1 INFO-VAX	Tue, 08 Mar 2005	Volume 2005 : Issue 133       Contents:1 1997's LAND attack can DOS WIN-XP and 2003 server  Bono vs. Carley  Re: Bono vs. Carley  Re: Bono vs. Carley  Re: Bono vs. Carley 0 Re: Build problem - GD (but we're almost there!)  Re: Change MAC address on DE500?  Re: Change MAC address on DE500?  Re: DECC floating point question  Re: DECC floating point question  Re: DECC floating point question Re: FT Linux - five 9's  Re: FT Linux - five 9's  Re: FT Linux - five 9's  RE: FT Linux - five 9's  Re: FT Linux - five 9's  Re: FT Linux - five 9's  Re: FT Linux - five 9's  Re: FT Linux - five 9's * Re: Has anyone had any bad T4 experiences? Intel and high-k& Re: new Itanium after Tukwila: Poulson Re: Oracle RDBE Re: PCSI product install failure - DDIS-E-TNF, invalid element syntax E Re: PCSI product install failure - DDIS-E-TNF, invalid element syntax  Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL  Re: Sending SMTP e-mail from DCL Re: Shannon opines on HP Re: Shavlik Randolph Fan Club  Shavlik Randolph Fan Club  Shavlik Randolph Fan Club , Re: Sony vs HP: a comparison of CEO oustings, Re: Sony vs HP: a comparison of CEO oustings, RE: Sony vs HP: a comparison of CEO oustings, Re: Sony vs HP: a comparison of CEO oustings2 Re: Starting JAVA after Upgrading to OpenVMS 7.3.2M Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers M Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers M Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers P Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers trP Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers tr Re: VMS startup problem? Re: VMS startup problem?< Re: What is the Difference between Shadow and Mirrored disk?  F ----------------------------------------------------------------------   Date: 7 Mar 2005 13:51:33 -0800  From: jordan@ccs4vms.com: Subject: 1997's LAND attack can DOS WIN-XP and 2003 serverC Message-ID: <1110232293.265029.197830@l41g2000cwc.googlegroups.com>    Per the article ( G http://www.securityfocus.com/archive/1/392354/2005-03-02/2005-03-08/0 ) @ and subsequent slashdot yammering it looks like some versions of@ win2003 server and XPSP2 (and perhaps SP1a according to a coupleE slashdotters) can be DOS'd by the LAND attack.  If the local software * firewall is enabled the attack is blocked.  G I had to dig up some old articles to find out about LAND; our firewalls * still list it among the 'base' protection.  > Those of you with such boxes on your networks may wish to takeF appropriate steps, or if sledgehammers and chainsaws aren't allowed in5 the peecee correctional process, at least be aware...    ------------------------------  $ Date: Mon, 7 Mar 2005 21:27:32 +0100, From: "Dr. Dweeb" <5msg0h202@sneakemail.com> Subject: Bono vs. Carley- Message-ID: <d0idfj$12fc$1@news.cybercity.dk>   G The race for World Bank president is hotting up with noted minstrel and K heavyweight intellectual in the field of international affairs and banking, * U2s own Bono, being mooted as a contender.D http://money.cnn.com/2005/02/26/news/newsmakers/bc.people.bono.reut/  	 Dr. Dweeb    ------------------------------  $ Date: Mon, 7 Mar 2005 15:46:57 -0500# From: "John Smith" <a@nonymous.com>  Subject: Re: Bono vs. Carley, Message-ID: <xomdnTfReIVZILHfRVn-2Q@igs.net>   Dr. Dweeb wrote:E > The race for World Bank president is hotting up with noted minstrel D > and heavyweight intellectual in the field of international affairs9 > and banking, U2s own Bono, being mooted as a contender. F > http://money.cnn.com/2005/02/26/news/newsmakers/bc.people.bono.reut/ >  > Dr. Dweeb     K How about Martha Stewart?.....she can tell all those developing nations how F to look better, help them with their hadicraft industries, and there'sL nothing that makes a good first impression of a country quite like the smellD of home-cooking when you step off the plane upon your arrival at the airport. passport in-hand.   --F OpenVMS    - The often imitated but never advertised operating system.C R.I.P. Alpha - I could have had an EV8 but all I got was this lousy  Itanic.    ------------------------------  % Date: Mon, 07 Mar 2005 13:10:11 -0800 , From: Ken Fairfield <my.full.name@intel.com> Subject: Re: Bono vs. Carley+ Message-ID: <d0ifvj$q0r$1@news01.intel.com>    John Smith wrote:  > Dr. Dweeb wrote: > E >>The race for World Bank president is hotting up with noted minstrel D >>and heavyweight intellectual in the field of international affairs9 >>and banking, U2s own Bono, being mooted as a contender. F >>http://money.cnn.com/2005/02/26/news/newsmakers/bc.people.bono.reut/ >> >>Dr. Dweeb  >  >  > M > How about Martha Stewart?.....she can tell all those developing nations how H > to look better, help them with their hadicraft industries, and there'sN > nothing that makes a good first impression of a country quite like the smellF > of home-cooking when you step off the plane upon your arrival at the > airport. passport in-hand.  > The difference being that Bono has the bona fides for the job.7 Martha Stewart has only celebrity and business success.   ; Either would be vetoed by the US on the basis of being able = to think for themselves, not being Republicans, and not being 8 loyal to El Presidente.  That last one, along with being; competent in <any> endeavor, more than disqualifies them...   	      -Ken  --  6 I don't speak for Intel, Intel doesn't speak for me...  
 Ken Fairfield ! D1C Automation VMS System Support " who:   kenneth dot h dot fairfield where: intel dot com   ------------------------------  $ Date: Mon, 7 Mar 2005 17:01:54 -0500# From: "John Smith" <a@nonymous.com>  Subject: Re: Bono vs. Carley, Message-ID: <U9adnTKCC_TIUrHfRVn-pQ@igs.net>   Ken Fairfield wrote: > John Smith wrote:  >> Dr. Dweeb wrote:  >>G >>> The race for World Bank president is hotting up with noted minstrel F >>> and heavyweight intellectual in the field of international affairs; >>> and banking, U2s own Bono, being mooted as a contender. H >>> http://money.cnn.com/2005/02/26/news/newsmakers/bc.people.bono.reut/ >>> 
 >>> Dr. Dweeb  >> >> >>B >> How about Martha Stewart?.....she can tell all those developing= >> nations how to look better, help them with their hadicraft E >> industries, and there's nothing that makes a good first impression F >> of a country quite like the smell of home-cooking when you step off@ >> the plane upon your arrival at the airport. passport in-hand. > @ > The difference being that Bono has the bona fides for the job.9 > Martha Stewart has only celebrity and business success.  > = > Either would be vetoed by the US on the basis of being able ? > to think for themselves, not being Republicans, and not being : > loyal to El Presidente.  That last one, along with being= > competent in <any> endeavor, more than disqualifies them...     G I don't think you'll make Ashcroft's Christmas card list this year  :-)    --F OpenVMS    - The often imitated but never advertised operating system.  D Alpha - I could have had an EV8 but all I got was this lousy Itanic.   ------------------------------   Date: 7 Mar 2005 16:13:32 -0800  From: mcbill20@yahoo.com9 Subject: Re: Build problem - GD (but we're almost there!) C Message-ID: <1110240812.393514.150110@l41g2000cwc.googlegroups.com>    JOUKJ wrote: > mcbill20@yahoo.com wrote: G > > Thanks. That worked. Taking out all the "#" lines allowed the build  to@ > > complete. I now just need to get the same thing to work with
 GDCHART. IF > > am having some issues with the GD library software though. The exe > > directory now contains:  > > 9 > > ANNOTATE.EXE;1          2344   4-MAR-2005 22:54:59.72  > > CIRCLETEXTTEST.EXE;19 > >                         3203   4-MAR-2005 22:55:14.84 9 > > DEMOOUT.GIF;2              9   4-MAR-2005 23:15:33.61 9 > > DEMOOUT.GIF;1              9   4-MAR-2005 23:09:45.06 9 > > DEMOOUT.PNG;2             18   4-MAR-2005 23:15:33.13 9 > > DEMOOUT.PNG;1             18   4-MAR-2005 23:09:44.63 9 > > DEMOOUTP.PNG;2            11   4-MAR-2005 23:15:33.89 9 > > DEMOOUTP.PNG;1            11   4-MAR-2005 23:09:45.30 9 > > FONTSIZETEST.EXE;1      2563   4-MAR-2005 22:55:17.33  > > FONTSIZETEST1.PNG;2 9 > >                            1   4-MAR-2005 23:21:04.21  > > FONTSIZETEST1.PNG;1 9 > >                            1   4-MAR-2005 23:14:09.28  > > FONTSIZETEST2.PNG;2 9 > >                            1   4-MAR-2005 23:21:04.70  > > FONTSIZETEST2.PNG;1 9 > >                            1   4-MAR-2005 23:14:09.83  > > FONTSIZETEST3.PNG;2 9 > >                            1   4-MAR-2005 23:21:05.17  > > FONTSIZETEST3.PNG;1 9 > >                            1   4-MAR-2005 23:14:10.29  > > FONTSIZETEST4.PNG;2 9 > >                            1   4-MAR-2005 23:21:05.65  > > FONTSIZETEST4.PNG;1 9 > >                            1   4-MAR-2005 23:14:10.75  > > FONTWHEELTEST.EXE;1 9 > >                         2565   4-MAR-2005 22:55:19.81 9 > > GD2COPYPAL.EXE;1         363   4-MAR-2005 22:55:01.89 9 > > GD2TIME.EXE;1            363   4-MAR-2005 22:55:21.96 9 > > GD2TOPNG.EXE;1          1021   4-MAR-2005 22:55:03.54 9 > > GDDEMO.EXE;1            1281   4-MAR-2005 22:55:24.15 9 > > GDPARTTOPNG.EXE;1       1021   4-MAR-2005 22:55:05.32 9 > > GDTEST.EXE;1            1680   4-MAR-2005 22:55:27.39 9 > > GDTESTFT.EXE;1          3229   4-MAR-2005 22:55:29.93 9 > > GDTOPNG.EXE;1            991   4-MAR-2005 22:55:07.05 9 > > PNGTOGD.EXE;1            991   4-MAR-2005 22:55:08.79 9 > > PNGTOGD2.EXE;1          1020   4-MAR-2005 22:55:10.57 9 > > TESTAC.EXE;1            1025   4-MAR-2005 22:55:32.51 9 > > WEBPNG.EXE;1             989   4-MAR-2005 22:55:12.95  > > F > > However, when I try tests (using GDDEMO, GDTEST or WEBPNG) I don't get D > > valid images. Actually, none of the existing PNG files that came withG > > the distribution seem to be valid either. I copied them to a PC and ? > > tried to view them but I get "drawing failed" with the .JPG 	 conersion G > > and "Could not complete this operation. The internal file signature  isE > > incorrect." with the .PNG files. This seems odd but then again, I  don't G > > know much about the GD software yet. I'll poke around some more and  > > also try building GDCHART. > >  > > Bill > > G > Strange. The sizes of the png files are wrong when I run gddemo on my , > system I get the following files created :' > Directory $DISK4:[JOUKJ.PUBLIC.gd.gd]  > " > DEMOOUT.GIF;1            192/192" > DEMOOUT.PNG;1            278/279" > DEMOOUTP.PNG;1           156/156 > # > Total of 3 files, 626/627 blocks.  >  > All 3 images look fine.  > D > If you have VMS7.3-2 I can packup my gd/gdchart directory for you. You  > can try my binaries. >  >  >  >             Jouk  E Thanks for the offer. Unfortunately I am still at 7.3-1 and am trying  to get hold of 7.3-2 CD's.     Bill   ------------------------------  $ Date: Mon, 7 Mar 2005 15:06:52 -05002 From: "Jonathan Boswell" <jsb.NOSP@M.cdrh.fda.gov>) Subject: Re: Change MAC address on DE500? 5 Message-ID: <xv2Xd.2763$Ny6.4399@mencken.net.nih.gov>   / "Crabs" <spamsucks@nospam.com> wrote in message ) news:iNCdnWWDyvte37TfRVn-3g@sunset.net... I > It is possible and actually quite simple to re-write the MAC address of J > the DE500 card.  What you need is the SROM programming tool kit from the4 > old Digital Semiconductor FTP site, here's a link:   Tom you genius!   H > Download and copy all the files to a Windows formatted bootable floppy  I You must mean format /s because this package is nearly 1MB all by itself, = not including various Windows utilities on my recovery disks.   H > disk. Place the DE500 card in a Windows 95/98/2000 system then boot toI > the floppy. Use SROMUTIL to read the existing information on the DE500,   L This fails with an "unable to access CSR9" error.  The utility does find theJ card and identifies it as a Digital Semiconductor 21143.  Got any ideas on why I can't access CSR9?   ------------------------------  % Date: Mon, 07 Mar 2005 13:32:10 -0800 # From: Tom Crabtree <spam@sucks.com> ) Subject: Re: Change MAC address on DE500? / Message-ID: <N_SdnWrKiKRxWrHfRVn-tA@sunset.net>   
