0 INFO-VAX	Sun, 04 Feb 2001	Volume 2001 : Issue 69      Contents: Re: Dave Cutler  Error message over LAT" How to configure DNS on our system& Re: How to configure DNS on our system& Re: How to configure DNS on our system& Re: How to configure DNS on our system& Re: How to configure DNS on our systemK Re: Is there a way to execute an Alpha OpenVMS GUI application on aremote s  Re: KZMSA Specifications Re: KZMSA Specifications Latest BIND exploit  Re: Latest BIND exploit  Re: Latest BIND exploit 	 Licensing  Re: Seeking FTP SERVER sources Re: Seeking FTP SERVER sources Re: Seeking FTP SERVER sources3 Re: Simple DECnet-IV Tasks Impossible under Phase-V  Re: Source listings contentsF Re: TCPIP 5.0A FTP Server Rejecting Connections - was RE: Where to get- Re: Unloading or deleting licenses in a batch - Re: Unloading or deleting licenses in a batch - Re: Unloading or deleting licenses in a batch - Re: Unloading or deleting licenses in a batch + Re: Unrecognized Individual Destination PDU ( Re: very sloooooow Glasfibre performance' Re: Writes flushed to Disk on Dismount? ' RE: Writes flushed to Disk on Dismount? ' RE: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' RE: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? 6 Re: [Q] How to Highlight a text in fortran 77 on VMS ?6 Re: [Q] How to Highlight a text in fortran 77 on VMS ?6 Re: [Q] How to Highlight a text in fortran 77 on VMS ?6 Re: [Q] How to Highlight a text in fortran 77 on VMS ?6 Re: [Q] How to Highlight a text in fortran 77 on VMS ?  F ----------------------------------------------------------------------  % Date: Sat, 03 Feb 2001 14:28:17 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca>  Subject: Re: Dave Cutler, Message-ID: <3A7C5BCF.9C67100A@videotron.ca>   Christof Brass wrote: = > headed Micro$oft systems. Why doesn't Micro$oft support the  > Alpha processor??   H Why should a company spend any money porting to an architecture that hasP nearly zero marketing and whose markets are being limited to just a few niches ?  M The Alpha chip may not be dead and there may be a few more iterations left in N its lifetime, but not only has it lost some of its speed/performance edge, butN Compaq has no interest in spreading Alpha to the rest of its product line. The- chip may not be dead, but its marketplace is.   M If Microsoft were to procductize Windows on Alpha, how would Compaq sell it ? M Since Microsoft is only interested in dealing with the Compaq parts that have L the permission to grow and advertise heavily, this excludes anything running
 on the Alpha.   J And of Compaq were to start selling Alphas at the lower end, how would itsD relationship with Intel go ? And would Compaq make any more money ?   N Besides, the current Alpha performance is probably either marginally higher or+ about the same as the 8086. So why bother ?   J You know, now that Microsoft is entering the hardware business, I think itN would be a riot if Microsoft were to start building Alpha based computers, putM NT on them and sell them with all the marketing might MS has. And it would be M an even bigger riot if Microsoft were to start selling VMS to all the markets K that Compaq refuses to sell VMS to. What a slap in the face and kick in the  butt Compaq would be getting.   L But in many ways, that is what Compaq deserves. Someone else should start toN sell VMS and Alpha in products Compaq refuses to sell them to and prove to theN world how incomopetent Compaq was when t decided to limit the market for alpha and vms.   ------------------------------  % Date: Sat, 03 Feb 2001 14:38:55 -0500 2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: Error message over LAT 7 Message-ID: <200102031439_MC2-C421-74CB@compuserve.com>   % Message text written by "Paul Oswald" F >Could anybody help please. I have a BOCA ticket printer attached to a DS90M H DECserver. I am getting an intermittent error message along the lines of  ? 'Duplicate ARP source address 172.17.254.10 received from 00-'.   J This IP address belongs to a Hub on the PC LAN. I have attached a copy of=   the port settings for reference.   ' Any help provided would be appreciated.  <   D         Uuencoding what is presumably ASCII text (the port settings)J imposes an unreasonable burden on those of use who read our mail on PCs n= ot equipped with UUDECODE!   J         If I wanted to read this gibberish, I'd have to copy it over to m= y G Alpha, find my UUDECODE program, and run it.  I very seldom do this and J never  for someone asking a question concerning the ASCII text that they = so thoughtfully encrypted!    ------------------------------  % Date: Sat, 03 Feb 2001 14:38:56 -0500 2 From: "Richard B. Gilbert" <DRAGON@compuserve.com>+ Subject: How to configure DNS on our system 7 Message-ID: <200102031439_MC2-C421-74CC@compuserve.com>   = Message text written by INTERNET:freestyle_london@my-deja.com H >We have a VMS Manager but he tells me he does not know how to configure DNS !   F The reason we want to configure DNS is so that we can use names rather than IP address's for printing.   F All of our printers are HP JetDirect cards running TCP/IP. He metioned4 a UCX table does that mean anything to anyone here ?  G What I am looking for is a resource (not a book) that I can give to him A to configure this. Also any guidelines people have on printing to E JetDirect would be great. As far as I know it prints to it via a port # number of something like 9001 etc..  <   J         You should not be surprised.  The average VMS system manager neve= r J has occasion to configure a DNS!   I have done it once in seventeen years=   as a VMS System manager.  A         It should be a lot easier than it actually is, and better  documented.  =    G         If all you want to do is to print from your VMS host, you don't  need a DNS, just a hosts file.G Your system manager can do a UCX SET HOST mumble /ADDRESS=3D10.11.12.13 0 /ALIAS=3Dxxx for each printer and there you are.  