1 INFO-VAX	Tue, 26 Jul 2005	Volume 2005 : Issue 414       Contents: Re: another CXX question% Another VMS system bites the dust :-( ) Re: Another VMS system bites the dust :-( ) Re: Another VMS system bites the dust :-( ) Re: Another VMS system bites the dust :-( ) Re: Another VMS system bites the dust :-( & Re: DEC Forms Calender and Time Panels& Re: DEC Forms Calender and Time Panels. improve file system performance related to BEA2 Re: improve file system performance related to BEA. improve file system performance related to BEA2 Re: improve file system performance related to BEA2 Re: improve file system performance related to BEA2 Re: improve file system performance related to BEA2 Re: improve file system performance related to BEA2 Re: improve file system performance related to BEA2 Re: improve file system performance related to BEA2 Re: improve file system performance related to BEA2 Re: improve file system performance related to BEA Re: Modbus TCP for VMS?  Re: Mozilla and ODS5 Re: Mozilla and ODS5 Re: Mozilla and ODS5, Printing while defaulted to remote directory0 Re: Printing while defaulted to remote directory0 Re: Printing while defaulted to remote directory( Re: Response issues on GS1280, VMS 7.3-2( Re: Response issues on GS1280, VMS 7.3-2( Re: Response issues on GS1280, VMS 7.3-2 Re: Storage shelf questions P Re: Two New Itanium 2s Feature Faster Front-Side Bus - Is this new Montecito inf VMS Backup for Windows Re: VMS Backup for Windows Re: VMS Backup for Windows Re: VMS Backup for Windows$ xemacs efs problem since openVMS 7.3( Re: xemacs efs problem since openVMS 7.3= Re: [OT] Rounding v Truncation, was: Re: Platform Support vs. = Re: [OT] Rounding v Truncation, was: Re: Platform Support vs. = Re: [OT] Rounding v Truncation, was: Re: Platform Support vs. = Re: [OT] Rounding v Truncation, was: Re: Platform Support vs.   F ----------------------------------------------------------------------  # Date: Tue, 26 Jul 2005 13:54:36 GMT 4 From: "Ed Vogel" <edward.vogel_stop_the_spam@hp.com>! Subject: Re: another CXX question 2 Message-ID: <wgrFe.9250$T%5.8015@news.cpqcorp.net>  9 "Chip Coldwell" <coldwell@gmail.nospam> wrote in message  ; news:Pine.LNX.4.61.0507251411190.20335@frank.harvard.edu... J > It speaks highly of the group at ZKO that you're willing to investigate 6 > these issues when they're raised by a mere hobbyist.       Thanks Chip.  D     A bug is a bug.  If a hobbyist reports it, a paying customer may+     still encounter it.  We want to fix it.   H     It turns out this problem was a regression introduced in the latest 
 patch kit.)     We plan to update the patch kit soon.   5     We want to thank you for finding this regression.        Take care,       Ed   ------------------------------  % Date: Tue, 26 Jul 2005 09:06:57 +0200 - From: Didier Morandi <prenom.nom@freesurf.fr> . Subject: Another VMS system bites the dust :-(4 Message-ID: <42e5e118$0$14994$626a14ce@news.free.fr>   (from http://tinyurl.com/9nq8w)   / "Sidebar: Miami-Dade Rolling Out New 911 System    News Story by Marc L. Songini   H JULY 25, 2005 (COMPUTERWORLD) - In an effort to get more information to R emergency personnel, Miami-Dade County is rolling out a new $8 million 911 system.  P The system will rely on specialized hardware from Motorola Inc. and integration 3 software from Burlington, Mass.-based Attunity Ltd.   J The system, slated to be implemented next month, will enable personnel to P rapidly access up-to-date information to speed up emergency response, said Jose B Alvarez, an IT director for the Miami-Dade Fire Rescue department.  O The new technology can link the 911 system to other county applications, which  P would allow, for example, the closest available rescue vehicle to be dispatched B to a given location by using an Automatic Vehicle Location system.  N To date, most 911 systems are self-contained for security and reliability and N generally don't interface well with other applications. Keeping the necessary Q emergency information current is very labor intensive, but using interfaces, the  M new system will be able to automatically update personnel, vehicle and radio  # data, as well as operational plans.   H According to analyst Lisa Pierce at Forrester Research Inc., "Real-time O integration with other systems and departments has always been a critical weak   link" of 911 systems.   O Overall, agencies and municipalities are challenged technologically, and there  J is little automation or sophistication in their 911 systems, Alvarez said.  Q "Miami-Dade is probably the exception," he said, noting its plans to enhance the  J system in-house with interfaces to make it more suitable for the county's  specific needs.   N The new package replaces a homegrown legacy system based on the well-traveled ( OpenVMS operating system, Alvarez said."   FYI.   D. --  1 Didier MORANDI - Expert informaticien - VMS / SAP 0   13 chemin du Gu, 1213 Petit-Lancy (GE) Suisse0 Tl. : +33(0)6 7983 6418 ~ www.didiermorandi.com   ------------------------------    Date: 26 Jul 2005 02:49:46 -0700 From: maher_rj@hotmail.com2 Subject: Re: Another VMS system bites the dust :-(A Message-ID: <1122371386.532830.4310@g43g2000cwa.googlegroups.com>   
 Hi Didier,  E Is that the whole article or a summary? I ask only because I couldn't E find where it mentioned exactly what this new wonder technology (that F gave them "Real-time integration with other systems and departments. . .") was.  G Motorola software to go with the hardware? Or is this the same Attunity + Ltd that provides the same software on VMS?   G Hope they have better luck than the new ambulance system that went like  here a couple of years back!   Regards Richard Maher   G PS. I'll bet you a beer that the VMS systems aren't turned off any time E soon. But then I'm sure we're all "tchnologically challenged" when it  comes to defining "replace".   ------------------------------  % Date: Tue, 26 Jul 2005 10:52:44 -0400 2 From: "Stanley F. Quayle" <squayle@insight.rr.com>2 Subject: Re: Another VMS system bites the dust :-(. Message-ID: <42E615FC.20263.340795A@localhost>  - On 26 Jul 2005 at 9:06, Didier Morandi wrote: ! > (from http://tinyurl.com/9nq8w)  > 1 > "Sidebar: Miami-Dade Rolling Out New 911 System  [...] A > The new package replaces a homegrown legacy system based on the 8 > well-traveled OpenVMS operating system, Alvarez said."  D I had proposed CHARON-VAX as a replacement back in September 2004.  A They were already on their way to the replacement system.  *sigh*   
 --Stan Quayle  Quayle Consulting Inc.  