 Johnathan: > K > You must mean format /s because this package is nearly 1MB all by itself, ? > not including various Windows utilities on my recovery disks.  Yes. >  > H >>disk. Place the DE500 card in a Windows 95/98/2000 system then boot toI >>the floppy. Use SROMUTIL to read the existing information on the DE500,  >  > N > This fails with an "unable to access CSR9" error.  The utility does find theL > card and identifies it as a Digital Semiconductor 21143.  Got any ideas on > why I can't access CSR9? > 8 I don't have a clue.  I've never seen this error before.E Where are you located?  I'm on the "Left Coast". ;-) (Sacramento, CA) - Contact me off-line, tomc@sunset*no_spam*.net  remove the obvious spam munge.E I can either program your card for you or send you a DE450 with your  A address on it. I've got lots of DE450's, but no DE500's to spare.    TomC   ------------------------------  # Date: Mon, 07 Mar 2005 19:40:15 GMT & From: John Reagan <john.reagan@hp.com>) Subject: Re: DECC floating point question 2 Message-ID: <z62Xd.1171$hB1.1118@news.cpqcorp.net>   Dave Froble wrote:   > J > Has my memory failed me again?  I thought that D-float was converted to > > IEEE float for any colculations.  Is it G-float that's used?  C G float, not IEEE float.  In particular, there are CVTDG and CVTGD  0 instructions designed to perform the conversion.  F On Alpha, unless you explicitly ask for IEEE float, you will only use + VAX F, D, and G.  X float is software only.    --   John Reagan / HP Pascal/{A|I}MACRO for OpenVMS Project Leader  Hewlett-Packard Company    ------------------------------  # Date: Mon, 07 Mar 2005 19:42:29 GMT & From: John Reagan <john.reagan@hp.com>) Subject: Re: DECC floating point question 1 Message-ID: <F82Xd.1172$hB1.502@news.cpqcorp.net>    briggs@encompasserve.org wrote:    > E > If my memory serves, D float is indeed converted to IEEE float for  < > calculations.  And you lose 3 bits of precision due to the
 > conversion.    Nope, D is converted to G.   > E > On VAX, you have the choice of running the C compiler in G floating A > mode or in D floating mode.  The default is D floating.  If you A > run in G floating mode, you want to be sure to link against the  > G floating run time library.  G You must be using a very old version of VAX C since you shouldn't have  ) to "know" which library you link against.    > ? > On Alpha, you also have a choice.  But, presumably due to the ? > poor hardware support for D floating, the default on Alpha is B > G floating.  I assume that want to take care to link against theD > D floating run time library if you choose to override the default.  H Again, what "D floating run time library" are you talking about?  There J is only on C RTL for instance.  You just link and the right thing happens.     --   John Reagan / HP Pascal/{A|I}MACRO for OpenVMS Project Leader  Hewlett-Packard Company    ------------------------------  % Date: Mon, 07 Mar 2005 12:54:21 -0800 # From: "Tom Linden" <tom@kednos.com> ) Subject: Re: DECC floating point question ( Message-ID: <opsnaccviyzgicya@hyrrokkin>  5 On Mon, 07 Mar 2005 20:51:35 +0100, Keith Cayemberg   ! <keith.cayemberg@arcor.de> wrote:   ! > briggs@encompasserve.org wrote: A >> In article <112ov83r9b5og1f@corp.supernews.com>, Dave Froble     >> <davef@tsoft-inc.com> writes: >> >>> John Reagan wrote: >>>  >>>> Dave Froble wrote:  >>>> >>>>I >>>>> Now that I got in my required language promotion, I'll guess that   J >>>>> your problem is using DOUBLE (8 byte floating point) on an Alpha.   L >>>>> There is no such thing.  Alpha converts such to IEEE floating point,  L >>>>> does the calculations, and then converts it back.  Guess what, there  ; >>>>> are 3 less bits of prescision in IEEE floating point.  >>>>>  >>>>F >>>> Alpha has hardware support for VAX F, VAX G, IEEE S, and IEEE T  J >>>> formats as well as a few instructions to help with VAX D floating.   G >>>> The C program on Alpha should use VAX F and VAX G out of the box   * >>>> without any conversions to/from IEEE. >>>> >>> J >>> Has my memory failed me again?  I thought that D-float was converted  C >>> to IEEE float for any colculations.  Is it G-float that's used? I >>   If my memory serves, D float is indeed converted to IEEE float for   = >> calculations.  And you lose 3 bits of precision due to the  >> conversion.G >>  On VAX, you have the choice of running the C compiler in G floating B >> mode or in D floating mode.  The default is D floating.  If youB >> run in G floating mode, you want to be sure to link against the >> G floating run time library. A >>  On Alpha, you also have a choice.  But, presumably due to the @ >> poor hardware support for D floating, the default on Alpha isC >> G floating.  I assume that want to take care to link against the E >> D floating run time library if you choose to override the default. G >>  By default, on Alpha, you get to use a fully supported data type --  >> G floating. >>  	John Briggs > I > In the Archived Documents there are several dealing with VAX to Alpha   % > porting and compatibility issues... - > http://h71000.www7.hp.com/doc/archived.html  > J > You will find the following useful table in the "OpenVMS Compatibility  ' > Between VAX and Alpha" Manual here... S > http://h71000.www7.hp.com/doc/73final/documentation/pdf/OVMS_COMPAT_VAX_ALPHA.pdf  >  > . > Table 4–1 Floating-Point Data Type Support > 0 > _Data Type_          _On VAX_       _On Alpha_ > H > D53_float            Not supported. Supported. Using D53_float insteadF > float) (Default                     of D56_float drops three bits ofC > double-precision                    precision and yields slightly 8 > format)                             different results. >  > H > D56_ float (Default  Supported.     Not supported. You can obtain fullF > double-precision                    support by translating your codeI > format)                             with DECmigrate. Alternatively, you C >                                      can substitute D53_float for E >                                      D56_float, if your application G >                                      does not require the extra three 9 >                                      bits of precision.  > 0 > F_float              Supported.     Supported. > 0 > G_float              Supported.     Supported. > H > H_float (128-        Supported.     Not supported. You can obtain fullJ > floating-point)                     H_float support with DECmigrate. YouG >                                      can use it to translate the code C >                                      module that contains H_float I >                                      structures, or you can recode your E >                                      application, using a supported 1 >                                      data type.  > 0 > S_float (IEEE)       Not supported. Supported. > 0 > T_float (IEEE)       Not supported. Supported. > J > X_float (128-        Not supported. Supported by DEC Fortran Version 6.2C > floating-point                      and by DEC C Version 4.0. The J > (IEEE))                             X_float data format is not identicalK >                                      to H_float, but both cover a similar C >                                      range of values. For Fortran I >                                      applications, automatic conversion H >                                      between X_float memory format andI >                                      H_float on-disk is possible by use J >                                      of the /CONVERT compiler qualifier,E >                                      or the CONVERT= option on OPEN 2 >                                      statements. > 	 > Cheers!  >  > Keith Cayemberg   C PL/I on both VAX and AXP support D53_float, but not D56.  On VAX it $ ignores the 3 least significant bitsH http://www.kednos.com/pli/docs/REFERENCE_MANUAL/6291pro_007.html#data_22G if you try to specify a precision greater than 53, it will default to    H-Float.       --  C Using Opera's revolutionary e-mail client: http://www.opera.com/m2/    ------------------------------  $ Date: Mon, 7 Mar 2005 15:23:02 -0500# From: "rob kas" <bob@paychoice.com>   Subject: Re: FT Linux - five 9's0 Message-ID: <112pe1jh5c98f74@corp.supernews.com>  J      Well you could combine this with the NSA's addon security module for  Linux        and have a VMS clone.   #         http://www.nsa.gov/selinux/     /                                             Rob       / "John Smith" <a@nonymous.com> wrote in message  & news:ALednX-C_srkK7HfRVn-uw@igs.net...8 > On a carrier-grade fault-tolerant Xeon-based box...... > H > Stratus offers a variation on the basic Linux kernel that the company  > calls G > "ft Linux," which adds to the base 2.4 kernel a layer of software for C > managing storage, memory and redundant-processing components in a K > fault-tolerant environment. Since the end result is a modified version of L > Linux, Stratus provides support for the operating system. The company saysK > it has been contributing code to the Linux community that will let future L > versions of the kernel provide better support for fault tolerance. StratusJ > hopes, by the end of the year, to build its ftLinux on top of a standardE > Linux distribution such as Red Hat or SuSE Linux, says Denny Lane,  