J         If you want to be able to print from PC's you do need a DNS.  Get=  J him a set of UCX documentation, copies of RFCs 1032, 1033, 1034, 1035 and=  G 1183.  Get him a copy of "DNS and BIND" by Paul Albitz and Cricket Liu, J O'Reilly & Associates, Inc. 1994.  Give him a couple of weeks to review a= llH this stuff and another couple of weeks of trial and error to get the DNS working.  D         You could hire a consultant who has done this before and whoJ already has all the books, notes, and secret handshakes.  I might be able=  F to take a week or two off and do it for you next month.  (With luck, I= might manage the trick in a day but better allow for a week.)    ------------------------------  % Date: Sat, 03 Feb 2001 16:15:58 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca> / Subject: Re: How to configure DNS on our system , Message-ID: <3A7C7506.34C10A49@videotron.ca>   "Richard B. Gilbert" wrote: C >         It should be a lot easier than it actually is, and better 
 > documented.   M Actually, when I installed TCPIP V5.0, I didn't have any problems configuring N the BIND server the first time. Where it gets very un-obvious is when you wantL to make changes to it. the SET HOST command doesn't actually add the host toW the real database, you have to do a CONVERT/VMS to get that data to go into production.   N But in the initial menus of the configuration menu, it was easy to understand.  K Now, when I will have to figure out is how to get that node to act as a DNS S server that will serve my local names as well as names it gets from the internet...   G > Your system manager can do a UCX SET HOST mumble /ADDRESS=10.11.12.13 0 > /ALIAS=xxx for each printer and there you are.  E Don't forget the CONVERT command otherwise that doesn't seem to work.    ------------------------------  # Date: Sat, 03 Feb 2001 21:57:56 GMT 3 From: Tim Llewellyn <tim.llewellyn@cableinet.co.uk> / Subject: Re: How to configure DNS on our system / Message-ID: <3A7C7DD2.E3680FF7@cableinet.co.uk>    "Richard B. Gilbert" wrote:  >    I >         If all you want to do is to print from your VMS host, you don't   > need a DNS, just a hosts file.G > Your system manager can do a UCX SET HOST mumble /ADDRESS=10.11.12.13 0 > /ALIAS=xxx for each printer and there you are.   D In fact, I don't even bother putting the printers in the local hosts file, F just hardcode the IP address in the queue definition and let the queue manager 
 handle it.  E Saves  bit of time configuring queues and one less thing to go wrong.    regards    ------------------------------  $ Date: Sat, 3 Feb 2001 17:52:56 -0500  From: John Santos <JOHN@egh.com>/ Subject: Re: How to configure DNS on our system 5 Message-ID: <1010203174845.3117G-100000@Ives.egh.com>   - On Sat, 3 Feb 2001, Richard B. Gilbert wrote:   ? > Message text written by INTERNET:freestyle_london@my-deja.com J > >We have a VMS Manager but he tells me he does not know how to configure > DNS !  [...] K >         If you want to be able to print from PC's you do need a DNS.  Get K > him a set of UCX documentation, copies of RFCs 1032, 1033, 1034, 1035 and I > 1183.  Get him a copy of "DNS and BIND" by Paul Albitz and Cricket Liu, M > O'Reilly & Associates, Inc. 1994.  Give him a couple of weeks to review all J > this stuff and another couple of weeks of trial and error to get the DNS
 > working.  D There is a 3rd edition (Sept. 1998) that covers BIND V8.  I stronglyB recommend looking for this edition, since the recent VMS IP stacksF all do Version 8 BIND and the configuration files are quite different.   --   John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------   Date: 3 Feb 2001 23:59:57 GMT ) From: leslie@clio.rice.edu (Jerry Leslie) / Subject: Re: How to configure DNS on our system ' Message-ID: <95i61t$2qv$1@joe.rice.edu>   ! John Santos (JOHN@egh.com) wrote:  : F : There is a 3rd edition (Sept. 1998) that covers BIND V8.  I stronglyD : recommend looking for this edition, since the recent VMS IP stacksH : all do Version 8 BIND and the configuration files are quite different. :  : --  
 : John Santos  : Evans Griffiths & Hart, Inc. : 781-861-0670 ext 539  G According to the attached article from the comp.protocols.dns.bind news C group, there is a fourth addition that should be available shortly.    --Jerry Leslie      N ==============================================================================* From: "Cricket Liu" <cricket@VeriSign.com># Newsgroups: comp.protocols.dns.bind ' Subject: Re: DNS & BIND (O'Reilly book)   Date: 27 Jan 2001 11:46:58 -0800 Organization: none	 Lines: 18  Approved: usenet@vix.com( Message-ID: <94v8ji$m8j@pub3.rc.vix.com>E References: <94k81r$rf1@pub3.rc.vix.com> <94kbhb$st4@pub3.rc.vix.com> " NNTP-Posting-Host: pub3.rc.vix.com Mime-Version: 1.0 S X-Trace: isrv4.pa.vix.com 980624637 25045 204.152.186.34 (27 Jan 2001 19:43:57 GMT)  X-Complaints-To: abuse@isc.org+ NNTP-Posting-Date: 27 Jan 2001 19:43:57 GMT ( Return-Path: <bind-users-bounce@isc.org>J X-Original-Message-ID: <005a01c08897$baeb0ea0$7dc2a8ce@boulder.acmebw.com>