 ----------- Stanley F. Quayle, P.E. N8SQ  +1 614-868-1363 3 8572 North Spring Ct., Pickerington, OH  43147  USA 0 stan-at-stanq-dot-com       http://www.stanq.com) "OpenVMS, when downtime is not an option"    ------------------------------  % Date: Tue, 26 Jul 2005 13:15:36 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 2 Subject: Re: Another VMS system bites the dust :-(+ Message-ID: <42E66FB6.D8E2C26@teksavvy.com>    "Stanley F. Quayle" wrote:D > I had proposed CHARON-VAX as a replacement back in September 2004.C > They were already on their way to the replacement system.  *sigh*     C Did they replace it with tandem technology ? any other HP product ?     J Integration is "interesting" argument. I don't see why VMS system couldn'tL connect to other systems to pick the information and do all the fancy stuff.  N What I'd like to know is whether HP fought to maintain this VMS account or if,N as usual, they had already written off all VMS acocunts so they doN't feel bad when losing them.    ------------------------------  % Date: Tue, 26 Jul 2005 10:53:16 -0700 % From: DeanW <dean.woodward@gmail.com> 2 Subject: Re: Another VMS system bites the dust :-(7 Message-ID: <3f119ada0507261053619f3431@mail.gmail.com>   5 On 7/26/05, Stanley F. Quayle <stan@stanq.com> wrote: F > > But, that sounds like you were replacing hardware- not solving theD > > stated issue, "Integration"- or was there more to your proposal? >=20D > "Integration" is a frequent excuse for eliminating VMS.  There areE > gobs of integration options for VMS, but that's not the real issue.   A In law enforcement, "Integration" between 911, police (incident & A investigation) records, corrections, supervision (probation), the E courts- is the hot topic. In many places, that's at least four and as  many as six different systems.  C We're currently working an interface for a customer to pull 911/CAD B data to start the officer's incident reports, which (if someone isC arrested) can also pre-populates the the jail's booking data... The B police and jail systems are already "integrated" (They're both ourB product, and really, it's only one product), and interfaces to the courts have existed for years.  C I suspect the issue is that the Miami/Dade system was "home grown", > and nobody wanted to do the work to straighten it out. (HavingC encountered "home grown" before, I know how ugly the design can be, D but the memories are locked away and only accessible by handing me a beer... ;-)   E You're probably right that hardware (VAX) maintenance costs were also D an issue. In any case, they saw VMS as part of the problem- too many4 people associate "VMS" directly with "VAX". Too bad.   ------------------------------  % Date: Tue, 26 Jul 2005 08:01:20 -0500 ' From: "Earl Lakia" <elakia@hotmail.com> / Subject: Re: DEC Forms Calender and Time Panels 1 Message-ID: <h9qdnayvMNO6qXvfRVn-3w@netnitco.net>   E Currently using VT340s and Reflections emulators.  X-windows wouldn't F be an option.  However, if there is an x-windows version around, maybe I could convert it.    -earl   H "Bob Koehler" <koehler@eisner.nospam.encompasserve.org> wrote in message- news:febOKuUFIMqg@eisner.encompasserve.org... @ > In article <67idnXAy_NfWrXzfRVn-pg@netnitco.net>, "Earl Lakia" <elakia@hotmail.com> writes:C > > Anyone developed a Calender Panel or Clock panel for DEC Forms? H > > Something similar to what is available in the Visual Basic and other
 > > Microsoft  > > Environments?  > > H > > I can't justify writing my own, and it's ok to call me lazy as I am. > G >    Does it have to be in forms, or can you use the X-windows versions  >    that ship with VMS? >    ------------------------------  % Date: Tue, 26 Jul 2005 13:09:32 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> / Subject: Re: DEC Forms Calender and Time Panels , Message-ID: <42E66E4B.F097CFB2@teksavvy.com>   Earl Lakia wrote:  > G > Currently using VT340s and Reflections emulators.  X-windows wouldn't H > be an option.  However, if there is an x-windows version around, maybe > I could convert it.   M If you were FMS based instead of that DecForms thing, you could use ALL-IN-1, # it has built in calendar functions.    ------------------------------    Date: 26 Jul 2005 05:06:23 -0700! From: "Nazim" <nmanser@progis.de> 7 Subject: improve file system performance related to BEA C Message-ID: <1122379583.047365.230980@g44g2000cwa.googlegroups.com>    Hi all,   A we are running BEA weblogic on an alphaserver ES47 with 8 GB RAM. 1 openVMS version is 7.3-2 with the latest patches.   G we nioticed that the startup of BEA weblogic takes about 2-3 min, while $ in windows and unix it takes 25 sec.  C during the statup phase we noticed that there are a lot of exec RMS  activity  * the bea account has the following settings   $ show rms_default6 $!default values for sequential files block=1 buffer=11 $ set rms_default/sequential/block=127/buffer=255 + $!default values for indexed files buffer=1 # $ set rms_default/indexed/buffer=20  $!default values for extents $ set rms_default/extent=65000 $!   and   . RMS_SEQFILE_WBH   1    0    0   1 Coded-valu D  A Having read the manuals guide to file system applications and RMS B utilities reference, we want to set the following attibutes to allG files which belongs in the BEA directory tree to override some defaults % settings especially for the hotfiles.   C one method is to create the fdl file, create an empty file with the C following FDL attributes, so that BEA weblogic accesses these files  with the right FDL parameters.  E The problem is how to reduce the work for doing it with all the files = in the directory, my colleage cannot believe that there is no ; systemwide or diskwide possibility to set these parameters.   = FILE                      new setting         default setting   0 best_try_contiguous       yes                 no/ extension                 65000               0 0 deferred_write            yes                 no/ window_size               255                 0 0 file_monitoring           yes                 no   CONNECT   0 read_ahead                yes                 no0 write_behind              yes                 no      D we have no chance to interact with the programm which deals with theB files as it is java based, and java chooses always the RMS and FDL default settings.      any hint will be welcome.    regards,   Nazim Manser   ------------------------------  % Date: Tue, 26 Jul 2005 14:33:59 +0200 $ From: "Hein" <hein.nomail@hp.nomail>; Subject: Re: improve file system performance related to BEA , Message-ID: <42e62db4$1@usenet01.boi.hp.com>  - "Nazim" <nmanser@progis.de> wrote in message  = news:1122379583.047365.230980@g44g2000cwa.googlegroups.com... 	 > Hi all,  : I > we nioticed that the startup of BEA weblogic takes about 2-3 min, while & > in windows and unix it takes 25 sec. : F > during the statup phase we noticed that there are a lot of exec RMS 
 > activity  J It might be worth your while to further quantify the activity through RMS 
 STATISISTICS.   Check out my rms_stats tool in: 6 tp://h71000.www7.hp.com/freeware/freeware60/rms_tools/E You would have to enable stats on all 'suspect files' before the Bea   startup.J Then activate rms_stats with for example a 10 minute interval, all stats,  and that list of suspect files. D After the bea startup see the accumulated activity per suspect file.  , > the bea account has the following settings >  > $ show rms_default8 > $!default values for sequential files block=1 buffer=13 > $ set rms_default/sequential/block=127/buffer=255   J This is wrong. For starters the default is now 32 blocks and 2 buffers if  RAH is selected.M With extensive testing I have rarely seen an advantage in anything more than  
 8 buffers.@ I suggest you reduce to: SET RMS/SEQ/BLO=96/BUF=4 and try again.  - > $!default values for indexed files buffer=1 % > $ set rms_default/indexed/buffer=20   L Again, this is wrong. The default is 'deepest index + 2' which often equals  4 or 5. . Here, the suggest /buf=20 is not a bad choice.M They may have more benefit from global buffers IF, and only if, you can come  J up with a process to pre-open those files and keep them open. Hint: check " that rms_tools freeware directory.   > $!default values for extents  > $ set rms_default/extent=65000  H Fine, but possibly overkill. How big are files created/grown during the  startup.M This may cause a lot of work that will be thrown away, notably if the dik is   fragmented. * I would try a value like 1024 and compare.  0 > RMS_SEQFILE_WBH   1    0    0   1 Coded-valu D  ! That would be good for log files.   M Of course, a well written / ported application would not require any of this   and have it all built in.   C > Having read the manuals guide to file system applications and RMS  : E > one method is to create the fdl file, create an empty file with the E > following FDL attributes, so that BEA weblogic accesses these files   > with the right FDL parameters.  > Unlikely. You can only set PERMANENT file attributes with FDL.J Typically: allocation, extent, (best-try-)contig, bucket-size for non-seq  files.I Other, DYNAMIC/CONNECT attributes are only valid if the program actually  M uses the FDL op open the file which is unlikely. Pre-created files would not   remember those attribute.   ? > FILE                      new setting         default setting  > 2 > best_try_contiguous       yes                 no1 > extension                 65000               0 2 > deferred_write            yes                 no1 > window_size               255                 0 2 > file_monitoring           yes                 no  6 Window size can be set device-wide through SET VOLUME.4 The FDL control is DYNAMIC and will not be remember.K Deffered write is dynamic, not remembered and only applies to shared files.   	 > CONNECT 2 > read_ahead                yes                 no2 > write_behind              yes                 no  L Dynamic attributed. Ignored / Forgotten unless the fdl file is used for the  actual open by Bea: Unlikely  F > we have no chance to interact with the programm which deals with theM > files as it is java based, and java chooses always the RMS and FDL default   > settings.   A Work with the vendor (and HP) to improve the VMS implementation?! L Of course that will not provide a fix 'tomorrow', but maybe in some nearish  future?   