 > director > of product marketing.  >  > --H > OpenVMS    - The often imitated but never advertised operating system. > F > R.I.P. Alpha - I could have had an EV8 but all I got was this lousy 	 > Itanic.  >  >    ------------------------------  $ Date: Mon, 7 Mar 2005 15:43:33 -0500# From: "John Smith" <a@nonymous.com>   Subject: Re: FT Linux - five 9's, Message-ID: <2MOdnf5zzvllIbHfRVn-hQ@igs.net>   rob kas wrote:0 > "John Smith" <a@nonymous.com> wrote in message( > news:ALednX-C_srkK7HfRVn-uw@igs.net...9 >> On a carrier-grade fault-tolerant Xeon-based box......  >>H >> Stratus offers a variation on the basic Linux kernel that the company >> callsH >> "ft Linux," which adds to the base 2.4 kernel a layer of software forD >> managing storage, memory and redundant-processing components in aA >> fault-tolerant environment. Since the end result is a modified G >> version of Linux, Stratus provides support for the operating system. > >> The company says it has been contributing code to the LinuxG >> community that will let future versions of the kernel provide better F >> support for fault tolerance. Stratus hopes, by the end of the year,G >> to build its ftLinux on top of a standard Linux distribution such as 3 >> Red Hat or SuSE Linux, says Denny Lane, director  >> of product marketing. >  >  > G >      Well you could combine this with the NSA's addon security module  > for Linux  >       and have a VMS clone.  > % >         http://www.nsa.gov/selinux/  >  > 1 >                                             Rob      Not really..... ' http://www.nsa.gov/selinux/info/faq.cfm   & 17. Is it approved for government use?H Security-enhanced Linux is not part of any currently approved version ofK Linux and has no special or additional approval for government use over any  other version of Linux.   F OpenVMS    - The often imitated but never advertised operating system.  L  R.I.P. Alpha - I could have had an EV8 but all I got was this lousy Itanic.   ------------------------------  # Date: Mon, 07 Mar 2005 22:37:58 GMT . From: "robert kas" <rkas60@nospam.verizon.net>  Subject: Re: FT Linux - five 9's, Message-ID: <aJ4Xd.92173$g16.70072@trndny08>  
     Greetings   @   Prehaps not now but everything has to start with a first step.                             Rob      >  > Not really..... ) > http://www.nsa.gov/selinux/info/faq.cfm  > ( > 17. Is it approved for government use?J > Security-enhanced Linux is not part of any currently approved version ofJ > Linux and has no special or additional approval for government use over  > any  > other version of Linux.  >    ------------------------------  $ Date: Mon, 7 Mar 2005 18:17:03 -0500' From: "Main, Kerry" <kerry.main@hp.com>   Subject: RE: FT Linux - five 9'sR Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB594D46@tayexc19.americas.cpqcorp.net>   > -----Original Message------ > From: John Smith [mailto:a@nonymous.com]=20  > Sent: March 7, 2005 3:44 PM  > To: Info-VAX@Mvb.Saic.Com " > Subject: Re: FT Linux - five 9's >=20 > rob kas wrote:2 > > "John Smith" <a@nonymous.com> wrote in message* > > news:ALednX-C_srkK7HfRVn-uw@igs.net...; > >> On a carrier-grade fault-tolerant Xeon-based box......  > >>A > >> Stratus offers a variation on the basic Linux kernel that=20 
 > the company 
 > >> calls@ > >> "ft Linux," which adds to the base 2.4 kernel a layer of=20 > software forF > >> managing storage, memory and redundant-processing components in aC > >> fault-tolerant environment. Since the end result is a modified : > >> version of Linux, Stratus provides support for the=20 > operating system. @ > >> The company says it has been contributing code to the Linux= > >> community that will let future versions of the kernel=20  > provide betterH > >> support for fault tolerance. Stratus hopes, by the end of the year,7 > >> to build its ftLinux on top of a standard Linux=20  > distribution such as5 > >> Red Hat or SuSE Linux, says Denny Lane, director  > >> of product marketing. > >  > >  > > < > >      Well you could combine this with the NSA's addon=20 > security module 
 > > for Linux  > >       and have a VMS clone.  > > ' > >         http://www.nsa.gov/selinux/  > >  > > 3 > >                                             Rob  >=20 >=20 > Not really..... ) > http://www.nsa.gov/selinux/info/faq.cfm  >=20( > 17. Is it approved for government use?B > Security-enhanced Linux is not part of any currently approved=20 > version of8 > Linux and has no special or additional approval for=20 > government use over any  > other version of Linux.  >=20    D Well, my personal thoughts are that one of the challenges faced withH layering FT or clustering software on top of an OS *that you do not own*8 is that whenever mandatory or critical or even importantG kernel/security/ driver patches are released, there is a potential that 0 these patches will break the ISV add-on package.  G The ISV can have specific views on what they support, but that does not H help much with many Customers when it comes to mandatory and/or security patches.  G That is certainly one of the big issues faced by the Eng group that did D the Digital Windows NT clustering package in the old days before theF Microsoft "Wolf pack" cluster product became widely adopted (for those that were around back then).=20   B Cust would add some mandatory or security fix and then the DigitalC Windows NT Clustering software would break e.g. something would not B fail-over or fail-back properly. And of course, after applying theE kernel patch, they would not test the cluster software so when it was * needed a few days later, the fun began.=20   Finger pointing galore.    :-)    Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  $ "OpenVMS has always had integrity .. Now, Integrity has OpenVMS .."   ------------------------------  $ Date: Mon, 7 Mar 2005 18:55:34 -0500# From: "John Smith" <a@nonymous.com>   Subject: Re: FT Linux - five 9's, Message-ID: <WcKdnVw-dpFkdLHfRVn-uw@igs.net>   robert kas wrote:  >     Greetings  > B >   Prehaps not now but everything has to start with a first step.    L Yeah.....like one full-page ad for VMS in the WSJ with no mention of Itanic.   --F OpenVMS    - The often imitated but never advertised operating system.  4 I could have had an EV8 but all I got was this lousy Itanic.    ------------------------------   Date: 8 Mar 2005 01:56:18 GMT ( From: bill@cs.uofs.edu (Bill Gunshannon)  Subject: Re: FT Linux - five 9's, Message-ID: <394f22F5t1o61U1@individual.net>  0 In article <112pe1jh5c98f74@corp.supernews.com>,& 	"rob kas" <bob@paychoice.com> writes:L >      Well you could combine this with the NSA's addon security module for  > Linux   H Thus the reason for listing Red Hat as one of the future homes.  Red Hat; already ships an SE version that includes all the NSA work.    bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------   Date: 8 Mar 2005 02:00:34 GMT ( From: bill@cs.uofs.edu (Bill Gunshannon)  Subject: Re: FT Linux - five 9's, Message-ID: <394fa2F5t1o61U2@individual.net>  , In article <2MOdnf5zzvllIbHfRVn-hQ@igs.net>,& 	"John Smith" <a@nonymous.com> writes: > rob kas wrote:1 >> "John Smith" <a@nonymous.com> wrote in message ) >> news:ALednX-C_srkK7HfRVn-uw@igs.net... : >>> On a carrier-grade fault-tolerant Xeon-based box...... >>> I >>> Stratus offers a variation on the basic Linux kernel that the company 	 >>> calls I >>> "ft Linux," which adds to the base 2.4 kernel a layer of software for E >>> managing storage, memory and redundant-processing components in a B >>> fault-tolerant environment. Since the end result is a modifiedH >>> version of Linux, Stratus provides support for the operating system.? >>> The company says it has been contributing code to the Linux H >>> community that will let future versions of the kernel provide betterG >>> support for fault tolerance. Stratus hopes, by the end of the year, H >>> to build its ftLinux on top of a standard Linux distribution such as4 >>> Red Hat or SuSE Linux, says Denny Lane, director >>> of product marketing.  >> >> >>H >>      Well you could combine this with the NSA's addon security module >> for Linux >>       and have a VMS clone. >>& >>         http://www.nsa.gov/selinux/ >> >>2 >>                                             Rob >  >  > Not really..... ) > http://www.nsa.gov/selinux/info/faq.cfm  > ( > 17. Is it approved for government use?J > Security-enhanced Linux is not part of any currently approved version ofM > Linux and has no special or additional approval for government use over any  > other version of Linux.  >   D Just because it has "no special or additional approval" doesn't meanD it is unapproved.  The government (or at least that part of DOD that? I am currently dealing with) is going Linux in a big way to the B detriment of Solaris which used to be the chosen son.  (Not that IB necessarily agree with the move, only stating it as a fact and one I need to learn to deal with.)     bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  $ Date: Mon, 7 Mar 2005 21:13:28 -0500# From: "John Smith" <a@nonymous.com>   Subject: Re: FT Linux - five 9's, Message-ID: <pomdnc4TdZKklrDfRVn-gA@igs.net>   Main, Kerry wrote: >> -----Original Message----- + >> From: John Smith [mailto:a@nonymous.com]  >> Sent: March 7, 2005 3:44 PM >> To: Info-VAX@Mvb.Saic.Com# >> Subject: Re: FT Linux - five 9's  >> >> rob kas wrote: 2 >>> "John Smith" <a@nonymous.com> wrote in message* >>> news:ALednX-C_srkK7HfRVn-uw@igs.net...; >>>> On a carrier-grade fault-tolerant Xeon-based box......  >>>>> >>>> Stratus offers a variation on the basic Linux kernel that >> the company
 >>>> calls= >>>> "ft Linux," which adds to the base 2.4 kernel a layer of  >> software for F >>>> managing storage, memory and redundant-processing components in aC >>>> fault-tolerant environment. Since the end result is a modified 7 >>>> version of Linux, Stratus provides support for the  >> operating system.@ >>>> The company says it has been contributing code to the Linux: >>>> community that will let future versions of the kernel >> provide better H >>>> support for fault tolerance. Stratus hopes, by the end of the year,4 >>>> to build its ftLinux on top of a standard Linux >> distribution such asn5 >>>> Red Hat or SuSE Linux, says Denny Lane, director  >>>> of product marketing. >>>. >>>e >>>e9 >>>      Well you could combine this with the NSA's addonM >> security module
 >>> for LinuxC >>>       and have a VMS clone.t >>>o' >>>         http://www.nsa.gov/selinux/s >>>  >>>E3 >>>                                             Rob  >> >> >> Not really.....* >> http://www.nsa.gov/selinux/info/faq.cfm >>) >> 17. Is it approved for government use?f@ >> Security-enhanced Linux is not part of any currently approved