 X-Priority: 3  X-Msmail-Priority: Normal 2 X-Mailer: Microsoft Outlook Express 5.50.4133.24008 X-Mimeole: Produced By Microsoft MimeOLE V5.50.4133.2400$ Errors-To: bind-users-bounce@isc.org X-List-Id: <bind-users.isc.org>     H > >Should I buy the book that only covers V8, or should I wait till theyK > >publish something that covers V9? Is there already a version that covers  V9?  > J > Cricket will probably be able to respond on how soon he expects the next > edition to be released.   E The fourth edition of "DNS and BIND" is in production right now.  The K time it takes to get through production can vary a little, but I expect the . new edition to be available in a month or two.  , FYI, it covers BIND through 8.2.3 and 9.1.0.   cricket    ------------------------------  % Date: Sat, 03 Feb 2001 23:15:52 -0500 # From: sol gongola <sol@adldata.com> T Subject: Re: Is there a way to execute an Alpha OpenVMS GUI application on aremote s' Message-ID: <3A7CD778.6C1F@adldata.com>   7 You can't force a terminal emulator session to open or  7 be created from VMS any more than you can force a real  6 vt terminal to magically exist :) unless you are using" something like X windows or xterm.  6 On the other hand, if a second vt emulator session is : already open on  your PC (or you have a second VT terminal; available) and you know its terminal number/ID, you should  4 be able have some code open the terminal and use it.   sol gongola    1 > On Fri, 02 Feb 2001 13:31:29 +0000, Nigel Arnot & > <sysmgr@maxwell.ph.kcl.ac.uk> wrote: > 
 > Hi folks > ? > Sorry, I should have meant "CUI" the ole Character-based User 0 > Interface (or is that "Cheap User Interface"?) > E > Basically we have users coming in from PCs (via Telnet sessions) or H > dumb ole VT220 terminals accessing the main application running on theG > DEC 4700 machines (app was built back in the 80's).  From within that E > main application, we would need to somehow remote login (rlogin? or B > telnet?) into the Compaq Alpha server and lauching another "CUI"A > application.   This would be displayed on the user's PCs telnet $ > session or on the VT220 terminals. >  > Hope this is clearer.  >  > Thanks > 
 > >> Hi folks  > >>K > >> We have users running a GUI VAX VMS application on a DEC 4700 (running I > >> VAX VMS 5.5-2).   There is also another GUI application that resides 5 > >> on an Alpha DS20E machine running OpenVMS 7.1-2.  > >>K > >> We would like to include in the menu in the application running on the K > >> DEC 4700 the name of the remote application.  When a user selects this J > >> menu item, we need to somehow automatically log the user to the AlphaH > >> DS20E and start the application.  This should be transparent to theF > >> user (hs/she doesn't even know that the application that they are' > >> accessing are on another machine).  > >>J > >> Can this be done?  Or are there any tools/3rd party packages that can
 > >> do this?  > >> > > @ > >Assuming by GUI, you mean X-windows, it should be quite easy. > > B > >The menu application first needs to establish where the user is@ > >coming from (ie that which is needed to execute a SET DISPLAYH > >that will send subsequent X-displays back to that user. $SHOW DISPLAYE > >(from an appropriate context!) is one way to get this information.  > > H > >Next, it has to tell the DS20E to run that application in the contextK > >of the chosen user. The way I'm familiar with is DECNET IV task-to-task. C > >You define the task in the DS20's task database (a .COM or .EXE) > > >and fire it up from the menu app just by opening the "file"> > >called DS20"user password"::"task=name" for read and write. > > F > >This will fire up a process in the user's context running the task-B > >defined .COM, and that should open the file SYS$NET read-write. > > J > >There's now a two-way channel between the menu app and the remote task.D > >So the nenu app can write the information necessary to define theF > >display application to that pipe, the task can read it, execute theH > >appropriate SET DISPLAY command to send subsequent X displays back toP > >the appropriate X-display station, and fire up the appropriate X-application. > > P > >DECnet V is much more complex to set up, but basically similar. AlternativelyJ > >you can use TCP/IP (UCX) and RSH or REXEC. If the DS20 app doesn't haveP > >to run in the context of a variable user, you can define a particular user-IDK > >to run it in the task definition, and miss the out "user password" field N > >(and the associated possible risk of passwords flying around your network).@ > >It's also not needed if the user-id needed on the DS20 is theM > >same as the user-id that the menu-app is running under and provided DECNET  > >proxy logins are enabled. > > N > >It might be easier (depending on what the user's GUI-display server is/are)O > >to define a shortcut on that platform that goes straight to the DS20, rather F > >than involving that middle-man menu app. There again, it might not. > > > > >> Thanks in advance (I'm a VAX newbie so go easy on me ...) > >> > >V > >You're welcome! > >eJ > >Appended an example, all in DCL. If this procedure is defined as a task) > >on a remote machine REMOTE, you can do  > >n > >@tell REMOTE dcl_command  > >o > >and see the results.t > >e > >       Yours, > >               Nigel Arnotc* > >               NRA@MAXWELL.PH.KCL.AC.UK > > G > >               "In the beginning there was nothing, which exploded."c > >sL > >$!-----------------------------------------------------------------------/ > >$ if f$mode().EQS."NETWORK" then goto remoteT5 > >$ if P1.EQS."" then inquire/nopunc p1 "_Node:    "e5 > >$ if P2.EQS."" then inquire/nopunc p2 "_Command: "p > >$ p1 = p1 - "::"  > >$ if p1.eqs."" then $exit  > >$ on controly then goto close% > >$ open/write net 'p1'::"task=tell" 8 > >$ write net P2+" ''P3' ''P4' ''P5' ''P6' ''P7' ''P8'" > >$ type net: > >$ close: close nete	 > >$ exit  > >$remote:i7 > >$ if "''tell_first_use'".nes."" then goto setup_done P > >$! there is a bug somewhere that prevents us getting local DEC tables through0 > >$! the user's UAF entry by proxy login. So.... > >$ set command/tables=sys_qec:dcl_tables.exe > >$ tell_first_use == "FALSE" > >$setup_done:r > >$ open/write net sys$net: > >$ read net commandd& > >$ write sys$output "TELL> ",COMMAND > >$ ass net.out sys$output 
 > >$ set noont > >$ 'command' > >$ stat = $statush > >$ deass sys$output  > >$ set on 3 > >$ if f$search("net.out").eqs."" then goto no_outo > >$ copy net.out net: > >$ delete net.out;*e > >$ goto out_done > >$no_out:fG > >$ write net "%TELL-I-NO_OUTPUT, command did not write to SYS$OUTPUT"?