 Good luck!   Hein.    ------------------------------    Date: 26 Jul 2005 06:47:10 -0700! From: "Nazim" <nmanser@progis.de> 7 Subject: improve file system performance related to BEA C Message-ID: <1122385629.983273.232040@g43g2000cwa.googlegroups.com>    Hi all,   A we are running BEA weblogic on an alphaserver ES47 with 8 GB RAM. 1 openVMS version is 7.3-2 with the latest patches.   G we nioticed that the startup of BEA weblogic takes about 2-3 min, while $ in windows and unix it takes 25 sec.  C during the statup phase we noticed that there are a lot of exec RMS  activity  * the bea account has the following settings   $ show rms_default6 $!default values for sequential files block=1 buffer=11 $ set rms_default/sequential/block=127/buffer=255 + $!default values for indexed files buffer=1 # $ set rms_default/indexed/buffer=20  $!default values for extents $ set rms_default/extent=65000 $!   and   . RMS_SEQFILE_WBH   1    0    0   1 Coded-valu D  A Having read the manuals guide to file system applications and RMS B utilities reference, we want to set the following attibutes to allG files which belongs in the BEA directory tree to override some defaults % settings especially for the hotfiles.   C one method is to create the fdl file, create an empty file with the C following FDL attributes, so that BEA weblogic accesses these files  with the right FDL parameters.  E The problem is how to reduce the work for doing it with all the files = in the directory, my colleage cannot believe that there is no ; systemwide or diskwide possibility to set these parameters.   = FILE                      new setting         default setting   0 best_try_contiguous       yes                 no/ extension                 65000               0 0 deferred_write            yes                 no/ window_size               255                 0 0 file_monitoring           yes                 no   CONNECT   0 read_ahead                yes                 no0 write_behind              yes                 no      D we have no chance to interact with the programm which deals with theB files as it is java based, and java chooses always the RMS and FDL default settings.      any hint will be welcome.    regards,   Nazim Manser   ------------------------------    Date: 26 Jul 2005 08:20:59 -0700! From: "Nazim" <nmanser@progis.de> ; Subject: Re: improve file system performance related to BEA C Message-ID: <1122391259.187520.243310@g14g2000cwa.googlegroups.com>    thanks for your hints,   i have set the rms defaults    SET RMS/SEQ/BLO=96/BUF=4  $ and tested sequential read operation  0 $java -cp .:log4j.jar "FileCheckSum" log4j_1.zipF 2005-07-26 16:19:06,688  INFO [main] (FileCheckSum.java:24) - Starting reading file log4j_1.zipB 2005-07-26 16:20:28,364  INFO [main] (FileCheckSum.java:29) - Read 316415 bytes from log4j_1.zip   B during the read the CPU load was 149% (ES80 with 2 CPUS, 2 GB RAM) 70 % user mode 70 % kernel mode 10 % exec mode   time for rreading: 82 sec.  G i have turned the file statistics on on the log4j.zip (file to be read)  $rmsstats -o=A -r log4j.zip    here is the output  ?   0.00 IO/Ops.  Ops: 316524  Cache:   0% Global:  0%  log4j.zip @ Local  Cache trials       0  hit      0  read     0  write     0@ Global Cache trials       0  hit      0  read     0  write     0  ? Locks:             Enqueue     Dequeue     Convert         Wait 	 Block-ast L -----------------+----------+-----------+-----------+------------+----------@  Shared file:             0           0           0            0    0@  Local buffer:            0           0           0            0    0@  Global buffer:           0           0           0            0    0@  Append locks:            0           0           0            0    0@  Global section:          0           0           0            0    0@  Data record:             0           0           0            0  F Open:           1  Close:        1  Con:          1  Discon:         1F Rewind:         0  Flush:        0  Extend        0  blocks:         0F Outbufquo:      0  Xqpqio:       1  Truncate      0  blocks:         0F Reads           0  Bytes:        0  Writes        0  bytes:          0  3 Find seq:       0  key:          0  rfa:          0 F Get  seq:       1  key:     316523  rfa:          0  bytes:  162060288F Put  seq:       0  key:          0                   bytes:          0F Deletes:        0                   Updates       0  bytes:          03 Splits:         0  multy:        0  OutbufQuo     0     ' on a laptop 1,7 GHz windows XP 1 GB ram    it takes only 4 seconds.   regards,   Nazim Manser   ------------------------------    Date: 26 Jul 2005 08:53:59 -0700! From: "Nazim" <nmanser@progis.de> ; Subject: Re: improve file system performance related to BEA B Message-ID: <1122393239.724898.90830@g44g2000cwa.googlegroups.com>   Hein,   F here is the output of monitor RMS while execution of the java programm  1                               RMS FILE OPERATIONS /                                  on node IPPS01 3                             26-JUL-2005 17:40:37.26 9 (Sequential)  $1$DGA5:[BEA.Karim.FileCheckSum]log4j.jar;1 C Active Streams:     0          CUR        AVE        MIN        MAX   C     $GET Call Rate    (Seq)   0.00       0.00       0.00       0.00 C                       (Key)   0.00       0.00       0.00       0.00 C                       (RFA)   0.00       0.00       0.00       0.00 C     $FIND Call Rate   (Seq)   0.00       0.00       0.00       0.00 C                       (Key)   0.00       0.00       0.00       0.00 C                       (RFA)   0.00       0.00       0.00       0.00 C     $PUT Call Rate    (Seq)   0.00       0.00       0.00       0.00 C                       (Key)   0.00       0.00       0.00       0.00 C     $READ Call Rate           0.00       0.40       0.00      12.66 C     $WRITE Call Rate          0.00       0.00       0.00       0.00 C     $UPDATE Call Rate         0.00       0.00       0.00       0.00 C     $DELETE Call Rate         0.00       0.00       0.00       0.00 C     $TRUNCATE Call Rate       0.00       0.00       0.00       0.00 C     $EXTEND Call Rate         0.00       0.00       0.00       0.00 C     $FLUSH Call Rate          0.00       0.00       0.00       0.00    ------------------------------    Date: 26 Jul 2005 08:54:22 -0700! From: "Nazim" <nmanser@progis.de> ; Subject: Re: improve file system performance related to BEA C Message-ID: <1122393262.721872.176780@o13g2000cwo.googlegroups.com>    Hein,   F here is the output of monitor RMS while execution of the java programm  1                               RMS FILE OPERATIONS /                                  on node IPPS01 3                             26-JUL-2005 17:40:37.26 9 (Sequential)  $1$DGA5:[BEA.Karim.FileCheckSum]log4j.jar;1 C Active Streams:     0          CUR        AVE        MIN        MAX   C     $GET Call Rate    (Seq)   0.00       0.00       0.00       0.00 C                       (Key)   0.00       0.00       0.00       0.00 C                       (RFA)   0.00       0.00       0.00       0.00 C     $FIND Call Rate   (Seq)   0.00       0.00       0.00       0.00 C                       (Key)   0.00       0.00       0.00       0.00 C                       (RFA)   0.00       0.00       0.00       0.00 C     $PUT Call Rate    (Seq)   0.00       0.00       0.00       0.00 C                       (Key)   0.00       0.00       0.00       0.00 C     $READ Call Rate           0.00       0.40       0.00      12.66 C     $WRITE Call Rate          0.00       0.00       0.00       0.00 C     $UPDATE Call Rate         0.00       0.00       0.00       0.00 C     $DELETE Call Rate         0.00       0.00       0.00       0.00 C     $TRUNCATE Call Rate       0.00       0.00       0.00       0.00 C     $EXTEND Call Rate         0.00       0.00       0.00       0.00 C     $FLUSH Call Rate          0.00       0.00       0.00       0.00    ------------------------------    Date: 26 Jul 2005 11:19:51 -0500 From: briggs@encompasserve.org; Subject: Re: improve file system performance related to BEA 3 Message-ID: <Fzcgx$jtg1FS@eisner.encompasserve.org>   g In article <1122391259.187520.243310@g14g2000cwa.googlegroups.com>, "Nazim" <nmanser@progis.de> writes:  > thanks for your hints, >  > i have set the rms defaults  >  > SET RMS/SEQ/BLO=96/BUF=4 > & > and tested sequential read operation > 2 > $java -cp .:log4j.jar "FileCheckSum" log4j_1.zipH > 2005-07-26 16:19:06,688  INFO [main] (FileCheckSum.java:24) - Starting > reading file log4j_1.zipD > 2005-07-26 16:20:28,364  INFO [main] (FileCheckSum.java:29) - Read > 316415 bytes from log4j_1.zip    ^^^^^^ > D > during the read the CPU load was 149% (ES80 with 2 CPUS, 2 GB RAM) > 70 % user mode > 70 % kernel mode > 10 % exec mode >  > time for rreading: 82 sec. > I > i have turned the file statistics on on the log4j.zip (file to be read)  > $rmsstats -o=A -r log4j.zip  >  > here is the output > A >   0.00 IO/Ops.  Ops: 316524  Cache:   0% Global:  0%  log4j.zip                         ^^^^^^   H 316524 bytes in 316524 RMS calls.  I think we have identified a problem.  E It appears that both the application programmer and the programmer(s) G responsible for the Java port have never heard of the word "buffering".   B The Java port should be able to handle single character reads fromE the application without having to go out to RMS to satisfy every such ! request.  It should do buffering.   B The application should not use single character reads in the firstI place.  (I don't know enough Java to know if there is an fgets equivalent  available, however).   	John Briggs   ------------------------------  % Date: Tue, 26 Jul 2005 18:34:23 +0200 $ From: "Hein" <hein.nomail@hp.nomail>; Subject: Re: improve file system performance related to BEA * Message-ID: <42e6660c@usenet01.boi.hp.com>  - "Nazim" <nmanser@progis.de> wrote in message  = news:1122391259.187520.243310@g14g2000cwa.googlegroups.com...  > thanks for your hints, >  > i have set the rms defaults  > SET RMS/SEQ/BLO=96/BUF=4& > and tested sequential read operation  H Ok, at some point it would be intersting to see if this result is a lot , different from the original 127/255 setting.  D > during the read the CPU load was 149% (ES80 with 2 CPUS, 2 GB RAM)2 > 70 % user mode, 70 % kernel mode, 10 % exec mode  5 This suggests little or no room for RMS improvements. L A good chunk of the time is already user mode, so you can only expect a few  % improvement, not 10x. J There is also a lot of kernel mode, probably the XFC doing its read ahead  and buffer managent.3 Most of that seems to be overlapping the user time. J Again, if that was even much more optimized it would still only influence ) the end result a little, not a magnitude.    > time for rreading: 82 sec. > I > i have turned the file statistics on on the log4j.zip (file to be read)  > $rmsstats -o=A -r log4j.zip  : @ >  0.00 IO/Ops.  