 >> version ofa6 >> Linux and has no special or additional approval for >> government use over any >> other version of Linux. >> >a >fF > Well, my personal thoughts are that one of the challenges faced withE > layering FT or clustering software on top of an OS *that you do notD? > own* is that whenever mandatory or critical or even importantmD > kernel/security/ driver patches are released, there is a potential7 > that these patches will break the ISV add-on package.m > E > The ISV can have specific views on what they support, but that doesmE > not help much with many Customers when it comes to mandatory and/or  > security patches.h >iE > That is certainly one of the big issues faced by the Eng group that F > did the Digital Windows NT clustering package in the old days beforeF > the Microsoft "Wolf pack" cluster product became widely adopted (for$ > those that were around back then). > D > Cust would add some mandatory or security fix and then the DigitalE > Windows NT Clustering software would break e.g. something would not5D > fail-over or fail-back properly. And of course, after applying theG > kernel patch, they would not test the cluster software so when it wast) > needed a few days later, the fun began./ >- > Finger pointing galore.i    ( So where's the carrier grade FT VMS box?     --F OpenVMS    - The often imitated but never advertised operating system.  =  I could have had an EV8 but all I got was this lousy Itanic.-   ------------------------------   Date: 7 Mar 2005 13:39:35 -08000" From: dave.baxter@bannerhealth.com3 Subject: Re: Has anyone had any bad T4 experiences? C Message-ID: <1110231575.475466.196970@f14g2000cwb.googlegroups.com>   D I wouldn't say "bad experiences".    In my experience, (granted, notA very deep), T4 is pretty good for "broad brush"-type reporting onl@ systems, however it doesn't really provide any granularity (i.e.A process level information).   In it's favour, it should should be.- pointed out that the over-head is pretty low.h  @ I must take the opportunity to support Kerry Main's plug for CA.@ They have a much improved attitude to VMS and have significantlyG overhauled their pricing structure.    We recently renewed out licenses C for CA's Unicenter product and even though we extended its use to anE whole other cluster, the renewal costs were only ~40% of the previous- cost.-  G Also, for those of us who got used to having DECPS to work with (in the05 good old Digital Days), it was worth working with CA.-  D Basically, Unicenter gives me the process-level granularity required0 for retro-active troubleshooting.   T4 does not.   Dave.e   David J Dachtera wrote:t > norm.raphael@metso.com wrote:  > >wD > > David J Dachtera <djesys.nospam@comcast.net> wrote on 03/01/2005 08:40:08 > > PM:p > >. > > > Beach Runner wrote:  > > > >- > > > > David J Dachtera wrote:, > > > >0 > > > > > Thomas Wirt wrote:	 > > > > >oF > > > > >>I am preparing to install T4 on my AS 4100's (VMS 7.3-1) and was2G > > > > >>wondering what experiences anyone else had had.  Does it worktA > > > > >>seamlessly?  Any crashes or performance hits?  Is it as  useful > > > as it looks?	 > > > > >a	 > > > > >oF > > > > > Trouble with T4 is, its a collector - ONLY. The truly useful bit -n > > the/D > > > > > reporting and analysis pieces - remain MIA (ever since the	 demise of  > > > > > PSPA and SPM).	 > > > > >  > > > > C > > > > No engineering and Colorado Support have tools they can usen with the@ > > > > data.  Chances are if you have a problem you'll get them	 involved.  > > >uB > > > Well, my problem is management asking for ad-hoc performanceD > > > reports/analysis on zero notice, not to mention wanting to see themG > > > regularly. (I.e., they expect to output like we used to give thems fromG > > > SPM and/or PSPA regularly and on-demand). Can they help me there?1 > >0D > > Have you considered PAWZ from http://www.perfcap.com who did ECP before, > > they were exiled to the wilds of Nashua. > G > I looked at that URL and chanced upon mention of a PSPA$READ image. IsC > was intrigued. Last I saw of PAWZ, it was WhineBloze GUI only andu sorely
 > lacking. >rG > Is anyone using PAWZ's VMS-side reporting capability to emulate SPM'sr' > and/or PSPA's ASCII histogram output?s >wD > (Hint: Notice that I keep repeating "ASCII histogram". That's your clue > as to what I'm looking for.) >o > -- > David J Dachtera > dba DJE Systems  > http://www.djesys.com/ >t+ > Unofficial OpenVMS Hobbyist Support Page:s$ > http://www.djesys.com/vms/support/ >t* > Unofficial Affordable OpenVMS Home Page:! > http://www.djesys.com/vms/soho/a >t$ > Unofficial OpenVMS-IA32 Home Page:! > http://www.djesys.com/vms/ia32/t >  > Coming soon:( > Unofficial OpenVMS Marketing Home Page   ------------------------------   Date: 7 Mar 2005 19:07:37 -0600i+ From: young_r@encompasserve.org (Rob Young)t Subject: Intel and high-k 3 Message-ID: <fpiYWIlhVpiQ@eisner.encompasserve.org>   $ 	Intel high-k article, January 2004:  9 http://www.intel.com/update/contents/it01041.htm#moreinfo   N The current challenge is to continue to push the 45 nm node, and the effort isH evolving from pure research to development work with Intel manufacturingK engineers. Intel is now on-track to implement the high-K dielectric and new H metal gate materials as part of our 45 nm manufacturing process in 2007.   Summaryn  O Moore's Law, which states that the number of transistors on integrated circuitseI doubles every processor generation (approximately every 24 months) is thetF foundation for the exponential growth of computing power and componentI integration at reduced cost. Current leakage and power dissipation issues K associated with the traditional SiO2 CMOS transistor gate dielectric impose 5 practical limits on the extensibility of Moore's Law.r  I After several years of effort, the Intel technology team has made a major K breakthrough that solves the chip power problem. Intel has identified a newoN material known as "high-K" to replace SiO2 as the gate dielectric, in additionH to new metals to replace the polysilicon gate electrode of NMOS and PMOS transistors.  M These new materials, when implemented with the correct process recipe, reducesN gate leakage by over 100-fold, while delivering record transistor performance.M Intel is on track to implement this new technology in the 45 nm node in 2007.n> The breakthrough will drive Moore's Law into the next decade.    ---g  B 	One chap commented that the power reduction is astounding.  Seems? 	with leakage reduced 100-fold, that isn't too idle of a boast.r   				Rob-   ------------------------------   Date: 7 Mar 2005 18:42:30 -0600e+ From: young_r@encompasserve.org (Rob Young) / Subject: Re: new Itanium after Tukwila: Poulson-3 Message-ID: <9l5KYwW$QG1$@eisner.encompasserve.org>m  K In article <87sm38g4w5.fsf@prep.synonet.com>, prep@prep.synonet.com writes:    > F > No one has production capable steppers as yet, plus the small detail& > of leakage and working high K gates. > G > Intel has a second problem in that IBM holds lots of the key patents.  >   : 	You may be confusing high-k with low-k.  IBM not too long 	ago made this bold statement:    - http://www.eetimes.com/story/OEG20020610S0076    By David Lammers   EE Times June 11, 2002 (6:32 a.m. ET)    M Davari said the high-k insulator challenge is so thorny that it may even come I later than the introduction of a vertical, dual-gate transistor structureRO another means of improving current drive. The vertical dual-gate structures are O difficult to fabricate, however, and have been seen as a radical departure from , today's horizontally planar CMOS structures.  (         and this from that same article:  K Strained silicon must be brought to manufacturing first, because the high-kEL insulators studied thus far create a drag on electron mobility. A shift to aI high-k insulator without the prior introduction of strained silicon would;J result in a device with less performance than transistors that use silicon dioxide as the gate insulator.  K For that reason, IBM plans to introduce strained silicon at the 65-nm node, B though Davari voiced a small possibility that it would come later.  L         Note that Intel is saying they solved the major high-k problem, with 	a breakthrough:  	1 http://www.eetimes.com/semi/news/OEG20031215S0067   , Companies put high-k on fast track for 45 nm   By David Lammers   EE Times" December 15, 2003 (2:27 p.m. ET)      L Last year, Max Fischetti, a staff member at IBM Corp.'s T.J. Watson ResearchN Center (Yorktown Heights, N.Y.), reported that the high-k metallic oxides moreL easily create soft optical phonons that couple with electrons in the channelO region, reducing electron mobility significantly even as leakage at the gate is 
 minimized.  H At IEDM, Shimal Datta, representing a team working at Intel's componentsH research lab in Hillsboro, Ore., said that while Intel has confirmed theN IBM-reported mobility degradation, it was able to reduce the phonon scatteringJ by using a titanium nitride metal gate above the hafnium oxide. The use ofO strained silicon, with strain introduced by a thick capping layer and a relaxed2J silicon germanium buffer layer comprising only 10 percent germanium atoms,1 further helped overcome the mobility degradation.5  H "Our electron mobility was enhanced by the metal gate. Combined with theM strained silicon, we are back above the universal mobility curve of silicon,"a Datta said.    ---   0 	The above showing up here in c.o.v. - Jan 2004:  Q http://groups-beta.google.com/group/comp.os.vms/msg/17f906cbb4b5fa21?dmode=sources  D 	And at the time of the announcement , it created quite the industry> 	buzz as IBM publiclally states:  "not yet - hard problem" and6 	Intel states:  "We got it."   From the above article:  1 http://www.eetimes.com/semi/news/OEG20031215S0067o  L "Several researchers at the 2003 International Electron Devices Meeting hereF last week said their companies' schedules for high-k introduction wereE accelerated after Intel Corp. announced in early November that it hadIM identified a high-k oxide and two metals for the gate electrodes, with likely   introduction at the 45-nm node."   				Rob2   ------------------------------   Date: 7 Mar 2005 20:17:31 -08002) From: "aniranshu" <aniranshu@yahoo.co.in>o Subject: Re: Oracle RDBnB Message-ID: <1110255451.087801.51900@o13g2000cwo.googlegroups.com>  	 Hi Keith,m   Those were real helpful.   Thanks mate :-) !!   ------------------------------   Date: 7 Mar 2005 12:46:03 -0800M$ From: "Ed Wilts" <ewilts@ewilts.org>N Subject: Re: PCSI product install failure - DDIS-E-TNF, invalid element syntaxC Message-ID: <1110228363.598473.243090@l41g2000cwc.googlegroups.com>s   Hi Charlie,.  B First of all, thanks for the comments on this thread.  Between theD information I was getting on the original call I logged and what has@ been discussed here, I might actually be able to recover from myD problems.  I have not only identified that the footprint matches butE I've identified the specific block that was corrupted and can go backa@ in time to before the corruption (the block has a nice timestampB telling me when it was corrupted).  I'm pulling the tapes from ourE offsite storage and will be restoring my PCSI database before my lastwG set of patches (which is when SYS V6.0 got installed and nailed me) andeF then re-applying the patches from there (saving the recovery data thisE time around).  Overall, I think may have gotten off lucky and believe E that simply restoring the database and re-applying a few patches will? get me currrent again.  D I think that there should be tools to fix a corrupt PCSI database inF the future or the "database" design should be such that it can recoverB itself somewhat more gracefully than requiring a full VMS install.E After all, vmsinstal never had these issues (it had others, but never 
 this bad).  D To everybody else:  it might not be a bad idea to make sure you have1 multiple good backups of your PCSI database filese4 ([vms$common...]*.*pcsi*) that you can restore from.  