 > >$out_done:o0 > >$ if .not.stat then write net f$message(stat) > >$ close net   ------------------------------  % Date: Sat, 03 Feb 2001 15:11:21 -050012 From: rdeininger@mindspring.com (Robert Deininger)! Subject: Re: KZMSA SpecificationsrL Message-ID: <rdeininger-0302011511210001@user-2ivebes.dialup.mindspring.com>  U In article <3A7C362F.9A6B0C25@mmaz.com>, "Barry Treahy, Jr." <treahy@mmaz.com> wrote:h  < > Anyone know of a detailed source for the KZMSA-AA adapter?  [ I see it is mentioned in "Guidlines for Cluster Configurations", as an XMI-to-SCSI adapter.h  : Maybe you need an old VAX-era Systems and Options Catalog?  ] Or try searching for KZMSA at northernlight.com?  I see lots of hits, but I didn't read them.h   The only one I clicked was:hI     http://www.compaq.pl/alphaserver/options/asgs140/asgs140_options.htmlt  Y Which appears to know something about a KZMSA-AB, but my broser can't get past the menus.    -- i Robert Deininger rdeininger@mindspring.come   ------------------------------  % Date: Sat, 03 Feb 2001 17:41:55 -0700s+ From: "Barry Treahy, Jr." <treahy@mmaz.com> ! Subject: Re: KZMSA Specifications ( Message-ID: <3A7CA553.5879250C@mmaz.com>   Robert Deininger wrote:A  L > > Anyone know of a detailed source for the KZMSA-AA adapter specification? > ] > I see it is mentioned in "Guidlines for Cluster Configurations", as an XMI-to-SCSI adapter.  > < > Maybe you need an old VAX-era Systems and Options Catalog?  ^ I have several old SOC's (95 & 96), but nothing that specifically talks of the KZMSA.  Some ofa the web references imply that this is an Alpha only XMI to SCSI adapter.  I hoped that by located ] a technical specification for it to determine if I can make it work for my project (one woulde\ think this would be easy at Compaq but guess again).  Too bad Compaq didn't leave prior year_ SOC's on-line.  With the price of storage, it is impossible to understand why Compaq insists onh9 removing all of the old historical VAX & VMS information.h   > The only one I clicked was:fK >     http://www.compaq.pl/alphaserver/options/asgs140/asgs140_options.html1  \ Like most of Compaq's VAX and VMS lobotomized site, this leads to be page not found error...   Barrye   --  ? Barry Treahy, Jr  *  Midwest Microwave  *  Vice President & CIOi  A E-mail: Treahy@mmaz.com * Phone: 480/314-1320 * FAX: 480/661-7028k   ------------------------------  % Date: Sat, 03 Feb 2001 14:38:58 -0500r2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: Latest BIND exploit7 Message-ID: <200102031439_MC2-C421-74CD@compuserve.com>h  B Message text written by "Stephen Eickhoff (remove the - to reply)"G >I am a hobbyist running DEC TCP/IP 5.0 on OpenVMS VAX 7.1. What are myg optionsp> for updating BIND to secure my system from the latest exploit? <u  H         It would be easier to address your question if we knew just what  "exploit" you were asking about.   ------------------------------  % Date: Sat, 03 Feb 2001 15:46:57 -0500 F From: "Stephen Eickhoff (remove the - to reply)" <s-eickhoff@capu.net>  Subject: Re: Latest BIND exploit( Message-ID: <3A7C6E41.C455969E@capu.net>   "Richard B. Gilbert" wrote:- > D > Message text written by "Stephen Eickhoff (remove the - to reply)"I > >I am a hobbyist running DEC TCP/IP 5.0 on OpenVMS VAX 7.1. What are my 	 > options @ > for updating BIND to secure my system from the latest exploit? > <  > J >         It would be easier to address your question if we knew just what" > "exploit" you were asking about.  = This one: http://news.cnet.com/news/0-1003-201-4638816-0.html    Excerpt:I 	The vulnerabilities are mostly so-called buffer overflows, which allow acK         specially formatted command to cause a computer to crash or executenM         arbitrary code. The flaws affect BIND versions 4 and 8, but have been 4         corrected in versions 4.9.8, 8.2.3 and 9.1.    -- r" ----------------------------------          Stephen Eickhoffs           Ardmore, PAb" ----------------------------------   ------------------------------  % Date: Sat, 03 Feb 2001 23:24:23 -0500v2 From: "Richard B. Gilbert" <DRAGON@compuserve.com>  Subject: Re: Latest BIND exploit7 Message-ID: <200102032324_MC2-C42A-BFAE@compuserve.com>s  B Message text written by "Stephen Eickhoff (remove the - to reply)" >"Richard B. Gilbert" wrote: > =y  D > Message text written by "Stephen Eickhoff (remove the - to reply)"J > >I am a hobbyist running DEC TCP/IP 5.0 on OpenVMS VAX 7.1. What are my=  	 > options @ > for updating BIND to secure my system from the latest exploit? > <  > =o  J >         It would be easier to address your question if we knew just wha= t1" > "exploit" you were asking about.  = This one: http://news.cnet.com/news/0-1003-201-4638816-0.html&   Excerpt:H         The vulnerabilities are mostly so-called buffer overflows, which allow auJ         specially formatted command to cause a computer to crash or execu= teH         arbitrary code. The flaws affect BIND versions 4 and 8, but have been5         corrected in versions 4.9.8, 8.2.3 and 9.1. =    <k  " Apply the latest patches and pray!  D There's some chance the Compaq TCP/IP Services for VMS (formerly andJ forever known as UCX) is not running the raw Berkeley code but I wouldn't=  C count on it.  