Ops: 316524  Cache:   0% Global:  0%  log4j.zipB > Local  Cache trials       0  hit      0  read     0  write     0  B I'm not sure why it is not counting reads. It should, best I know.5 Verify with ANAL/SYS... SET PROC... SHOW PROC/RMS=FSB ? That will show the data that rms_stats is supposed to be using,   H > Get  seq:       1  key:     316523  rfa:          0  bytes:  162060288   good.      In an other reply Nazim wrote:H > here is the output of monitor RMS while execution of the java programm  ; > (Sequential)  $1$DGA5:[BEA.Karim.FileCheckSum]log4j.jar;1 D >Active Streams:     0          CUR        AVE        MIN        MAX : D >    $READ Call Rate           0.00       0.40       0.00      12.66  5 So what we see here is that java is using 'block IO'. 8 The SET RMS /BUF/BLO are not directl in effect for this.J But if this is really the C-RTL doing its thing, then the SET RMS options L will be honored indirectlt (because the C RTL requests the values and tries H to honor them). If they are private calls to SYS$READ, then the SET RMS ! options are likely to be ignored.      Hein,    ------------------------------    Date: 26 Jul 2005 11:41:09 -0500 From: briggs@encompasserve.org; Subject: Re: improve file system performance related to BEA 3 Message-ID: <oquyOIn8$uw$@eisner.encompasserve.org>   g In article <1122391259.187520.243310@g14g2000cwa.googlegroups.com>, "Nazim" <nmanser@progis.de> writes: H > Get  seq:       1  key:     316523  rfa:          0  bytes:  162060288H                               ^^^^^^                           ^^^^^^^^^/ I missed this line when I commented previously.   C It looks to me as if we have one get for the first byte in the file ? and 316,523 "keyed" gets for the remaining 316,523 bytes.  That E probably translates into RMS "direct" access (possible for sequential ! files with fixed length records).   B Dividing 162060288 by 316523 it looks like we're reading 512 bytes for every byte in the file.   F Somebody has absolutely no clue about how to write an application, how to port one or both.   	John Briggs   ------------------------------    Date: 26 Jul 2005 10:20:57 -0700! From: "Nazim" <nmanser@progis.de> ; Subject: Re: improve file system performance related to BEA C Message-ID: <1122398457.695809.223060@g44g2000cwa.googlegroups.com>   5 here is the output of $analyze/system during the read    set process/index=xxx  show process /rms=fsb    FSB Address: 032CE000  ----------- $ OPEN:           1.  CLOSE:        0.$ CONNECT:        1.  DISCONN:      0.$ REWIND:         0.  FLUSH:        0.$ EXTEND:         0.  blocks:       0.$ TRUNCATE:       0.  blocks:       0.  6 FIND seq:       0.  key:          0.  rfa:          0.> GET  seq:       0.  key:          0.  rfa:          0.  bytes: 0.> PUT  seq:       0.  key:          0.                    bytes: 0.$ UPDATE:         0.  bytes:        0. DELETE:         0.  $ READ:          38.  bytes:  2433059.$ WRITE:          0.  bytes:        0.  G LOCAL  CACHE attempts:       0.  hits:       0.  read:       0.  write:       0. G GLOBAL CACHE attempts:       0.  hits:       0.  read:       0.  write:       0.  GLOBAL BUFFER INTERLOCKING: F    GBHSH Intlck Collisions:            0        GBH Intlck Collisions:        0D    GBHSH Held at Rundown:              0        GBH Held at Rundown:        0  ? LOCKS:            Enqueue    Dequeue    Convert       Block-ast ;  Shared file:           0.         0.         0.         0. ;  Local buffer:          0.         0.         0.         0. ;  Global buffer:         0.         0.         0.         0. ;  Shared append:         0.         0.         0.         0. ;  Global section:        0.         0.         0.         0. 0  Data record:           0.         0.         0.   XQP QIO:                1.  F BUCKET SPLIT (1) :      0.  SPLIT (N) :       0.  OUTBUFQUO:        0.  E DEV1 .. DEV5:    00000000   00000000   00000000   00000000   00000000          FSB Address: 033A4000  ----------- $ OPEN:           1.  CLOSE:        0.$ CONNECT:        1.  DISCONN:      0.$ REWIND:         0.  FLUSH:        0.$ EXTEND:         0.  blocks:       0.$ TRUNCATE:       0.  blocks:       0.  6 FIND seq:       0.  key:          0.  rfa:          0.> GET  seq:       0.  key:          0.  rfa:          0.  bytes: 0.> PUT  seq:       0.  key:          0.                    bytes: 0.$ UPDATE:         0.  bytes:        0. DELETE:         0.  $ READ:           1.  bytes:      479.$ WRITE:          0.  bytes:        0.  G LOCAL  CACHE attempts:       0.  hits:       0.  read:       0.  write:       0. G GLOBAL CACHE attempts:       0.  hits:       0.  read:       0.  write:       0.  GLOBAL BUFFER INTERLOCKING: F    GBHSH Intlck Collisions:            0        GBH Intlck Collisions:        0D    GBHSH Held at Rundown:              0        GBH Held at Rundown:        0  ? LOCKS:            Enqueue    Dequeue    Convert       Block-ast ;  Shared file:           0.         0.         0.         0. ;  Local buffer:          0.         0.         0.         0. ;  Global buffer:         0.         0.         0.         0. ;  Shared append:         0.         0.         0.         0. ;  Global section:        0.         0.         0.         0. 0  Data record:           0.         0.         0.   XQP QIO:                1.  F BUCKET SPLIT (1) :      0.  SPLIT (N) :       0.  OUTBUFQUO:        0.  E DEV1 .. DEV5:    00000000   00000000   00000000   00000000   00000000      FSB Address: 034AA000  ----------- $ OPEN:           1.  CLOSE:        0.$ CONNECT:        1.  DISCONN:      0.$ REWIND:         0.  FLUSH:        0.$ EXTEND:         0.  blocks:       0.$ TRUNCATE:       0.  blocks:       0.  6 FIND seq:       0.  key:          0.  rfa:          0.> GET  seq:       1.  key:     316489.  rfa:          0.  bytes:
 162042880.> PUT  seq:       0.  key:          0.                    bytes: 0.$ UPDATE:         0.  bytes:        0. DELETE:         0.  $ READ:           0.  bytes:        0.$ WRITE:          0.  bytes:        0.  G LOCAL  CACHE attempts:       0.  hits:       0.  read:       0.  write:       0. G GLOBAL CACHE attempts:       0.  hits:       0.  read:       0.  write:       0.  GLOBAL BUFFER INTERLOCKING: F    GBHSH Intlck Collisions:            0        GBH Intlck Collisions:        0D    GBHSH Held at Rundown:              0        GBH Held at Rundown:        0  ? LOCKS:            Enqueue    Dequeue    Convert       Block-ast ;  Shared file:           0.         0.         0.         0. ;  Local buffer:          0.         0.         0.         0. ;  Global buffer:         0.         0.         0.         0. ;  Shared append:         0.         0.         0.         0. ;  Global section:        0.         0.         0.         0. 0  Data record:           0.         0.         0.   XQP QIO:                1.  F BUCKET SPLIT (1) :      0.  SPLIT (N) :       0.  OUTBUFQUO:        0.  E DEV1 .. DEV5:    00000000   00000000   00000000   00000000   00000000    ------------------------------    Date: 26 Jul 2005 10:52:19 -0700! From: "Nazim" <nmanser@progis.de> ; Subject: Re: improve file system performance related to BEA C Message-ID: <1122400339.158569.147110@g43g2000cwa.googlegroups.com>   6 here i can put the source code of the java application       import java.io.*;  import org.apache.log4j.Logger;    public class FileCheckSum  {        static Logger logger;      public FileCheckSum()      {      }   *     public static void main(String args[])     { * /*  17*/        String filename = args[0];/ /*  18*/        File file = new File(filename);  /*  20*/        try                    { 2 /*  20*/            FileInputStream fistream = new FileInputStream(file);5 /*  21*/            BufferedInputStream istream = new  BufferedInputStream(fistream);  /*  23*/            int sum = 0;E /*  24*/            logger.info("Starting reading file " + filename); 2 /*  26*/            while(istream.available() > 1)                       { 7 /*  26*/                int read_byte = istream.read();  /*  27*/                sum++;                       } @ /*  29*/            logger.info("Read " + sum + " bytes from " +
 filename);                   } * /*  30*/        catch(Exception exception)                   {                    }       }        static       {> /*  10*/        logger = Logger.getLogger(FileCheckSum.class);      } }      regards,   Nazim Manser   ------------------------------  % Date: Tue, 26 Jul 2005 08:16:02 -0500 ' From: "Earl Lakia" <elakia@hotmail.com>   Subject: Re: Modbus TCP for VMS?1 Message-ID: <efOdnTJIi7MJqnvfRVn-qA@netnitco.net>   9 The communication library mentioned does more than simply > get a set of registers from a PLC.  However, contacting vendorD might also get more favorable pricing.  In fact a group of customers5 are serving a VMS global section to thier HMIs (e.g., ? iFix, InTouch, and Cimplicity).  Also a couple of DCL utilities 0 for adhock reads and for DCL simulation scripts.   -earl  lakia<n o s p a m>@ipact.com  2 "tadamsmar" <tadamsmar@yahoo.com> wrote in message< news:1122041647.878734.85250@g49g2000cwa.googlegroups.com...G > I am a developer for a VMS-based control system.  I need to interface G > to devices that have a Modbus TCP interface.  We are considering this * > as an alternative to RS-232 interfacing. > C > I don't have any experience with control and data acquistion over C > TCP/IP, but it seems that just about every new instrument that we A > consider interfacing these days supports TCP/IP communications.  > $ > Looks like I can get this package: > I > http://www.ipact.com/Products/Modicon/TCP%20Communication%20Library.htm  > 7 > but it will cost about $15,000 total for our 5 nodes.  > F > Anyone familiar with a "roll your own" or freeware approach to this? > 4 > Anyone have any experience with the Ipact product? >    ------------------------------    Date: 26 Jul 2005 07:29:28 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)  Subject: Re: Mozilla and ODS5 3 Message-ID: <EALtz2HJmDxZ@eisner.encompasserve.org>   a In article <_IudnWmq7bv85HjfRVn-oQ@adelphia.com>, "John E. Malmberg" <wb8tyw@qsl.network> writes:  > Mark Berryman wrote:F >> Does anyone know the magic incantation to get Mozilla to save filesK >> using ODS5 syntax, rather than converting them to ODS2 (and would you be  >> willing to share it)? > E > It appears to be hardcoded into the VMS_JACKETS shared image, so a  K > change would be required there to fix it.  There may be some other parts  % > of Mozilla that need to be changed.   E    Mozilla accesses VMS_JACKETS as a shared image.  It is possible to D    get the VMS_JACSKETS source from www.hp.com/go/vms and modify it.  B    I did that just because I want to see /logical-name rather thanB    /physical-name for disks when I open a file chooser in Mozilla.   ------------------------------  % Date: Tue, 26 Jul 2005 15:30:48 +0000 - From: David B Sneddon <dbsneddon@bigpond.com>  Subject: Re: Mozilla and ODS5 * Message-ID: <42E65728.6050605@bigpond.com>  ! Bob Koehler was overheard to say: c > In article <_IudnWmq7bv85HjfRVn-oQ@adelphia.com>, "John E. Malmberg" <wb8tyw@qsl.network> writes:  >  >>Mark Berryman wrote: >>F >>>Does anyone know the magic incantation to get Mozilla to save filesK >>>using ODS5 syntax, rather than converting them to ODS2 (and would you be  >>>willing to share it)? >>E >>It appears to be hardcoded into the VMS_JACKETS shared image, so a  K >>change would be required there to fix it.  There may be some other parts  % >>of Mozilla that need to be changed.  >  > G >    Mozilla accesses VMS_JACKETS as a shared image.  It is possible to F >    get the VMS_JACSKETS source from www.hp.com/go/vms and modify it. > D >    I did that just because I want to see /logical-name rather thanD >    /physical-name for disks when I open a file chooser in Mozilla.   Bob,  > Do you have a URL that points to the source?  I tried a search& and had a look around but had no luck.   Regards, Dave.  --  D David B Sneddon (dbs)  VMS Systems Programmer  dbsneddon@bigpond.comD Sneddo's quick guide ...     http://www.users.bigpond.com/dbsneddon/D DBS freeware     http://www.users.bigpond.com/dbsneddon/software.htm   ------------------------------    Date: 26 Jul 2005 10:48:44 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: Mozilla and ODS5 3 Message-ID: <DMZHJ+WO+66Y@eisner.encompasserve.org>   q In article <EALtz2HJmDxZ@eisner.encompasserve.org>, koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:   G >    Mozilla accesses VMS_JACKETS as a shared image.  It is possible to F >    get the VMS_JACSKETS source from www.hp.com/go/vms and modify it. > D >    I did that just because I want to see /logical-name rather thanD >    /physical-name for disks when I open a file chooser in Mozilla.  E That sounds like the original Mozilla author did not read the release H notes for VMS V2.0 which explained why one should eschew physical names.   ------------------------------    Date: 26 Jul 2005 09:01:28 -0700# From: "Bobby" <colemanr7@yahoo.com> 5 Subject: Printing while defaulted to remote directory C Message-ID: <1122392995.558027.261060@z14g2000cwz.googlegroups.com>    Hi all,   F I have a print device and queue configured on machine A (both point to7 a TCP printer).  I then do a "set default machineB"user G pass"::drive:[dir]" to machine B.  If I then proceed to print a file in C this remote directory by any means, i.e. print xxx.txt or otherwise > direct it to the device (LRA0:) or queue (sys$print) I get the following error:  B "%PRINT-E-REMOINSF,node name specification or /REMOTE missing from
 command line"   E Importantly, I must print using an explicit device name (i.e., LRA0:) @ since the software that is doing the printing does not recognize> queues. I am not logged into the remote host so I don't reallyG understand why this happens.  Any light on my dim understanding will be  much appreciated.    Thanks,  Bobby    ------------------------------  # Date: Tue, 26 Jul 2005 17:13:38 GMT 7 From: John Malmberg <malmberg@dskwld.zko.dec.compaq.hp> 9 Subject: Re: Printing while defaulted to remote directory 1 Message-ID: <6buFe.9255$C66.293@news.cpqcorp.net>    Bobby wrote:	 > Hi all,  > H > I have a print device and queue configured on machine A (both point to9 > a TCP printer).  I then do a "set default machineB"user # > pass"::drive:[dir]" to machine B.   F Many applications will not work correctly when you do that.  And many F will work, but will have other problems in regard to speed or machine  loading.  F If you can not log in directly to that system, I would recommend that K you look into NFS, DFS, or clustering so that the files appear to be local.   D I recommend that you not set your default directory to remote nodes.  I Also, I recommend to use proxies on DECNET, and only use the "user pass"  F feature for testing problems.  And at the systems I used to manage, I G disabled the ability of all accounts to use that notation and required  * only the use of network proxies on DECNET.  ' >  If I then proceed to print a file in E > this remote directory by any means, i.e. print xxx.txt or otherwise @ > direct it to the device (LRA0:) or queue (sys$print) I get the > following error: > D > "%PRINT-E-REMOINSF,node name specification or /REMOTE missing from > command line"   ? The syntax to copy the file to the LRA0: device is to use COPY  . machineb"user pass"::drive:[dir]file.ext LRA0:  G Of course that will only work if you have a printer connected to LRA0:  C or LRA0: spooled to a print queue.  A network printer would not be  # connected to LRA0: as stated below.   H I will try to get syntax to print the file through a print queue right, F but I never do that because you can not specify the remote queue name.  5 PRINT/REMOTE machineb"user pass"::drive:[dir]file.ext   3 That will print on the queue SYS$PRINT on machineb.   G > Importantly, I must print using an explicit device name (i.e., LRA0:) B > since the software that is doing the printing does not recognize@ > queues. I am not logged into the remote host so I don't reallyI > understand why this happens.  Any light on my dim understanding will be  > much appreciated.   F The error message above will only occur when the software understands B print queues and is not sending data directly to the LRA0: device.  G The fact that you have some unknown software application in play means  H that only someone with specific knowledge of that software and probably ? the configuration of your system can give you specific answers.   F Now LRA0: is the name of a parallel port, not a TCP printer.  You can F spool the LRA0: device to a print queue where the printer is actually I somewhere else.  Generally that only works for very small configurations.   I For network print operations that do not use a VMS visible terminal type  F device name and generic queues, a dummy LAT device can be created and  used as a spooled device.   F Once a spooled device is assigned to a queue, then any file copied to H that device will print on that queue with the default settings for that  print queue.  F But with out knowing more specifics, there are too many variations in E TCP printing and in the use of generic print queues to give out more   useful advice.   -John ! malmberg@dskwld.zko.dec.compaq.hp  Personal Opinion Only    ------------------------------  % Date: Tue, 26 Jul 2005 13:25:23 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 9 Subject: Re: Printing while defaulted to remote directory , Message-ID: <42E67201.7CDA61FC@teksavvy.com>   Bobby wrote:D > "%PRINT-E-REMOINSF,node name specification or /REMOTE missing from > command line"     L The print command simply gives the queue system a *file-id*, not a full fileL specification to queue. The printing symbiont is then expected to have local7 access to this file when it gets around to printing it.   J Your "set default node"user pass"::  will work interactively, but when youK print, you give the symbiont a file ID which  resides on a different system E and it then doesn't even know which node it resides on. (and may have L interesting side effects if the same file id exists on your local system and9 the symbiont would then try to access that file instead).     M With conventional serial print queues, you could set a serial device /SPOOLED I and associate a print queue to it. In the above case, you could then COPY % node"user pass"::[dir]file.ext  TXA5:    Look at HELP SET DEV /SPOOLED   J You might be able to create an LTAnnn device and then associate it to yourN real queue name. If you copy to that device, the syetem *MIGHT* send it to theA print queue instead of attempting to spew the data out of LTAnnn:    ------------------------------  % Date: Tue, 26 Jul 2005 02:33:36 -0400 ' From: Dave Froble <davef@tsoft-inc.com> 1 Subject: Re: Response issues on GS1280, VMS 7.3-2 0 Message-ID: <11ebm3ntmsiop6b@corp.supernews.com>   Lee Mah wrote: > Keith Parris wrote:  > 
 >> Lee wrote:  >>@ >>>     All I/O of production data files is co-ordinated throughD >>>     in-house developed library routines.  For every record read,B >>>     inserted, updated, or deleted, a translation of the XXXXXX@ >>>     logical is performed to see if a production data file isC >>>     being maintained (CONVERT/RECLAIM).  If the XXXXXX value is B >>>     anything other than OFF, then extra processing takes placeA >>>     for a data file that is being maintained.  The processing E >>>     that takes place on a file being maintained is a "journalling > >>>     concept.  This allows us to maintain a file with usersC >>>     actively entering data.  Once the maintenance is completed, 9 >>>     the users access the new data file transparently.  >> >> >>C >> Instead of translating the logical name for each I/O, you might  * >> consider one of a couple of approaches:J >> 1) Within the application, use a TQE to fire off a periodic routine to G >> translate the XXXXXX logical name and set a flag indicating whether  G >> maintenance work is in progress. Then for each I/O, first test this  H >> flag. Let's say you decide to do this polling once every 10 seconds. E >> Then you simply need to $WAIT slightly more than 10 seconds after  F >> defining the logical name XXXXXXX to be something other than "OFF" 4 >> before you actually start working with the files.J >> 2) At application startup time, use $ENQW to acquire a Protected Write J >> mode lock on a resource named XXXXXX (or whatever) with a Blocking AST D >> queued. Once this lock is granted, set a flag indicating that no G >> maintenance work is going on. Test this flag for each I/O. When you  I >> need to do file maintenance work, grab and hold a lock with EXclusive  G >> mode for the duration of the maintenance work. In the applications'  F >> Blocking AST routine, convert the PW lock down to NL mode, set the E >> flag so as to indicate that maintenance work is now going on, and  H >> queue an asynchronous request with $ENQ to convert your lock from NL ; >> to PR mode lock, with a completion AST which clears the  " >> "maintenance-in-progress" flag. >  >  >  > Thanks Keith,  > . > I'll forward your suggestion to the DB team.  H Without getting into the details, I'll second Keith's idea of using the I lock manager for what is, by your description above, basically a locking  D type of task.  Using logicals for such is a mis-use of the logicals 2 capability, and such mis-use can lead to problems.  B That said, getting a good grasp of usage of the DLM can be, well,  difficult comes to mind.   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Tue, 26 Jul 2005 11:12:32 +0200 $ From: "Hein" <hein.nomail@hp.nomail>1 Subject: Re: Response issues on GS1280, VMS 7.3-2 , Message-ID: <42e5fe7d$1@usenet01.boi.hp.com>  0 "Lee" <lytmah@telusplanet.net> wrote in message ' news:7BeFe.173767$on1.59196@clgrps13... 