 Thanks again,t	    .../EdD   ------------------------------  # Date: Mon, 07 Mar 2005 21:58:55 GMTu3 From: hammond@not@peek.ssr.hp.com (Charlie Hammond)iN Subject: Re: PCSI product install failure - DDIS-E-TNF, invalid element syntax1 Message-ID: <z84Xd.1187$yG1.156@news.cpqcorp.net>r  D In article <1110228363.598473.243090@l41g2000cwc.googlegroups.com>, & "Ed Wilts" <ewilts@ewilts.org> writes:  8 >First of all, thanks for the comments on this thread.    2 You are welcome.  I'm glad they have been helpful.  F >I think that there should be tools to fix a corrupt PCSI database ...  B If we were prescient and knew what was going to go wrong, we could@ do this.  Unfortunately, solving the general case of an unknown/F unexpected error is a bit difficult.  Consider the case at hand, where) PCSI itself did not cause the corruption.m  = BTW, if you have the PCSI kit for a product you can do a much @ better PRODUCT REGISTER -- one that will recapture almost all of< the information in the database.  To do this, do NOT use the; SYS$UPDATE:PCSI$REGISTER_PRODUCT.COM procedure.  Instead don+ a PRODUCT REGISTER PRODUCT similar to this:o  <     $ PRODUCT REGISTER PRODUCT <name> /SOURCE=<kit_location>  D For example, if you want to register the products that are installedA along with OpenVMS, you could mount the kit, lets say on DKA400:,  and then  8     $ PRODUCT REGISTER PRODUCT * /SOURCE:dka400:[KITS.*]  @ This will present a list of products and offer you the choice to% choose the ones you want to register.r  K For other products, you would mount a different CD and point to a differentc4 directory on it -- or maybe you keep kits on-line???  G >the future or the "database" design should be such that it can recoverdC >itself somewhat more gracefully than requiring a full VMS install.   E The full information needed to recover from an unpredictable problem  E would require that ALL the kits for installed products are available. " In the general case, that is hard.  E >To everybody else:  it might not be a bad idea to make sure you havee2 >multiple good backups of your PCSI database files5 >([vms$common...]*.*pcsi*) that you can restore from.d  ) Regular backups are *ALWAYS* a good idea.a   -- hJ       Charlie Hammond -- Hewlett-Packard Company -- Ft Lauderdale  FL  USAF           (hammond@not@peek.ssr.hp.com -- remove "@not" when replying)J       All opinions expressed are my own and not necessarily my employer's.   ------------------------------   Date: 7 Mar 2005 14:14:03 -0500i7 From: "Gareth V. Williams" <graff@cfa0.cfa.harvard.edu>I% Subject: Sending SMTP e-mail from DCLl. Message-ID: <422ca7fb@cfanews.cfa.harvard.edu>  E I recently moved my home VMS box behind our firewall router, to avoid@B problems with my SSH sessions into the office VMS machines hangingD I can forward X window and SSH packets from the router to my box andC I have also set up SMTP packets to be forwarded.  However, I cannotc? send e-mail to my home machine.  Mail sent to "gareth@<firewalln: router IP address>" bounces back with the following error:H %TCPIP-E-SMTP_NOSUCHUSER, no such user, GARETH@<firewall router IP addr>E If I make a manual connection to port 25 at <firewall router IP addr> @ I can send e-mail if I set the "RCPT TO: GARETH", but that is no help with the DCL MAIL command.h  C I have looked at SFF, but that simply calls the MAIL facility to docD the work of transfrring the data and so I cannot use it to send mail. to "GARETH" and have it arrive at my home box.  C So I've been looking at the DCL solution to direct communication toE> port 25 that has been discussed on this newsgroup in the past.C I came up with a short test script to check that it worked.  In thee@ example command file below WORK_EMAIL is my work e-mail address,D WORK_DOMAIN is my work domain and ROUTER_IP is the IP address of theC firewall router: all are defined in the portion of the command files not shown here.t  E $! Attempt to send SMTP e-mail from DCL on work system to home systemE7 $ telnet/create_session/protocol=none 'ROUTER_IP' 25 99p! $ open/read/write smtplink TNA99:.& $ write smtplink "HELO ''WORK_DOMAIN'"+ $ write smtplink "MAIL FROM: ''WORK_EMAIL'"r" $ write smtplink "RCPT TO: gareth" $ write smtplink "DATA"r: $ write smtplink "This is a test message.  Please ignore." $ write smtplink "." $ write smtplink "QUIT"p $ close smtplink $ telnet/delete_session 99  ' Upon executing the command file, I get:f  ) %TELNET-S-CRSES, Session created on TNA99n   and nothing else.t  C On my home box (where I have REPLY/ENABLE=(NETWORK,SECURITY) set) Io8 see the following as soon as I execute the command file:  8 %%%%%%%%%%%  OPCOM   7-MAR-2005 12:40:13.91  %%%%%%%%%%%$ Message from user INTERnet on SCULLYF INTERnet ACP SMTP Accept Request from Host: 131.142.11.101 Port: 50257   but no mail is delivered.s  A The previously-posted DCL code appears to have worked for others.f5 Why isn't mine?  Any hints as to what I'm doing wrongi would be appreciated.n  9 Both office and home boxes are VMS V7.3-2, TCPIP V5.4-15.i     Gareth   -- dH ------------------------------------------------------------------------H Gareth V. Williams, MS 18, 60 Garden Street, Cambridge, MA 02138, U.S.A.+ Associate Director, IAU Minor Planet CenterUH gwilliams@cfa.harvard.edu        http://cfa-www.harvard.edu/iau/mpc.html7 OpenVMS & RISC OS: refined choices in operating systems    ------------------------------   Date: 07 Mar 2005 19:57:16 GMT/ From: Thierry Dussuet <thierry@dussuet.lugs.ch>a) Subject: Re: Sending SMTP e-mail from DCLo/ Message-ID: <slrnd2pcgt.tt.thierry@MARS.Family>   E On 2005-03-07, Gareth V. Williams <graff@cfa0.cfa.harvard.edu> wrote:iG > I recently moved my home VMS box behind our firewall router, to avoid D > problems with my SSH sessions into the office VMS machines hangingF > I can forward X window and SSH packets from the router to my box andE > I have also set up SMTP packets to be forwarded.  However, I cannot A > send e-mail to my home machine.  Mail sent to "gareth@<firewallu< > router IP address>" bounces back with the following error:J > %TCPIP-E-SMTP_NOSUCHUSER, no such user, GARETH@<firewall router IP addr>G > If I make a manual connection to port 25 at <firewall router IP addr>.B > I can send e-mail if I set the "RCPT TO: GARETH", but that is no! > help with the DCL MAIL command.   L Did you set the mail server to recognize the firewall router ip such that itP accepts mail for it?  Or/And maybe you'd like to set up some dns/dyndns to pointL to the ip, and set the mail server to have it as its domain name.  I have itD here like this and it works (worked with TCPIP, works with Multinet)M Also, maybe you have to make sure that the IP is enclosed in square brackets,o like GARETH@[123.123.123.123]V   Thierry    ------------------------------  * Date: Mon, 7 Mar 2005 19:46:29 +0000 (UTC). From: klewis@LUMINA.MITRE.ORG (Keith A. Lewis)) Subject: Re: Sending SMTP e-mail from DCL . Message-ID: <d0ib2l$r1f$1@newslocal.mitre.org>   "Gareth V. Williams" <graff@cfa0.cfa.harvard.edu> writes in article <422ca7fb@cfanews.cfa.harvard.edu> dated 7 Mar 2005 14:14:03 -0500:n8 >$ telnet/create_session/protocol=none 'ROUTER_IP' 25 99  I I didn't know about telnet/create_session.  I will have to read about it!i  " >$ open/read/write smtplink TNA99:' >$ write smtplink "HELO ''WORK_DOMAIN'"c, >$ write smtplink "MAIL FROM: ''WORK_EMAIL'"# >$ write smtplink "RCPT TO: gareth"n >$ write smtplink "DATA"; >$ write smtplink "This is a test message.  Please ignore."  >$ write smtplink "."o >$ write smtplink "QUIT" >$ close smtplink. >$ telnet/delete_session 99e    9 >%%%%%%%%%%%  OPCOM   7-MAR-2005 12:40:13.91  %%%%%%%%%%%M% >Message from user INTERnet on SCULLYtG >INTERnet ACP SMTP Accept Request from Host: 131.142.11.101 Port: 50257. >  >but no mail is delivered.  L One thing you're doing "wrong" -- after "DATA" there should be a full rfc822K message.  Headers, blankline, body, then ".".  The headers section needs tocG contain some minimum set -- I think it's {to, from, subject}.  Date and.@ Message-ID are optional (I think) but they are almost universal.  D I'm not sure that this is what's causing your problem but you should  probably fix it as a first step.  0 --Keith Lewis              klewis {at} mitre.org> The above may not (yet) represent the opinions of my employer.   ------------------------------    Date: 07 Mar 2005 16:35:40 -05003 From: Rich Alderson <news@alderson.users.panix.com>n) Subject: Re: Sending SMTP e-mail from DCL . Message-ID: <mddy8czqgur.fsf@panix5.panix.com>  0 klewis@LUMINA.MITRE.ORG (Keith A. Lewis) writes:  N > One thing you're doing "wrong" -- after "DATA" there should be a full rfc822M > message.  Headers, blankline, body, then ".".  