UCX and True64 TCP/IP share a common code base. . . .:   Or don't run a Bind server.s   ------------------------------  % Date: Sat, 03 Feb 2001 22:37:37 -0800w) From: Wayne Holland <wholland@tscnet.com>  Subject: LicensingO Message-ID: <C849ADC255F581E4.294898C2CCF425FE.8EB43A6F948788E6@lp.airnews.net>a  E If I get a used VAX with its O/S intact, and I eventually fire it up,tA and find that the license has expired,  what do I do?? (hobbyist)H   ------------------------------   Date: 3 Feb 2001 16:45:55 -0500h9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)i' Subject: Re: Seeking FTP SERVER sourceso3 Message-ID: <azdRmzKcf24l@eisner.encompasserve.org>w  Y In article <95h9t1$4gk$1@personal.myavista.com>, "Lee Gillie" <Lee@NOSPAModp.com> writes:e  ) > Have C, C++, Basic, Macro32 (no BLISS).   A By "Have" I presume you are talking about programming experience.e  7 The Bliss compiler is on the Freeware Discs these days.h   ------------------------------  $ Date: Sat, 3 Feb 2001 19:11:14 -0500  From: John Santos <JOHN@egh.com>' Subject: Re: Seeking FTP SERVER sources:5 Message-ID: <1010203190438.3117I-100000@Ives.egh.com>t  % On Sat, 3 Feb 2001, Lee Gillie wrote:N  F > Looking for a freeware, openware, shareware FTP server with sources.? > I need basic FTP, and to add extensions via the SITE command. D > Have C, C++, Basic, Macro32 (no BLISS).  This is for VAX with UCX.G > I've done this before, for NT, writing from scratch, but hoping there.E > will be an existing VMS project I can use to jumpstart this effort.IF > Even to just let me know that you are aware this doesn't exist would > help.  > < > Thanks in advance for any hints.  Best regard - Lee Gillie  	 Hi, Lee -   A MGFTP includes sources, but it is written in BLISS.  AFAIK, BLISSgA is now freeware and is on the Freeware CD, but I suppose learningb it is the obstacle.    -- o John SantosH Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  $ Date: Sat, 3 Feb 2001 19:26:50 -0700& From: "Lee Gillie" <Lee@NOSPAModp.com>' Subject: Re: Seeking FTP SERVER sourcese0 Message-ID: <95iec4$epm$1@personal.myavista.com>   John -  B Hope everything going well at EG&H. I saw the MadGoat version. TheB way new languages are being tossed out there today, I would rather= write FTP from scratch, in something that I know, than try to1> maintain something in a "foreign" (one-of-a-project) language.A When too many syntaxes are stuffed in one ear, others run out thevA other. I'm going to do it in C++ with DecThreads, and provide foru: deriving classes to add functionality to the base service.   Thanks for your reply. - Lee --F ______________________________________________________________________F Lee Gillie, CCP                                Remove NOSPAM to E-MailF Online Data Processing, Inc. - 3501 N. Haven -  Spokane, WA 99207-8500  - "John Santos" <JOHN@egh.com> wrote in message-/ news:1010203190438.3117I-100000@Ives.egh.com... ' > On Sat, 3 Feb 2001, Lee Gillie wrote:e > H > > Looking for a freeware, openware, shareware FTP server with sources.A > > I need basic FTP, and to add extensions via the SITE command. F > > Have C, C++, Basic, Macro32 (no BLISS).  This is for VAX with UCX.I > > I've done this before, for NT, writing from scratch, but hoping thereiG > > will be an existing VMS project I can use to jumpstart this effort. H > > Even to just let me know that you are aware this doesn't exist would	 > > help.t > >t> > > Thanks in advance for any hints.  Best regard - Lee Gillie >i > Hi, Lee -d >oC > MGFTP includes sources, but it is written in BLISS.  AFAIK, BLISS=C > is now freeware and is on the Freeware CD, but I suppose learning= > it is the obstacle.c >k > --
 > John Santoss > Evans Griffiths & Hart, Inc. > 781-861-0670 ext 539 >O   ------------------------------  # Date: Sat, 03 Feb 2001 18:24:04 GMTe8 From: Veli =?iso-8859-1?Q?K=F6rkk=F6?= <korkko@decus.fi>< Subject: Re: Simple DECnet-IV Tasks Impossible under Phase-V( Message-ID: <3A7C17B2.A4E0FF74@decus.fi>  ; Well, this SHOULD BE REALLY EASY afterall. Suppose you have.5 done an upgrade from DECnet IV to this Phase V stuff.7	 Thereforeo9 you have an old NETNODE_REMOTE.DAT in place, typically in1   	SYS$SYSTEM:NETNODE_REMOTE.DAT  3 If you do not have this (e.q. because this is a new8	 install),$< just copy if from another system. Then using DECNET_REGISTER utility    	$ MCR DECNET_REGISTER  : and from menu of options select the option for EXPORT. The firstV; item to specify is the naming system. Specify that you wantI to< export from Phase IV (vs. DECdns and LOCAL which are related to< Phase IV). Specify that you want to export all nodes, i.e. * andV2 specify a name for the export file, e.q. NODES.DAT  < Then the utility creates file NODES.DAT that can be imported
 next, i.e.; from main menu select the item for IMPORT. Specify that youd want6 to import to LOCAL namespace, specify import file name
