >  >Larry Kilgallen schrieb:  > A >> On an Alpha system with the latest copy of VMS I have verifyed I >> that the VOLSHAD license is present and the SHADOWING system parameter ! >> has a value of 2 after reboot.  >>K >> But I see no DSA device, and attempting a mount does not see one either.  > C >Did you set a nonzero ALLOCLASS parameter? Which MOUNT command did F >issue? What is the error message if you perform the comand manually ?  H Zero ALLOCLASS, I have found, is one of the major flaws people encounter setting up volume shadowing.   --  K VAXman- A Bored Certified VMS Kernel Mode Hacker   VAXman(at)TMESIS(dot)COM              5   "Well my son, life is like a beanstalk, isn't it?"     ------------------------------  + Date: Sat, 25 Mar 2006 16:51:14 +0000 (UTC) 7 From: moroney@world.std.spaamtrap.com (Michael Moroney) - Subject: Re: Where do DSA devices come from ? ( Message-ID: <e03se2$quq$2@pcls4.std.com>  / Kilgallen@SpamCop.net (Larry Kilgallen) writes:   ? >On an Alpha system with the latest copy of VMS I have verifyed G >that the VOLSHAD license is present and the SHADOWING system parameter  >has a value of 2 after reboot.   I >But I see no DSA device, and attempting a mount does not see one either.   H There are no DSA devices present until you create and mount a shadowset.) What do you mean by "attempting a mount"?   C To create and mount a new 1 member shadowset from an existing disk  
 $n$ddcu: try:   F $ MOUNT DSAn:/SHADOW=$n$ddcu: label, and a DSAn: device will magically appear, mounted.   ------------------------------    Date: 25 Mar 2006 10:55:57 -0600- From: Kilgallen@SpamCop.net (Larry Kilgallen) - Subject: Re: Where do DSA devices come from ? 3 Message-ID: <0nDJYVQX5odd@eisner.encompasserve.org>   U In article <00A53387.D7CE90BC@SendSpamHere.ORG>,   VAXman-  @SendSpamHere.ORG writes: z > In article <1143302790.196510.96330@t31g2000cwb.googlegroups.com>, "Ralf Gaertner" <ralf.gaertner@t-systems.com> writes: >> >> >> >>Larry Kilgallen schrieb: >>B >>> On an Alpha system with the latest copy of VMS I have verifyedJ >>> that the VOLSHAD license is present and the SHADOWING system parameter" >>> has a value of 2 after reboot. >>> L >>> But I see no DSA device, and attempting a mount does not see one either. >>D >>Did you set a nonzero ALLOCLASS parameter? Which MOUNT command didG >>issue? What is the error message if you perform the comand manually ?  > J > Zero ALLOCLASS, I have found, is one of the major flaws people encounter > setting up volume shadowing.  B My problem turned out to be that I had specified /SHADOW after theA volume label, but things worked when I moved it before the volume @ label.  That violates my understanding of how DCL commands work. Oh well.   ------------------------------  % Date: Sat, 25 Mar 2006 13:07:49 -0500 3 From: "Richard B. Gilbert" <rgilbert88@comcast.net> - Subject: Re: Where do DSA devices come from ? : Message-ID: <Z7SdnSp-XZZoG7jZnZ2dnUVZ_uudnZ2d@comcast.com>   Larry Kilgallen wrote:  @ > On an Alpha system with the latest copy of VMS I have verifyedH > that the VOLSHAD license is present and the SHADOWING system parameter  > has a value of 2 after reboot. > J > But I see no DSA device, and attempting a mount does not see one either.  H The DSA device is a "logical" or "pseudo" device and ISTR is created by F   MOUNT /SHADOW.  Sorry, but it's been about eight years since I last I had a volume shadowing license available to play with or I might be able  I to tell you more.  The DSA device represents the disks in the shadow set.    ------------------------------  % Date: Sat, 25 Mar 2006 10:47:10 -0800 # From: "Tom Linden" <tom@kednos.com> - Subject: Re: Where do DSA devices come from ? ) Message-ID: <op.s6zfswt8zgicya@hyrrokkin>   8 On Sat, 25 Mar 2006 10:07:49 -0800, Richard B. Gilbert   <rgilbert88@comcast.net> wrote:    > Larry Kilgallen wrote: > A >> On an Alpha system with the latest copy of VMS I have verifyed I >> that the VOLSHAD license is present and the SHADOWING system parameter ! >> has a value of 2 after reboot. F >>  But I see no DSA device, and attempting a mount does not see one  
 >> either. > K > The DSA device is a "logical" or "pseudo" device and ISTR is created by   I >   MOUNT /SHADOW.  Sorry, but it's been about eight years since I last   L > had a volume shadowing license available to play with or I might be able  K > to tell you more.  The DSA device represents the disks in the shadow set.   + You will see something like with show dev d   L DSA2:                   Mounted              0  DECPLI       123811944       1   5   A $42$DKA1000:   (FREJA)  ShadowSetMember      0  (member of DSA2:) A $42$DKA1100:   (FREJA)  ShadowSetMember      0  (member of DSA2:)    ------------------------------   End of INFO-VAX 2006.168 ************************                                                                                                                                                                                                          qUj%eSa]}Fj(_]sX)H\r#Xߕu03bw=t{²ǞހX,o_Q(%(?~yf/kP86	Mõ>2}p{:籧ccʿ'}F1r%>ho7ʊLK3˛|쯡|wKxH_{kzq}}Xxx>4[ETG /b}X6W%I=xa#{#퍸zgo?hfO>'?EwBo>oewBO(7Z/.O%io!17>{ϑ!~D˄?~c<*49c)^ڛ;/r`G"{VNyt2h7N^`P̭7T~p\זO?T|v>~y/ߒ[
\v}I)>i_>v ![}|"Nkr'rk
|,n[U
qxdCφk)uW8ˢ> {S瑟-
~Lwo&i.Z=| =O7P(.An],>1_}>o}
~$L
<	[lO_qU|׾J}crvu6?HUx6o&ӿː0yPW<o,	\|%lэ.<Ms^?>#߹~í /f_E~&7/WѦ~'A5'H*"ϻ\V['8OH!L?ch i~[CecıNB`Mq"M{"k{SҫM9 R(B*-R望.