The headers section needs touI > contain some minimum set -- I think it's {to, from, subject}.  Date andgB > Message-ID are optional (I think) but they are almost universal.  H The only thing required by RFC821 (SMTP) is text data following the DATAM command; I have frequently tested my email system by putting random data intoiO the message.  The MUA may be unhappy with the result, but the MTA doesn't care.r  O Only braindead mail programs pay any attention to what is in the data--all thatg5 is required for delivery is in the protocol commands.b  K (And I've been responsible for RFC821/RFC822-compliant e-mail systems since-K  1983, with considerable experience in implementation; I'm not just talkingm  through my hat.)    --  L Rich Alderson                                       | /"\ ASCII ribbon     |L news@alderson.users.panix.com                       | \ / campaign against |L "You get what anybody gets. You get a lifetime."    |  x  HTML mail and    |L                          --Death, of the Endless    | / \ postings         |   ------------------------------  # Date: Mon, 07 Mar 2005 22:15:49 GMTr# From: hoff@hp.nospam (Hoff Hoffman)u) Subject: Re: Sending SMTP e-mail from DCLt1 Message-ID: <po4Xd.1191$2J1.805@news.cpqcorp.net>f  h In article <422ca7fb@cfanews.cfa.harvard.edu>, "Gareth V. Williams" <graff@cfa0.cfa.harvard.edu> writes:F :I recently moved my home VMS box behind our firewall router, to avoidC :problems with my SSH sessions into the office VMS machines hanging E :I can forward X window and SSH packets from the router to my box andoD :I have also set up SMTP packets to be forwarded.  However, I cannot@ :send e-mail to my home machine.  Mail sent to "gareth@<firewall; :router IP address>" bounces back with the following error:yI :%TCPIP-E-SMTP_NOSUCHUSER, no such user, GARETH@<firewall router IP addr>/  C   Your firewall is not passing through mail to the interior of your E   network -- most won't, of course, unless specifically configured tof    allow it, for obvious reasons.  E   Depending on the particulars of the firewall, you will need to have E   an IP address visible behind the firewall, or you will have to haveeC   the firewall configured to forward the incoming SMTP traffic to aiF   protected SMTP server located behind the firewall.  (If the firewallD   is forwarding traffic for you, you will want to be very careful toD   ensure that you are not configuring your network to relay spam, asE   there are folks that can and do probe all networks looking for open F   SMTP relays, and that subsequently do use that information to launchF   spam -- and there are other folks that black-list networks with openD   SMTP relays, cutting the network off from various communications.)  D   Various of the more expensive firewalls will have SMTP servers andB   filtering, virus-scanning and other such features built in.  (ItE   would not surprise me to learn there is a cut-down "embedded" LinuxuD   firewall around that can provide at least some of this, when usingA   an old PC as a dedicated home-brewed Linux-based firewall.  TheKE   commercial products with SMTP servers, and spam and virus filteringD-   features, tend to be higher-end offerings.)c  B   If your firewall is correctly configured to protect your networkB   and you have not opened up the SMTP path, you should not be ableD   to circumvent the firewall protections with DCL-based or SFF mail.    N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faqaN  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.come   ------------------------------  % Date: Mon, 07 Mar 2005 17:16:42 -0500r- From: JF Mezei <jfmezei.spamnot@teksavvy.com>>) Subject: Re: Sending SMTP e-mail from DCLrB Message-ID: <1110233049.a10540dcf531f57da8ce00752255d50c@teranews>   "Gareth V. Williams" wrote:mA > send e-mail to my home machine.  Mail sent to "gareth@<firewallx< > router IP address>" bounces back with the following error:J > %TCPIP-E-SMTP_NOSUCHUSER, no such user, GARETH@<firewall router IP addr>G > If I make a manual connection to port 25 at <firewall router IP addr>2B > I can send e-mail if I set the "RCPT TO: GARETH", but that is no! > help with the DCL MAIL command.k  @ You may wish to set the logging logicals for the SMTP server AND
 receiver. ; http://www.vaxination.ca/vms/index.html and choose the smtpe configuration section.    E When you send to user@ip.ad.dr.ess   , you need to make sure that the . VMS host actually knows about that ip address.  H For instance, if you internet address on the wan side is 69.45.23.45 butG your VMS host, behind a NAT router knows itself as 10.0.0.34, then whenyG it gets a message destined for gareth@69.45.23.45 it woudln't know thati it belonged to itself.    E > I have looked at SFF, but that simply calls the MAIL facility to dorF > the work of transfrring the data and so I cannot use it to send mail0 > to "GARETH" and have it arrive at my home box.  B SFF does not call VMSmail. It submits a job to the SMTP symbiont.     $ > $ write smtplink "RCPT TO: gareth"  D This tells the receiving VMS system to simply add whatever the local? domain name is is deliver the message to the local user gareth.u  ? But real emails have a RCPT TO: that contains a fully qualifiednG user/domain and the receiving smtp server will try to resolve the domaneI portion to see if the email needs to be forwarded to another host or not.g   ------------------------------   Date: 7 Mar 2005 17:44:38 -0500e7 From: "Gareth V. Williams" <graff@cfa0.cfa.harvard.edu>n) Subject: Re: Sending SMTP e-mail from DCLs. Message-ID: <422cd956@cfanews.cfa.harvard.edu>  / Keith A. Lewis <klewis@lumina.mitre.org> wrote:o : "Gareth V. Williams" <graff@cfa0.cfa.harvard.edu> writes in article <422ca7fb@cfanews.cfa.harvard.edu> dated 7 Mar 2005 14:14:03 -0500:a9 :>$ telnet/create_session/protocol=none 'ROUTER_IP' 25 99i  K : I didn't know about telnet/create_session.  I will have to read about it!e  A   Neither did I until I went digging in the archive...  Ain't VMSO
 wonderful :-)h  # :>$ open/read/write smtplink TNA99:y( :>$ write smtplink "HELO ''WORK_DOMAIN'"- :>$ write smtplink "MAIL FROM: ''WORK_EMAIL'"a$ :>$ write smtplink "RCPT TO: gareth" :>$ write smtplink "DATA"s< :>$ write smtplink "This is a test message.  Please ignore." :>$ write smtplink "." :>$ write smtplink "QUIT"  :>$ close smtplink :>$ telnet/delete_session 99    : :>%%%%%%%%%%%  OPCOM   7-MAR-2005 12:40:13.91  %%%%%%%%%%%& :>Message from user INTERnet on SCULLYH :>INTERnet ACP SMTP Accept Request from Host: 131.142.11.101 Port: 50257 :> :>but no mail is delivered.h  N : One thing you're doing "wrong" -- after "DATA" there should be a full rfc822M : message.  Headers, blankline, body, then ".".  The headers section needs touI : contain some minimum set -- I think it's {to, from, subject}.  Date and>B : Message-ID are optional (I think) but they are almost universal.  G   I've not found that to be the case when I've made a manual connection I to the SMTP port on my home machine.  Once I've done the "MAIL FROM:" and E the "RCPT TO:", I just do "DATA" and type a test message, without anyi3 RFC822 headers and the test message gets delivered.h  F : I'm not sure that this is what's causing your problem but you should" : probably fix it as a first step.  E   I modified the command file.  Still didn't work :-(  But thanks forn the thought.  	    Garethd     -- eH ------------------------------------------------------------------------H Gareth V. Williams, MS 18, 60 Garden Street, Cambridge, MA 02138, U.S.A.+ Associate Director, IAU Minor Planet Center.H gwilliams@cfa.harvard.edu        http://cfa-www.harvard.edu/iau/mpc.html7 OpenVMS & RISC OS: refined choices in operating systemsn   ------------------------------   Date: 7 Mar 2005 18:07:54 -0500t7 From: "Gareth V. Williams" <graff@cfa0.cfa.harvard.edu>x) Subject: Re: Sending SMTP e-mail from DCLn. Message-ID: <422cdeca@cfanews.cfa.harvard.edu>   Hoff Hoffman <hoff> wrote:E :   Your firewall is not passing through mail to the interior of youroG :   network -- most won't, of course, unless specifically configured tot" :   allow it, for obvious reasons.  ?   I have configured the firewall to pass port 25 traffic to thee; VMS box.  I know this forwarding works: I can make a manualtF telnet/port=25 connection to the IP address of the firewall router and? my home VMS box responds.  My problem is that I can't do by thes( DCL route that seems to work for others.  H :   protected SMTP server located behind the firewall.  (If the firewallF :   is forwarding traffic for you, you will want to be very careful toF :   ensure that you are not configuring your network to relay spam, as  D   The firewall is configured to forward SMTP to my VMS box, which is configured to not relay mail.   