 NODES.DAT,: error file name NODES.ERR (or whatever) and for the import template< LOCAL:.* plus e.q. register or update for the actual action.  : And that is it. You have migrated all 65000(?) DECnet node name/address9 pairs from your Phase IV database to PhaseV LOCAL system.  (Or even toI DECdns).   _velie   "David J. Dachtera" wrote: >  > "Laurie L. Vandenberg" wrote:7 > >. > > Perhaps using  > > $MCR DECNET_REGISTER > > option 2 > > will work for you. > ' > Tried that - no schmae. Need the NCL.  >  > -- > David J. Dachteray > dba DJE Systemsu > http://www.djesys.com/ > < > Unofficial Affordable OpenVMS Home Page and Message Board:! > http://www.djesys.com/vms/soho/  > H > This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings > is to be expected. > B > Feel free to exercise your rights of free speech and expression. > H > However, attacks against individual posters, or groups of posters, are > strongly discouraged.f   ------------------------------  % Date: Sat, 03 Feb 2001 21:15:47 -0500 , From: "Glenn C. Everhart" <Everhart@GCE.com>% Subject: Re: Source listings contentsg' Message-ID: <3A7C7503.648553E3@GCE.com>   @ There are also things excluded because they are considered trade@ secrets (e.g. the GEM code generator), some things because their> authors don't want them used as examples (considering them tooB ugly for public exposure), some things where patent apps are being@ made, and some where they just don't want to have some info fall= into competitors' hands (not much of this, but I believe some > bits of cluster code are omitted for this reason). I believe a@ few bits are also excluded because they are considered VERY VERY> likely to be redone and they don't want anyone using the calls> mentioned, lest their code suddenly and irretrievably break in a future minor release.   @ There are probably a few more categories, but I know of examples? of all the above. (No, I do not care to be more specific; don't # know as it would be appropriate...)e   Glenn Everhart   Wayne Sewell wrote:e > i > In article <jazs4jZs$EpU@eisner.decus.org>, Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) writes:s\ > > In article <3A735C46.BB21A735@infopuls.com>, Christof Brass <brass@infopuls.com> writes: > >  > >n= > >> From the FAQ VMS7: "OpenVMS VAX and OpenVMS Alpha sourcec% > >> listings CD-ROM sets include themA > >> source listings of most of OpenVMS, ...". I also read in covT& > >> that these sets are not complete. > >>; > >> *Why?* Is anybody out there who possesses the sources?e > >eE > > They exclude material covering pending patents, material coveringe. > > unannounced features and the LMF listings. > D > Ah, yes.  LMF, the license *management* tool that is not a license > *enforcement* tool.l >  > --Q > ===============================================================================-O > Wayne Sewell, Tachyon Software Consulting  (281)812-0738  wayne@tachysoft.xxxa: > http://www.tachysoft.xxx/www/tachyon.html and wayne.htmlM > change .xxx to .com in addresses above, assuming you are not a spambot  :-)iQ > ===============================================================================tD > Cute Girl, to Curly: "Oh, what a beautiful head of bone you have!"   ------------------------------  $ Date: Sat, 3 Feb 2001 17:32:44 -0500  From: John Santos <JOHN@egh.com>O Subject: Re: TCPIP 5.0A FTP Server Rejecting Connections - was RE: Where to gete5 Message-ID: <1010203171818.3117F-100000@Ives.egh.com>b  + On Thu, 1 Feb 2001, Gloria Griffith wrote: X  R > Ok, I have tried all of your great suggestions and I am not making any progress. [...]c  3 > I have checked the protection on the login files.  >  > Directory SYS$COMMON:[SYSMGR]r > " > SYLOGIN.COM;6 (RWED,RWED,RE,RWE) >   E World WRITE access to sylogin.com???  This sounds real bad to me, butDC probably isn't the cause of your problem (unless someone inserted ad "$logout" in it.)h  C What does SYLOGIN.COM do?  If it does anything that won't work from A a network process, such as $ SET TERMINAL commands, it will breaksA FTP.  You need to dispatch on F$MODE and only execute interactiveDB commands for interactive processes, etc.  See SYLOGIN.TEMPLATE for instructions and examples.   > Directory SYS$COMMON:[SYSMGR]e > ! > LOGIN.COM;12 (RWED,RWED,RE,RWE)o  A Irrelevent, unless you are FTP'ing to the SYSTEM account.  If so,e the same provisos apply.  T > The FTP server looks like it ought  to be working but something is wrong because IR > have tried to FTP to this machine many times and there are not any new log filesO > being generated.  The log file I sent you earlier was from Jan 19 when thingsRT > stopped working (of course I do not know what happened that day....this machine isT > used for production and the apps people are always doing something without telling" > me) and there are no more since.  G No new log files?  What is the version number of the top-most log file?-J ;32767?  If so, you need to delete them all or rename them down in versionE numbers (i.e. to keep 20 versions, rename ;32748 to ;1, ;32749 to ;2,2E etc. through ;32767 to ;20.)  The easiest way to do this is to rename$" them all twice, using a wild card:  ( $ RENAME TCPIP$FTP_SERVER.LOG;* TEMP.LOG( $ RENAME TEMP.LOG;* TCPIP$FTP_SERVER.LOG  B (Doing a wild-card rename like this inverts the order of the file F versions, so doing it twice restores the original order, but gets the I version numbers to range from 1 to NN instead of from 32767-NN to 32767.)o   > U > This has got to be something that I am just not seeing. Thank you for your help, ifbA > you want to throw any more advice my way, I will appreciate it!u >  > Gloria Griffitht   Hope this helps.   -- e John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  % Date: Sat, 03 Feb 2001 14:44:31 -0500d2 From: rdeininger@mindspring.com (Robert Deininger)6 Subject: Re: Unloading or deleting licenses in a batchL Message-ID: <rdeininger-0302011444310001@user-2ivebes.dialup.mindspring.com>  p In article <3A7B7B1D.95496C86@capu.net>, "Stephen Eickhoff (remove the - to reply)" <s-eickhoff@capu.net> wrote:  N > Is there any way of deleting and/or unloading expired licenses more than oneN > at a time? On my hobbyist system, I foolishly loaded ALL the layered productM > licenses. Now they've expired, and my poor VAX emits a symphony of beeps oni	 > bootup.r > -- >     I recently threw toghether a COM file to get rid of all expired licenses in a license database. I can email it if anyone is interested.A   The trick is to note that LICENSE LIST has a bunch of qualifiers that the other license commands lack.  I make a listing of interesting licenses, then filter that to make LICENSE DISABLE commands.  Disabled licenses are easy to delete.i  w Though in the present case, it might be best to just start over with a new database, and load it with the new licences.o   -- h Robert Deininger rdeininger@mindspring.com    ------------------------------  % Date: Sat, 03 Feb 2001 15:42:45 -0500gF From: "Stephen Eickhoff (remove the - to reply)" <s-eickhoff@capu.net>6 Subject: Re: Unloading or deleting licenses in a batch( Message-ID: <3A7C6D45.933278E7@capu.net>   Dave wrote:n   > D > License delete * /expire=[old license expiry date here] appears toI > delete only those licenses with that expiry date.  I did that to remove?, > the old licenses from my Hobbyist systems.  lH That is not a valid parameter on my VAX. Maybe you have an Alpha system?   -- A" ----------------------------------          Stephen Eickhoff(           Ardmore, PAh" ----------------------------------   ------------------------------  % Date: Sat, 03 Feb 2001 14:23:52 +0000d& From: Dave <DRichardson@interbaun.com>6 Subject: Re: Unloading or deleting licenses in a batch- Message-ID: <3A7C1478.307EBB18@interbaun.com>u  / Stephen Eickhoff (remove the - to reply) wrote:p > 
 > Dave wrote:o >  > >OF > > License delete * /expire=[old license expiry date here] appears toK > > delete only those licenses with that expiry date.  I did that to removei. > > the old licenses from my Hobbyist systems. > C > That is not a valid parameter on my VAX. Maybe you have an Alpha i	 > system?C  E Yep, 7.2-1, but /expire isn't valid there either.  I've misrememberedo# something.  I probably used either:   : $ License disable * /authorization=decus-can-[my number]-*# $ license delete * /status=disabled    or   $ License delete /all   3 as I have only the hobbyist licenses on the machine5  E Then I ran the .com files (edited one for each node, plus the LP) fort loading the licenses.g   >  > --$ > ---------------------------------- >          Stephen EickhoffS >           Ardmore, PAh$ > ----------------------------------    9 I apologize for the mis-information in my origional post.o  A Perhaps Compaq will extend the LMF to include a "$ License delete G /status=expired"?  This would be usefull to the CSLG and Hobbiest userse* who have a plethoria of term limited PAKs.   Dave Richardson-   ------------------------------  $ Date: Sat, 3 Feb 2001 18:34:44 -0500  From: John Santos <JOHN@egh.com>6 Subject: Re: Unloading or deleting licenses in a batch5 Message-ID: <1010203182514.3117H-100000@Ives.egh.com>e  C On Fri, 2 Feb 2001, Stephen Eickhoff (remove the - to reply) wrote:p  N > Is there any way of deleting and/or unloading expired licenses more than oneN > at a time? On my hobbyist system, I foolishly loaded ALL the layered productM > licenses. Now they've expired, and my poor VAX emits a symphony of beeps onh	 > bootup.o  H Hobbyist licenses (like ASAP/CSA licenses) are delivered in a .COM file.? I made a copy of the .com file and used an ad hoc TECO macro toc> massage all the license commands into license delete commands.? (Keeping the product name and authorization codes, but deletingi everything else.)w  B Of course, you need to use last year's .COM file to delete the old< licenses, as this year's will have new /AUTHORIZATION codes.     -- - John SantosS Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  # Date: Sun, 04 Feb 2001 02:18:14 GMT0( From: Jay Olson <jjo@triton.com.no.spam>4 Subject: Re: Unrecognized Individual Destination PDU1 Message-ID: <3A7CADE5.6050005@triton.com.no.spam>A   Gnter Gratzer wrote:    > Hi!eF > After rebooting our Alpha server we got the following message during