 > Hein wrote: 3 >> "Lee" <lytmah@telusplanet.net> wrote in message  * >> news:yKSBe.131110$9A2.88439@edtnps89...J >>>Users run approx. 1,000 commands in the same format when they log into  >>>the cluster:  : 3 >>>$       DEF/NOLOG/JOB   FILE1          FILE1.DAT  : H > Those were just examples. We have different production domains on the $ > VMScluster  based on the username.  I I still think there are better ways. Notably instead of using a per-file   logical, I wouldI look towards using some [rooted.] device logicals to identify a group of   files in a 'domain'.  F > The username determines what production procedure to execute.  E.g., > $ IF USRNAM .EQS. "USER1" 
 > $   THEN > $   @SYS$MANAGER:USER1.COM  I Hmm... nitpicking perhaps, but that seems a maintenance nightmare + slow  
 execution.2 I would hope the actual procedure looks more like:3 $user_file = f$search("sys$manager:"+USRNAM+".COM") D $if user_file.eqs."" then user_file = "sys$manager:USER_DEFAULT.com" $@'user_file      E >> So... what is the application trying to achieve looking up XXXXXX. G >> Is it a debug flag? A start/stop flag? An alternative data location?  :  > For every record read,< > inserted, updated, or deleted, a translation of the XXXXXX: > logical is performed to see if a production data file is% > being maintained (CONVERT/RECLAIM).   J Ah, good stuff. I like the concept, but the implementation seems far from  optimal.C First of all the polling mechanisme chosen is relatively expensive. M Second it would also seem too late: If some application has a file open, and  K maintenance is required, how does that application know to close it unless  J it happens to need an operation on the file? Unless... the file is opened H before each operation and closed afterwards? (Yuck!) Also... how do you 2 handle 'context' as use in for example 'get next'.I How does the maintenance application know when the file is available for    maintenance... cluster wide... ?M All the above problems can be solved with a relatively simple locking scheme   with blocking AST.M Application users for a file grab a corresponding unique CR lock, specifying   a blocking AST. F Maintenance users try to grab the same lock in PW or EX mode, causing  blocking AST to trigger.K On the blocking AST toggle a local 'maintenance requested' flag similar to   the logical name reading. M If not in a 'do not disturb' mode, then just close  the file and convert the   lock down to NL.K When coming out of 'do not disturb mode' check the 'maintenance requested'   flag. H Before starting an operation re-aquire lock  and open file if not in CR  mode.   J Also... CONVERT/RECLAIM is rarely adequate for maintenance and rarely the  most effective. G If this was not just an example, then please (re-)evaluate using plain   CONVERT/FDL/FAST/NOSORT/STAT  E Since this stuff is all in libraries already there is good good that  J improvements can be made without major re-writes of the application logic  itself.    Again, great concepts! Hope this helps some,    Hein.    ------------------------------  # Date: Tue, 26 Jul 2005 12:01:10 GMT " From: Lee <lytmah@telusplanet.net>1 Subject: Re: Response issues on GS1280, VMS 7.3-2 , Message-ID: <aCpFe.104611$wr.37375@clgrps12>   Hein wrote: 2 > "Lee" <lytmah@telusplanet.net> wrote in message ) > news:7BeFe.173767$on1.59196@clgrps13...  > 
 >>Hein wrote:  >>3 >>>"Lee" <lytmah@telusplanet.net> wrote in message  * >>>news:yKSBe.131110$9A2.88439@edtnps89... >>> K >>>>Users run approx. 1,000 commands in the same format when they log into   >>>>the cluster: >  > :  > 4 >>>>$       DEF/NOLOG/JOB   FILE1          FILE1.DAT >  > :  > H >>Those were just examples. We have different production domains on the $ >>VMScluster  based on the username. >  > K > I still think there are better ways. Notably instead of using a per-file   > logical, I wouldK > look towards using some [rooted.] device logicals to identify a group of   > files in a 'domain'. >  > F >>The username determines what production procedure to execute.  E.g., >>$ IF USRNAM .EQS. "USER1" 
 >>$   THEN >>$   @SYS$MANAGER:USER1.COM >  > K > Hmm... nitpicking perhaps, but that seems a maintenance nightmare + slow   > execution.4 > I would hope the actual procedure looks more like:5 > $user_file = f$search("sys$manager:"+USRNAM+".COM") F > $if user_file.eqs."" then user_file = "sys$manager:USER_DEFAULT.com" > $@'user_file >  >  >  > E >>>So... what is the application trying to achieve looking up XXXXXX. G >>>Is it a debug flag? A start/stop flag? An alternative data location?  >  > :  >  >>For every record read,< >>inserted, updated, or deleted, a translation of the XXXXXX: >>logical is performed to see if a production data file is% >>being maintained (CONVERT/RECLAIM).  >  > L > Ah, good stuff. I like the concept, but the implementation seems far from 
 > optimal.E > First of all the polling mechanisme chosen is relatively expensive. O > Second it would also seem too late: If some application has a file open, and  M > maintenance is required, how does that application know to close it unless  L > it happens to need an operation on the file? Unless... the file is opened J > before each operation and closed afterwards? (Yuck!) Also... how do you 4 > handle 'context' as use in for example 'get next'.K > How does the maintenance application know when the file is available for  " > maintenance... cluster wide... ?O > All the above problems can be solved with a relatively simple locking scheme   > with blocking AST.O > Application users for a file grab a corresponding unique CR lock, specifying   > a blocking AST. H > Maintenance users try to grab the same lock in PW or EX mode, causing  > blocking AST to trigger.M > On the blocking AST toggle a local 'maintenance requested' flag similar to   > the logical name reading. O > If not in a 'do not disturb' mode, then just close  the file and convert the   > lock down to NL.M > When coming out of 'do not disturb mode' check the 'maintenance requested'   > flag. J > Before starting an operation re-aquire lock  and open file if not in CR  > mode.  > L > Also... CONVERT/RECLAIM is rarely adequate for maintenance and rarely the  > most effective. I > If this was not just an example, then please (re-)evaluate using plain   > CONVERT/FDL/FAST/NOSORT/STAT > G > Since this stuff is all in libraries already there is good good that  L > improvements can be made without major re-writes of the application logic 	 > itself.  >  > Again, great concepts! > Hope this helps some,  >  > Hein.  >  >     4 Again, what I've provided here are just examples off the top of my head.   0 The coding was implemented over 10 years ago and1 has undergone minor changes.  However, use of the 2 procedures has enabled us to make the cluster-wide2 applications available continuously for many years1 (through migrations from VAX to Alpha, from third  party disks to Storageworks...) 1 I suppose it's time to look at making the process 1 more efficient.  It's a matter of availability of 
 resources.  / After the migration from ES45 to GS1280, we did 1 notice some improvement in batch processing times  (as expected).  1 Up to this point, the following factors have been 0 identified has possible suspects in the response problem:
    - Network. -    - Disk fragmentation.  Has been turned off ! 	for some time due to DVE issues. -    - CPU imbalance.  Changes were recommended  	and implemented.   # Comparison of the two environments:     Four ES45's			Two GS1280's *    Four nodes			4 nodes + 2 separate nodes+    Each node with 2 CPU's	2 nodes - 4 CPU's  				2 nodes - 6 CPU's "    No increase in processing load.$    No change in networking topology.  ) Your suggestions are greatly appreciated.    ------------------------------  % Date: Tue, 26 Jul 2005 10:47:28 -0400 - From: William Webb <william.w.webb@gmail.com> $ Subject: Re: Storage shelf questions7 Message-ID: <8660a3a10507260747300e107a@mail.gmail.com>   1 On 7/24/05, H Vlems <nospam@what.ever.com> wrote: 8 > The storgae shelf in question is labeled : DS-BA356-JDD > I got it together with two separately shipped personality modules:L >  70-31490-01 (no other part no on the label) with two 68 pin SCSI connect= ors @ >  70-33067-02 or DS-BA35X-FA with one very small SCSI connector >=20 > Questions:L > 1) does the shelf accept disks like the DS-RZ1DF-VW and/or the DS-RZ1CB-V= W?B > 2) the 70-31490-01 has one bank of dip switches. I'm looking for > documentation L > 3) the DS-BA35X-FA has two sets of dip switches, need documentation for t= his  > unit as wellL > 4) the $64000 question (:-): is this combination going to work with a KZP= SA > or KZPCM-Dx adapter? >=20 > Hans Vlems >=20 >=20 >=20  	 Hans: =20   @ Send me an email with a real address and I'll see what I can do.   WWWebb   --=20 C NOTE: This email address is only used for noncommerical VMS-related  correspondence. C All unsolicited commercial email will be deemed to be a request for 8 services pursuant to the terms and conditions located at# http://bellsouthpwp.net/w/e/webbww/    ------------------------------    Date: 26 Jul 2005 04:48:42 -0700 From: icerq4a@spray.seY Subject: Re: Two New Itanium 2s Feature Faster Front-Side Bus - Is this new Montecito inf C Message-ID: <1122378522.566550.278800@g47g2000cwa.googlegroups.com>    Bill Todd wrote: > Keith Parris wrote: ! > > norm.raphael@metso.com wrote:  > > ( > >>  Is this new Montecito information? > >  > > < > > I certainly hadn't seen this statement in public before: > > I > > "Intel said it expects that systems running on the Montecito CPU will E > > deliver up to twice the performance, up to three times the system L > > bandwidth, and more than 2 1/2 times as much on-die cache as the current& > > generation of Itanium processors." > I > Intel has been making statements along these lines for many months, but > > only recently has it become completely clear (e.g., by theirI > quantitative cache comparison above) that they referred to *per-socket* - > performance, not to *per-core* performance.  > I > In other words, they expect each Montecito core to offer about the same J > performance as a current Madison II core (not too surprising, as they'llH > clock only at 1.8 GHz nominal, 2.0 GHz max), but since there'll be twoI > of those cores per socket the per-socket performance will about double.  > G > More specifically, 'up to twice the performance' should be relatively I > easy to achieve where there are enough concurrent threads to distribute B > over the dual cores and where the overall chip bandwidth isn't aI > constraint (by contrast, single-thread performance won't improve much). I >   But 'up to three times the system bandwidth' will occur only with new G > accompanying chipsets (Intel isn't developing any, but HP and Fujitsu I > reportedly are) that can match the new FSB speeds and (again) only on a J > per-socket, not a per-core, basis - so best-case per-core improvement isH > more like 50% with the new chipsets, and just plugging Montecitos intoH > existing systems won't achieve any per-core bandwidth increase at all.J > And 'more than 2 1/2 times as much on-die cache' yet again doesn't applyJ > per-core since I'm pretty sure that the two cores don't share cache withH > each other, so any given thread will have only about 33% more cache to > use than it has today. > F > Intel is also being outright mendacious in other performance claims.& > For example, this Inquirer article (E > http://www.theinquirer.net/?article=24459 ) uncritically parrots an F > Intel press release touting Montecito as offering 60% higher LINPACKI > performance than POWER5, without noting the fact that they're comparing I > the performance of a quad-chip, 8-core Montecito system against that of I > a dual-chip, quad-core POWER5 system (IBM does sell a quad-chip, 8-core H > POWER5 system, of course, and it handily beats the LINPACK performance0 > reported for the comparable Montecito system). > # > Just to keep the record straight.  >  > - bill  7 That Intel press release was indeed a really sorry one.   G But you are also wrong, Montecito does beat Power5 in LINPACK. There is G this confusion about which LINPACK benchmark Intel and IBM use and most ( people don't understand the differences.   ------------------------------  % Date: Tue, 26 Jul 2005 15:12:47 +0200 ( From: "Rudolf Wingert" <win@fom.fgan.de> Subject: VMS Backup for Windows 3 Message-ID: <000101c591e3$b76d6250$994614ac@wat153>    Hello,    F Is there any freeware tool to read OpenVMS backup tapes (EXAbyte) on a Windows system?    TIA and best regards R. Wingert    ------------------------------    Date: 26 Jul 2005 06:56:06 -0700 From: dooleys@snowy.net.au# Subject: Re: VMS Backup for Windows B Message-ID: <1122386166.495456.58820@g47g2000cwa.googlegroups.com>   Rudolf Wingert wrote:  > Hello, > H > Is there any freeware tool to read OpenVMS backup tapes (EXAbyte) on a > Windows system?  > ! > TIA and best regards R. Wingert < The sig tape announcement posted earlier today contained one; for "other systems" - I don't know if this includes windows  Fall 2004 VMS SIG Tape contents  VMSBACKUP4-1-1.ZIP Phil   ------------------------------  % Date: Tue, 26 Jul 2005 09:55:17 -0400  From: norm.raphael@metso.com# Subject: Re: VMS Backup for Windows Q Message-ID: <OFB3095FD5.E91C25D4-ON8525704A.004C3453-8525704A.004C808A@metso.com>   C "Rudolf Wingert" <win@fom.fgan.de> wrote on 07/26/2005 09:12:47 AM:    > Hello, > H > Is there any freeware tool to read OpenVMS backup tapes (EXAbyte) on a > Windows system?  > ! > TIA and best regards R. Wingert  >   I wonder if this would help you:    B Glenn Everhart <Everhart@gce.com> wrote on 07/25/2005 07:01:58 PM:  	 > Folks - H > The Fall 2004 edition of the VMS SIG tapes has been mailed to folks on. > the tree. If you need to get on, contact me. >  > Glenn Everhart > Everhart@gce.com > Smyrna, Del 19977  >  > Contents of the tape:  >  > Fall 2004 VMS / LT SIG Tapes > [VMSLT04B...] Directory Tree >  [snip] > G > [.NET]   Net stuff Things off internet, not necessarily VMS specific.  Bible C >       has text of KJV Bible Blassic is a BASIC interpreter in src B >       BOTAN is a crypto lib CDRtools are for burning cd-r CryptoA >       is another crypto lib DVD-P-RW-Tools is for burning DVD+R > >       Firebird is a DBMS FSP client is for FSP, like FTP butC >       connectionless GNUPG is a free PGP version, crypto for all. > >       INGRESr#-SRC-1 is the Ingres DBMS released open source= >       MSORT is a sort with many options OPENSSL is a crypto < >       library and can generate certificates QEMU is an x86@ >       emulator TECOCLINUX is a TECOC dialect for linux in src.> >       Updates Pete Siemsen's older one UNFS3 is a reader for? >       linux disk file structures UNZIP and ZIP are updates to 9 >       those compression utils XPDF is to view pdf files  >  ***** G >       WINE is a windows emulator VMSBACKUP is a reader for vms backup  >       tapes on other systems *****  > 3 >       AAAREADME.TXT;1     BIBLE-KJV_4_18_TAR.GZ;1 2 >       BLASSIC-0_10_0.TGZ;1 BOTAN-1_4_6_TAR.BZ2;16 >       BRLCAD-7_0_2_TAR.BZ2;1 CDRTOOLS-2_01_TAR.BZ2;1= >       CRYPTO-128.ZIP;1 DVD-P-RW-TOOLS-5_21_ 4_10_8.TAR-GZ;1 > >       FIREBIRD-1_5_2_4731_TAR.BZ2;1 FSPCLIENT-0_0_9_TAR.GZ;1C >       FSPROXY-07.ZIP;1 GNUPG-1_4_1_TAR.BZ2;1 INGRESR3-SRC-1_TAR.B @ >       Z2;1 LIBECC-0_11_0_TAR.GZ;1                  MSORT.TGZ;13 >       OCRAD-0_9_TAR.BZ2;1 OPENSSL-0_9_7G_TAR.GZ;1 7 >       P7ZIP_4_14_01_SRC_TAR.BZ2;1 QEMU-0_6_1_TAR.GZ;1 7 >       SOCAT-1_4_1_0_TAR.GZ;1 TECOCLINUX0398_TAR.G Z;1 A >       UNFS3-0_9_12_TAR.GZ;1                   UNZIP552_TAR.GZ;1 4 >       VMSBACKUP4-1-1.ZIP;1 WINE-20050419_TAR.BZ2;11 >       XPDF-3_00PL1.PATCH;1 XPDF-3_00PL2.PATCH;1 + >       XPDF-3_00_TAR.GZ;1  ZIP231_TAR.GZ;1  >  [snip]   ------------------------------    Date: 26 Jul 2005 07:26:15 -0700 From: "Arie" <arie@zeepost.nl># Subject: Re: VMS Backup for Windows B Message-ID: <1122387975.927453.85680@g44g2000cwa.googlegroups.com>   Rudolf Wingert schreef:  > Hello, > H > Is there any freeware tool to read OpenVMS backup tapes (EXAbyte) on a > Windows system?  > ! > TIA and best regards R. Wingert   9 You could try our OpenVMS TapeReader utility for Windows: 4 http://www.emulatorsinternational.com/en/taperdr.htm  G It's not freeware but it has a trial period, perhaps that is sufficient  for your purpose.   