     Gareth   --  H ------------------------------------------------------------------------H Gareth V. Williams, MS 18, 60 Garden Street, Cambridge, MA 02138, U.S.A.+ Associate Director, IAU Minor Planet Center H gwilliams@cfa.harvard.edu        http://cfa-www.harvard.edu/iau/mpc.html7 OpenVMS & RISC OS: refined choices in operating systemso   ------------------------------  # Date: Mon, 07 Mar 2005 23:52:36 GMTl# From: hoff@hp.nospam (Hoff Hoffman)e) Subject: Re: Sending SMTP e-mail from DCLh1 Message-ID: <8P5Xd.1211$HO1.203@news.cpqcorp.net>a  h In article <422cdeca@cfanews.cfa.harvard.edu>, "Gareth V. Williams" <graff@cfa0.cfa.harvard.edu> writes:@ :  I have configured the firewall to pass port 25 traffic to the< :VMS box.  I know this forwarding works: I can make a manualG :telnet/port=25 connection to the IP address of the firewall router andn@ :my home VMS box responds.  My problem is that I can't do by the) :DCL route that seems to work for others.d  K   Are you getting messed up by the record formatting involved?  DCL prefers2J   to use formats other than stream lf, obviously (VFC, most commonly), andH   stream lf is quite often the prefered choice for record processing and.   for record formats within many IP utilities.    N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com.   ------------------------------  $ Date: Mon, 7 Mar 2005 21:25:13 +0100, From: "Dr. Dweeb" <5msg0h202@sneakemail.com>! Subject: Re: Shannon opines on HPl- Message-ID: <d0idb8$12ct$1@news.cybercity.dk>t   Bill Todd wrote: > JF Mezei wrote:c >o > ...n > F >> In both cases, Pfeiffer and Carly had their eyes set on growing theE >> company to compete against IBM. In both cases, they failed royallysG >> since they focused on the wintel crap instead of leveraging the real> >> enterprise stuff. >t > Wrong. >oA > Pfeiffer focused significantly, possibly even primarily, on thenH > enterprise stuff.  It may well have been a large part of the reason heA > got kicked out (because Ben Rosen favored a far more PC-centrich > strategy). >   I Yes, I thought there was a chance with Pfeiffer because it seemed like heDL understood that PCs were just something you needed when dealing with the bigK customers, but that the enterprise stuff and services were where the future!J big bucks would be.  Sadly, Rosen bumped him before we had a chance to see$ what sort of go he could make of it.  	 Dr. Dweeb   = > Or perhaps you meant to say 'Curly' rather than 'Pfeiffer'._ >A > - bill   ------------------------------   Date: 7 Mar 2005 12:21:59 -0800g From: tadamsmar@yahoo.comA& Subject: Re: Shavlik Randolph Fan ClubC Message-ID: <1110226919.298821.265460@f14g2000cwb.googlegroups.com>T  $ Sorry!  I posted to the wrong group!   ------------------------------   Date: 7 Mar 2005 12:20:29 -0800E From: tadamsmar@yahoo.comn" Subject: Shavlik Randolph Fan ClubC Message-ID: <1110226829.933744.178340@z14g2000cwz.googlegroups.com>o  0 I think we should start a fan club for Shav. The4 guy managed to put up impressive numbers last night.  2 He made %25 of Duke's fouls and while playing only 15 minutes.q  5 But that 15 minutes still makes him Duke's Sixth Man. / Solid. The next closest is Love with 6 minutes.a  1 Must be some intangible that awards him the loftyA- post of Duke's Six man.  I'm not sure what ita0 is.  I think is a rare combination of impressive- high school stats, 6'10'' and 98.6 degrees F.-  . Also, that hair.  I wonder if people think all. the NC flash-in-the-pans have Clay Aiken hair?   You can't coach that hair.   You can't coach 98.6.t   ------------------------------  * Date: Mon, 7 Mar 2005 22:01:39 +0000 (UTC) From: tadamsmar@yahoo.comt" Subject: Shavlik Randolph Fan Club$ Message-ID: <d0ij03$eeo$1@online.de>   illegal message cancelled    ------------------------------  $ Date: Mon, 7 Mar 2005 16:00:36 -05002 From: "Jonathan Boswell" <jsb.NOSP@M.cdrh.fda.gov>5 Subject: Re: Sony vs HP: a comparison of CEO oustings35 Message-ID: <Wh3Xd.2766$Ny6.4901@mencken.net.nih.gov>o  : "JF Mezei" <jfmezei.spamnot@teksavvy.com> wrote in message< news:1110181564.d5e691710c4c89601f1a387e3b21f247@teranews...I > It worries me that HP may in fact end up like Compaq, unable to attract C > a new LEADER and forced to appoint some accountant or MBA that ise* > without vision, leadership and backbone.   Maybe Ken Olsen is available.t   ------------------------------  $ Date: Mon, 7 Mar 2005 17:03:48 -0500# From: "John Smith" <a@nonymous.com> 5 Subject: Re: Sony vs HP: a comparison of CEO oustingsa, Message-ID: <ob2dnUExIKBaUrHfRVn-sA@igs.net>   Jonathan Boswell wrote:m< > "JF Mezei" <jfmezei.spamnot@teksavvy.com> wrote in message> > news:1110181564.d5e691710c4c89601f1a387e3b21f247@teranews...B >> It worries me that HP may in fact end up like Compaq, unable toD >> attract a new LEADER and forced to appoint some accountant or MBA3 >> that is without vision, leadership and backbone.I >v > Maybe Ken Olsen is available.      I always liked Jim Milton.   --F OpenVMS    - The often imitated but never advertised operating system.D Alpha - I could have had an EV8 but all I got was this lousy Itanic.   ------------------------------  $ Date: Mon, 7 Mar 2005 18:19:22 -0500' From: "Main, Kerry" <kerry.main@hp.com>u5 Subject: RE: Sony vs HP: a comparison of CEO oustingsdR Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB594D47@tayexc19.americas.cpqcorp.net>   > -----Original Message------ > From: John Smith [mailto:a@nonymous.com]=203 > Sent: March 7, 2005 5:04 PMi > To: Info-VAX@Mvb.Saic.Com27 > Subject: Re: Sony vs HP: a comparison of CEO oustingsl >=20 > Jonathan Boswell wrote:e> > > "JF Mezei" <jfmezei.spamnot@teksavvy.com> wrote in message@ > > news:1110181564.d5e691710c4c89601f1a387e3b21f247@teranews...D > >> It worries me that HP may in fact end up like Compaq, unable toF > >> attract a new LEADER and forced to appoint some accountant or MBA5 > >> that is without vision, leadership and backbone.e > >s! > > Maybe Ken Olsen is available.  >=20 >=20 > I always liked Jim Milton. >=20  F Me to - fellow Canadian who rose from Digital field Sales rep to HP NA Sales VP (before he left).   :-)b  
 Kerry Main Senior Consultanth HP Services Canada Voice: 613-592-4660t Fax: 613-591-4477l kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  $ "OpenVMS has always had integrity .. Now, Integrity has OpenVMS .."   ------------------------------  $ Date: Mon, 7 Mar 2005 21:17:25 -0500# From: "John Smith" <a@nonymous.com> 5 Subject: Re: Sony vs HP: a comparison of CEO oustingsh, Message-ID: <pomdnckTdZKklrDfRVn-gA@igs.net>   Main, Kerry wrote: >> -----Original Message-----v+ >> From: John Smith [mailto:a@nonymous.com]e >> Sent: March 7, 2005 5:04 PM >> To: Info-VAX@Mvb.Saic.Com8 >> Subject: Re: Sony vs HP: a comparison of CEO oustings >> >> Jonathan Boswell wrote:> >>> "JF Mezei" <jfmezei.spamnot@teksavvy.com> wrote in message@ >>> news:1110181564.d5e691710c4c89601f1a387e3b21f247@teranews...D >>>> It worries me that HP may in fact end up like Compaq, unable toF >>>> attract a new LEADER and forced to appoint some accountant or MBA5 >>>> that is without vision, leadership and backbone.t >>>o! >>> Maybe Ken Olsen is available.s >> >> >> I always liked Jim Milton.a >> >sH > Me to - fellow Canadian who rose from Digital field Sales rep to HP NA > Sales VP (before he left).    1 You meant to say 'was shown the door', don't you?a    6 HP - where 'Integrity' is just another decal on a box.   ------------------------------  # Date: Mon, 07 Mar 2005 19:27:37 GMTt! From: Nigel Barker <nigel@hp.com>t; Subject: Re: Starting JAVA after Upgrading to OpenVMS 7.3.2 8 Message-ID: <0r8p21dl995rkcnsb09t1nei7vgb1mfjfo@4ax.com>  : On 7 Mar 2005 10:22:20 -0800, gary.lively@gmail.com wrote:   >eD >After upgrading to OpenVMS 7.3.2, the current version of Java 1.1.8 >stopped working.  > F >I upgraded to version JAVA131 V1.3-15, but after a reboot it will not >start.  > 8 >My question is - how do I restart Java after a reboot ? >s >thanks   P Either put @SYS$COMMON:[JAVA$131.COM]JAVA$131_SETUP.COM FAST in the LOGIN.COM of( each Java user it put it in SYLOGIN.COM.   -- Nigel Barker Live from the sunny Cote d'Azur2   ------------------------------   Date: 7 Mar 2005 14:02:57 -0800i* From: thierry.uso@wanadoo.fr (Thierry USO)V Subject: Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers= Message-ID: <79c022e7.0503071402.775d0800@posting.google.com>l  a Roy Omond <Roy.Omond@BlueBubble.UK.Com> wrote in message news:<393bpoF5u1t9oU1@individual.net>...  > Gentle colleagues, > 1 > this is becoming a bit of a showstopper for me:e > > > A very simple pair of programs (Server and Client) are setupB > to transfer a quantity of data via a TCP Socket.  The backgroundF > to this is actually in trying to get this to work in UCX v3.3 ECO 3,@ > whereas it worked flawlessly (and as per the documentation) in< > vanilla UCX v3.3 (without ECO), running under VAX VMS 6.2. > > > The problem is easily demonstrated on VMS 7.3-2 (Alpha) with? > TCP/IP Services v5.4 ECO 4, as well as all the versions afters0 > UCX v3.3 ECO 3 and above (both VAX and Alpha). > H > The Server process establishes a socket, listens for an accept on the H > socket (which the Client sends), and sends an amount of data down the G > socket to the client.  That's it.  It works fine with data up to and  < > including 65,535 bytes.  Beyond this, it returns an error 0 > ("%system-f-ivbuflen, invalid buffer length").@ > I can find no mention in the documentation "TCP/IP Sockets APIA > and System Services Programming" of any such restriction on thea > amount of data to transfer.t >   C During the TCP connection establishment, the MSS (Max Segment Size) E may be negotiated using the TCP MSS option. 65535 is the max value of D the TCP MSS option (16-bit field)... A segment size > 65535 is a bad? choice (IP fragmentation...) and seems to be banned in some UCXa	 versions.    ------------------------------  % Date: Mon, 07 Mar 2005 14:03:02 -0800i# From: "Tom Linden" <tom@kednos.com>iV Subject: Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers( Message-ID: <opsnafjc1mzgicya@hyrrokkin>  G On Mon, 07 Mar 2005 16:14:45 -0500, Dave Froble <davef@tsoft-inc.com>  e wrote:  K > Actually, I'm a bit surprised that you could send more than 32767 bytes  kI > in one I/O, due to string lengths in descriptors using a word for the  cJ > length.  Ok, proves I shouldn't try to think.  That would likely be an  J > unsigned word, and it can indeed go up to 65535.  I'm guessing that is  D > your problem, a string longer than VMS can handle in a single I/O.  3 Can you use an unsigned for length in a descriptor?-   ------------------------------  # Date: Mon, 07 Mar 2005 23:20:53 GMTn# From: hoff@hp.nospam (Hoff Hoffman)-V Subject: Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers1 Message-ID: <pl5Xd.1206$oL1.823@news.cpqcorp.net>-  j In article <79c022e7.0503071402.775d0800@posting.google.com>, thierry.uso@wanadoo.fr (Thierry USO) writes:b :Roy Omond <Roy.Omond@BlueBubble.UK.Com> wrote in message news:<393bpoF5u1t9oU1@individual.net>...I :> The Server process establishes a socket, listens for an accept on the rI :> socket (which the Client sends), and sends an amount of data down the  H :> socket to the client.  That's it.  It works fine with data up to and = :> including 65,535 bytes.  Beyond this, it returns an error  1 :> ("%system-f-ivbuflen, invalid buffer length").S : D :During the TCP connection establishment, the MSS (Max Segment Size)F :may be negotiated using the TCP MSS option. 65535 is the max value ofE :the TCP MSS option (16-bit field)... A segment size > 65535 is a badu@ :choice (IP fragmentation...) and seems to be banned in some UCX
 :versions.  F   Transfers of 65536 and up require TCP/IP Services V5.5, or later (asG   available) -- this work involves changes within both the C RTL and inuF   TCP/IP.  These larger I/O transfers do not work reliably (if at all)F   on earlier TCP/IP Services releases.  For details on the I/O support@   in TCP/IP Services V5.5, please see the release documentation.  N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faqrN  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.comO   ------------------------------  % Date: Mon, 07 Mar 2005 16:14:45 -0500e' From: Dave Froble <davef@tsoft-inc.com> Y Subject: Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers tr-0 Message-ID: <112pghi6o89u456@corp.supernews.com>   Roy Omond wrote: > Dan Allen wrote: > J >> Well - documentation aside - TCP sockets are byte streams. The size of  >> theE >> actual TCP packets are not necessarily related to the application  
 >> read/writeiG >> sizes. Just read and write in chunk sizes of your choosing. The TCP .
 >> layer willt >> packetize it at whim anyway.> >  > : > Exactly, and that is what appears to be *not* happening. > C > I try a write of 65536 bytes, and it generates the error message.  >  > I.e. it's broken.t >   H It's been a while since I was doing this, and I used the System Service " interface, not the CRTL interface.  G Took a quick look.  It seems that I had to do writes to the socket one rG buffer at a time, breaking the entire message up based upon buffersize.p  H Actually, I'm a bit surprised that you could send more than 32767 bytes F in one I/O, due to string lengths in descriptors using a word for the G length.  Ok, proves I shouldn't try to think.  That would likely be an gG unsigned word, and it can indeed go up to 65535.  I'm guessing that is oB your problem, a string longer than VMS can handle in a single I/O.  G Even that is an assumption.  Are you attempting to do a single I/O, or 8" breaking it up into multiple I/Os?   Dave   ------------------------------  % Date: Tue, 08 Mar 2005 00:22:00 -0500n' From: Dave Froble <davef@tsoft-inc.com> Y Subject: Re: TCP/IP - Sockets appear to be restricted to maximum 65,535 byte transfers tro/ Message-ID: <112qd3515bma70@corp.supernews.com>    Tom Linden wrote:yI > On Mon, 07 Mar 2005 16:14:45 -0500, Dave Froble <davef@tsoft-inc.com>  t > wrote: > E >> Actually, I'm a bit surprised that you could send more than 32767  H >> bytes  in one I/O, due to string lengths in descriptors using a word F >> for the  length.  Ok, proves I shouldn't try to think.  That would G >> likely be an  unsigned word, and it can indeed go up to 65535.  I'm  J >> guessing that is  your problem, a string longer than VMS can handle in  >> a single I/O. >  > 5 > Can you use an unsigned for length in a descriptor?8  C I'd always thought (there I go again) that strings were limited to aE 32767, thus the sign bit wasn't used, irregardless of the data being C signed or unsigned.i  F Now, this is discussing descriptors.  However, the real topic is I/O, G and while I'd assume that descriptors would likely be involved, that's xD not a requirement if you specify a buffer starting address, and the B length of data with a LongWord, or an unsigned Word.  These are 2  seperate arguments in a QIO.  H I'd also ask, not that it's your fault either way, what is the use of a 3 negative number in the length word of a descriptor?e  G Casual reading of the above indicates that I haven't had the I/O users  B guide, nor the System Services manual, open in quite a while.  :-(  $ I hate it when I can't have any fun.   Dave   ------------------------------   Date: 7 Mar 2005 14:31:58 -0800'$ From: "AEF" <spamsink2001@yahoo.com>! Subject: Re: VMS startup problem?MC Message-ID: <1110234718.822076.121170@f14g2000cwb.googlegroups.com>a   Hoff Hoffman wrote:yD > In article <1110210707.252629.28730@o13g2000cwo.googlegroups.com>,& "AEF" <spamsink2001@yahoo.com> writes: > : 
 > :AEF wrote:f > :> John Willis wrote:  >V2 > :> > Having a bit of a problem starting VMS 5.1. >    [...]    > :> >>>b/r5:1 > :>C > :> will perform a conversation boot (use lower- or uppercase). Itl will5 > :> give you a SYSBOOT prompt. From there you can dos > :>" > :> SYSBOOT> SET STARTUP_P1 "MIN" > :>C > :> to perform a minumum boot which will skip SYSTARTUP_V5.COM and  probablyE > :> a couple of the other startup command procedures. If that works,- check-   [...]   ? > :> Be sure to set STARTUP_P1 back to "" when you've fixed thee problem! >:9 >   If y'all follow the sequence in the FAQ, you get this. automatically.F >   You won't need the reboot either, as the sequence shown in the FAQA >   can allow you to get a fully running system with the standarde startup G >   -- after making the correction, of course -- without having to makeI< >   the second reboot to get the environment back to normal.  F But if you had set STARTUP_P1 to "MIN", you'll only get a minimum bootC (at least SYSTARTUP_*.COM won't get run). Don't you need the reboot C with STARTUP_P1 reset to "" to get a full, normal boot after fixingrA things during the MIN session? Or is it okay to just manually run G SYS$SYSTEM:STARTUP.COM even though you have a minimal system already up  and running? Or something else?a   ------------------------------  # Date: Mon, 07 Mar 2005 23:31:01 GMT<# From: hoff@hp.nospam (Hoff Hoffman)-! Subject: Re: VMS startup problem?e1 Message-ID: <Vu5Xd.1207$oL1.624@news.cpqcorp.net>-  j In article <1110234718.822076.121170@f14g2000cwb.googlegroups.com>, "AEF" <spamsink2001@yahoo.com> writes:G :But if you had set STARTUP_P1 to "MIN", you'll only get a minimum booteD :(at least SYSTARTUP_*.COM won't get run). Don't you need the rebootD :with STARTUP_P1 reset to "" to get a full, normal boot after fixingB :things during the MIN session? Or is it okay to just manually runH :SYS$SYSTEM:STARTUP.COM even though you have a minimal system already up  :and running? Or something else?  C   Quoting from the relevent section of the OpenVMS Frequently Askedn   Questions (FAQ) document:e       "...G                    4  At the $ prompt, the system will now be acceptingiJ                       startup commands directly from the console. Type the1                       following two DCL commands:a                         $ SPAWNn+                       $ @SYS$SYSTEM:STARTUP,  F                    5  You should now see the dollar ($) prompt of DCL.     ..."    G   Again, please take a look at the entire related discussion within the E   FAQ, as it includes additional details on the various steps, and itsD   has additional background within those areas where there have beenC   discussions of and/or where there has been evidence of confusion.   E   If you should find errors in the FAQ or if its contents or commentsxH   should confuse or trouble you, please do let me know -- and preferablyG   with the background or the correction or such, of course -- as I haver#   the FAQ open for various updates.l  E   I don't mean any of this to appear rude, of course -- the FAQ is aniD   effort to maintain and certainly an effort to read, and there is aG   particular quantity of "fun" in simply keeping its URLs current.  AndrD   by popular demand, those HTML-format URLs are now hyperlinks.  ButG   there is a huge amount of information within the FAQ, and many of theiH   common questions and common discussions are obviously included.  ThereG   are probably too many discussions included, given how big the FAQ hash.   become in recent years.  But I digress.  :-)  N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faqfN  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.comu   ------------------------------  # Date: Mon, 07 Mar 2005 22:42:03 GMTd  From: John Santos <john@egh.com>E Subject: Re: What is the Difference between Shadow and Mirrored disk? + Message-ID: <%M4Xd.84547$uc.25049@trnddc04>-   Tom Linden wrote:aE > On Sat, 05 Mar 2005 02:07:24 GMT, John Santos <john@egh.com> wrote:  >  >> Tom Linden wrote: >>3 >>> On Fri, 04 Mar 2005 19:54:50 -0500, JF Mezei   n) >>> <jfmezei.spamnot@teksavvy.com> wrote:  >>>- >>>> Tom Linden wrote: >>>>E >>>>> can't be a quorum disk.  Suppose you have a shadow set, DSA0,  u >>>>> hysicallyy >>>>> consistingG >>>>> of two drives and connected by, say,shared scsi to nodes A and B.r >>>> >>>> >>>> >>>>    Q >>>> ============================/X/=====================================(scsi)  oF >>>>     |                |                   |                      |J >>>>   [node1]          [disk1]             [disk2]                [node2] >>>>L >>>> If you break the SCSI at X, then both nodes will still see their quorum? >>>> disk and each node will just work with the drive they see.v >>>u7 >>>   The picture I had in mind might be something likea >>>                      LAN& >>>            _______________________' >>>            |                      |w' >>>            |        SCSI          |s) >>>          NODE A ============== NODE Be >>>                       Io >>>                       I  >>>                      DSA0h >>>                  ----------s  >>>                  |         |" >>>                disk1     disk2 >> >>J >> This isn't an acurate picture.  DSA0: is not a hardware device, sittingD >> on a bus, as you've picture here.  It is a software fiction.  TheK >> physical disks are connected to one or more adapters as in JF's picture.: >  > K > Fiction or not, could it not be done? Why does it have to be a hardware  o	 > device? M > In my case, e.g. I have a BA356 with 3 shadow sets on a scsi shared amongstl5 > three nodes, which each of their own internal diskso >  >   E It can't be done because it is impossible.  You can't have a piece ofyB software hanging off a SCSI bus.  You have two choices to make the@ picture reflect what is possible.  If you want to use VMS volume? shadowing, then the picture shows DSA0 in the wrong place.  The E shadowing driver (two copies) lives in the two nodes, not hanging off F the common SCSI bus.  Then JF's picture applies and a severed SCSI busC (or, maybe if the provision's I've mentioned in another post apply, D a simultaneaously severed SCSI bus and LAN), will cause each node toF continue running, thinking it has access to the quorum disk, resulting/ in a partitioned cluster and eventual disaster.o  D The 2nd choice is using a hardware raid controller, and *NOT* volumeE shadowing, in which case the picture is still wrong, since a hardwaredC raid controller *NEVER* would show up as DSA0:.  It would be DRA0:,f; or a bunch of DKAnnn: devices (if it were an HSZx0.)  It is C perfectly okay to use a DRA or DKA raid partition as a quorum disk,dC provided it isn't also shadowed with VMS volume shadowing, in whichf+ case we're back to the 1st situation again.y   ya >>H >> Your picture is a more accurate depiction of a RAID controller, whichB >> would be DRA0:, not DSA0:, and which is fine to use for quorum. >>C >> In JF's picture, there would also have to be a LAN connection oftG >> some sort, since SCSI doesn't do SCS, and you would have to cut bothtB >> the LAN and the SCSI at the same time to partition the cluster. >> >> >> >  >  >      -- u John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------   End of INFO-VAX 2005.133 ************************