 Best regards,   
 Arie de Groot  Emulators International    ------------------------------  % Date: Tue, 26 Jul 2005 13:29:56 +0200 / From: chateau <"chateau64[NOSPAM]"@ifrance.com> - Subject: xemacs efs problem since openVMS 7.3 3 Message-ID: <42e61eb4$0$6807$636a15ce@news.free.fr>    Hi,   ; I work with xemacs on linux to edit file on server openVMS. = My problem is  since openvms 7.3 with ftp server KERBEROS_V4, D open a file with comamnd C-x C-f /user@host:src/toto.c doesn't work. xemacs edits an empty file In my ftp buffer , i have :   . ftp> quote rnfr $1$DKA0:[user.src]gip_launch.c: 350 File $1$DKA0:[GIP54.SRC]gip_depart.c; will be renamed.! ftp> quote cwd $1$dka0:[user.src]  250-CWD command successful. 0 250 New default directory is $1$DKA0:[GIP54.SRC]% ftp> dir " " /home/titi/tmp/efsazAixD  200 PORT command successful.7 150 Opening data connection for   (192.168.3.107,33106) % 226 LIST Directory transfer complete.  ftp>     if anyone have a solution. thanks in advance.   chateau    ------------------------------    Date: 26 Jul 2005 07:35:11 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) 1 Subject: Re: xemacs efs problem since openVMS 7.3 3 Message-ID: <z4yusDv+d9Dk@eisner.encompasserve.org>   e In article <42e61eb4$0$6807$636a15ce@news.free.fr>, chateau <"chateau64[NOSPAM]"@ifrance.com> writes:   ' > ftp> dir " " /home/titi/tmp/efsazAixD  > 200 PORT command successful.9 > 150 Opening data connection for   (192.168.3.107,33106) ' > 226 LIST Directory transfer complete.   D    There's the bug.  xemacs should be sending NLST (usually the userF    command ls) instead of LIST (the user command dir).  NLST returns aD    predicatable, machine readable list of file names.  LIST providesA    a vendor and OS specific human readable list of file names and     related data.   ------------------------------  % Date: Tue, 26 Jul 2005 02:25:08 -0400 ' From: Dave Froble <davef@tsoft-inc.com> F Subject: Re: [OT] Rounding v Truncation, was: Re: Platform Support vs.0 Message-ID: <11ebljsdihchb70@corp.supernews.com>   Simon Clubley wrote:\ > In article <11e6169idjh2n3a@corp.supernews.com>, Dave Froble <davef@tsoft-inc.com> writes: > F >>When I had a problem with Visual Basic she defended VB and wouldn't J >>consider it wrong.  So I'll ask, actually maybe take a poll.  If a real I >>value is placed into an integer variable, how many people would expect  I >>the fraction to be truncated?  I'd expect most would.  I did.  VAX/DEC  D >>BASIC truncates.  Well, VB rounds, and fractional parts of .5 and H >>greater round up.  Sure screwed with the logic from the converted VAX 1 >>Basic applications.  Sure screwed with my mind.  >> >  > O > I would regard rounding to be the correct thing to do from a maths viewpoint.   C I'll throw my rather aged degree in Mathematics onto the pile.  My  C memories of integer operation rules is that any fractional part is  ' always truncated.  Hey, could be wrong.   D > However, I created two examples, one in Ada and another in C which: > exhibit different behaviour. Ada rounds and C truncates:   <snip>  N > The Ada behaviour is specified in the Ada Language Reference Manual (section@ > 4.6, line 33), see: http://www.adapower.com/rm95/RM-4-6.html : > F > 	If the target type is an integer type and the operand type is real,B > 	the result is rounded to the nearest integer (away from zero if* > 	exactly halfway between two integers).   F Now this I have no problem with.  If there is a specification such as H above, fine, anyone who makes use of the fine manual can determine what 
 to expect.  F I could find no reference to such rules for Visual Basic.  Not saying 4 that they don't exist, just that I cannot find such.  H Regardless, my past experiences are that in a computer language, moving H a real number into an integer variable causes any fractional part to be B truncated.  Try it in MACRO-32.  Ok, so I'm a bit lazy, and don't E include the FIX function when that's the desired result.  That comes  A from over 30 years of programming and being used to a particular  6 behavior.  Perhaps I should learn to be more specific.   --  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: Tue, 26 Jul 2005 10:15:05 +0200 - From: Alex van Denzel <vandenzel@hotmail.com> F Subject: Re: [OT] Rounding v Truncation, was: Re: Platform Support vs.6 Message-ID: <42e5f109$0$2193$dbd43001@news.wanadoo.nl>  \ > In article <11e6169idjh2n3a@corp.supernews.com>, Dave Froble <davef@tsoft-inc.com> writes: > F >>When I had a problem with Visual Basic she defended VB and wouldn't J >>consider it wrong.  So I'll ask, actually maybe take a poll.  If a real I >>value is placed into an integer variable, how many people would expect  I >>the fraction to be truncated?  I'd expect most would.  I did.  VAX/DEC  D >>BASIC truncates.  Well, VB rounds, and fractional parts of .5 and H >>greater round up.  Sure screwed with the logic from the converted VAX 1 >>Basic applications.  Sure screwed with my mind.   
 Just using      Dim value as Integer     value = 1.5    MsgBox value    Or even       10 a% = 1.5    20 print a%  G One would expect different results in different BASIC dialects. If the  I previous poster is right VB rounds, at least his version. Other dialects  G would reject the assignment (can't assign a floating point value to an  ( integer variable), others will truncate.  	 Better is       10 a% = INT(1.5)     20 print a%  E the INT() function is _defined_ to truncate. IIRC the usual BASIC of  H twenty+ years ago didn't have a function to round off values. The trick  I used was INT(value + .5)   --   Alex   ------------------------------  % Date: Tue, 26 Jul 2005 13:06:29 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> F Subject: Re: [OT] Rounding v Truncation, was: Re: Platform Support vs., Message-ID: <42E66D94.1AA8573A@teksavvy.com>   Dave Froble wrote:I > Regardless, my past experiences are that in a computer language, moving I > a real number into an integer variable causes any fractional part to be 
 > truncated.    % Logically, I would tend to agree too.   L A real number is comprised of an integer followed by decimals. When you move7 it to an integer, only the integer part should be used.   F truncating is a logical operation. You know exactly what you will get.  R "rounding" is a subjective operation.  If you have 10.4444444444444444444444444446  N Is that machine going to see the 6, and then round each digit as it moves leftM to come up with integer 11, or will it stop after n decimals and see only "4"  and keep the integer at 10 ?   ------------------------------    Date: 26 Jul 2005 12:49:05 -0500B From: clubley@remove_me.eisner.decus.org-Earth.UFP (Simon Clubley)F Subject: Re: [OT] Rounding v Truncation, was: Re: Platform Support vs.3 Message-ID: <dT7MgTK37TU9@eisner.encompasserve.org>   \ In article <42E66D94.1AA8573A@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes: > N > A real number is comprised of an integer followed by decimals. When you move9 > it to an integer, only the integer part should be used.  > H > truncating is a logical operation. You know exactly what you will get. > T > "rounding" is a subjective operation.  If you have 10.4444444444444444444444444446 >   > Erm, that value is less than 10.5, so the answer is always 10.  P > Is that machine going to see the 6, and then round each digit as it moves leftO > to come up with integer 11, or will it stop after n decimals and see only "4"  > and keep the integer at 10 ?   That's not how it works... :-)  G As for rounding versus truncation, I prefer rounding because it results J in a more accurate result. IE: rounding 3.6 results in a difference of 0.47 versus truncation which results in a difference of 0.6.    Simon.   --  B Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP       7 Microsoft: The Standard Oil Company of the 21st century    ------------------------------   End of INFO-VAX 2005.414 ************************