 > startup: > : > %%%%%%%%%%%  OPCOM   3-FEB-2001 14:27:07.32  %%%%%%%%%%%$ > Message from user SYSTEM on DPMS01I > Event: Unrecognized Individual Destination PDU from: Node LOCAL:.DPMS01  > CSMA-CD Station CSMACD-0,s/ >         at: 2001-02-03-14:27:07.322+01:00Iinf : >         Data Link Header='AA0004000284AA0004003C840800'H9 >         eventUid   A1359B79-F9E0-11D4-A64C-44504D533031e9 >         entityUid  D185CB35-F9DB-11D4-8115-AA0004000284x9 >         streamUid  D47D2478-F9DB-11D4-8160-AA0004000284f > L > The message was comming continously. Can anybody explain me the meaning of > this message?  >   3 This message means that the system with MAC addresse4 AA-00-04-02-84 (i.e. DECnet address 33.2) was trying7 to send ethernet packets to the system with MAC addresst8 AA-00-04-03-84 (i.e. DECnet address 33.3) using ethernet7 protocol 0800 (which is TCP, I believe), and that theres9 was no process which was set up to receive these packets. < Simply put, one node was trying to initiate a TCP connection* and your TCP/IP stack was not yet running.  ( 	- Jay Olson (jjo "at" triton "dot" com) 	Triton Software Group LLC   ------------------------------  # Date: Sat, 03 Feb 2001 22:18:39 GMT4 From: Dirk Munk <munk@home.nl>1 Subject: Re: very sloooooow Glasfibre performance ' Message-ID: <3A7C83BC.8415BD68@home.nl>   @ With the help of a Compaq engineer the problem has been located.N The cache memory from both HSG80's was not mirrored. The reliability of todaysR memory is such that I decided that the benefits of more cache memory (512MB versusQ 256MB with mirrored cache) would outweigh the remote change that the cache memorytS would fail. However it seems that the present version of the HSG80 firmware doesn't P like this configuration when the controllers are in multibus failover mode. WithP mirrored cache all problems were over, and we reached the maximum write speed of
 the disks.     Dirk Munk wrote:   > "Main, Kerry" wrote: >b	 > > Dirk,: > >5G > > >>> When we restore a DBMS 7 (codasyl) database, the performance isrP > > miserable. We get about 30 - 40 IO's /second and see a queue length of aboutN > > 5 on the disks. The whole procedure takes appr. 3 - 4 hr.. In contrast theN > > same procedure on another (slower) ES40 takes appr. 30 minutes (to a Mylex > > raid controller). <<<r > >nH > > Fyi, 30-40 IO's per second is a pretty small number and something isH > > definately wrong if you are seeing a queue length of 5 on individual( > > spindles with this number of IO's .. > Q > It is not to 1 spindle, but to a stripe set of 2 mirrorsets (4 spindles total).s >  > >cO > > One thought does come to mind - are you restoring to a newly inited disk orsL > > to (or from?) one that is low in space and perhaps extremely fragmented? > >e< > > Are the same symptoms seen when restoring to a new disk? >S% > The disks are new and almost empty.8 >6 > >0: > > Is the restore from disk to disk or from tape to disk? >2' > From internal SCSI disk to fibredisk.e >O > >d > >> > > Regards, > >i > > Kerry Main > > Senior Consultanti > > Compaq Canada Inc. > > Professional Servicese > > Voice: 613-592-4660G > > Fax  :  819-772-7036  > > Email: Kerry.Main@Compaq.com > >o > > -----Original Message-----) > > From: Dirk Munk [mailto:munk@home.nl]t" > > Sent: January 29, 2001 6:38 PM > > To: Info-VAX@Mvb.Saic.Com-1 > > Subject: very sloooooow Glasfibre performanceu > >iJ > > We are experiencing a very disturbing performance problem with our new > > HSG80 setup. > > H > > We have a ES40 connected to a HSG80 cabinet (by means of a switch of > > course).B > > The ES40, the switch and the HSG80 are not in use by any other > > program/system etc.C > > C > > When we restore a DBMS 7 (codasyl) database, the performance issJ > > miserable. We get about 30 - 40 IO's /second and see a queue length of > > about 5 on the disks. C > > The whole procedure takes appr. 3 - 4 hr.. In contrast the same1I > > procedure on another (slower) ES40 takes appr. 30 minutes (to a MylexT > > raid controller).  > > K > > Of course writeback cache is enabled, and the max. cached transfer size, > > is 128 or higher.S > >TI > > We see the same problem on 2 ES40's, with their own switch and HSG80.c > >,I > > We use VMS 7.2-1 with all applicable patches for fibre and shadowing.a > >i" > > Has anyone seen this before ?? > >t > > Regards, > >h > > Dirk   ------------------------------  # Date: Sat, 03 Feb 2001 19:15:27 GMTt% From: Uwe Zessin <zessin@my-deja.com> 0 Subject: Re: Writes flushed to Disk on Dismount?) Message-ID: <95hlcb$ggd$1@nnrp1.deja.com>n  ( In article <3A7C1CC0.8040302@wi.rr.com>,'   Scott Vieth <svieth@wi.rr.com> wrote:  >n > Paul Repacholi wrote:h > ( > > Think it may be Glen. Consider this. > >-@ > > You tell ops there is a down time for work on the HSGs. TheyD > > dutifully shut down the apps, dismount the shadow set. The write& > > timer is 65000, ie about 20 hours. > >e@ > > Failed circus walks in, is told its shut down and whacks the > > breakers...t  A What is shut down? Access to the disks? The servers? The storage?e: Site management is more complicated with detached storage!   > > Good bye caches. >SG > The caches are protected by batteries.  How long were you thinking ofs, > having the HSG80s cut off from main power? > B > Also, did you do a SHUT OTHER and SHUT THIS before you let FieldE > Service start on your HSG80s? Did your Field Engineer execute thoseV/ > commands before he/she started on the HSG80s?h   I hope they do it these days... . Not all end-users are capable to manage a HSx.  D I was at a customer's site when the service person just shut off theB power, pulled the controller (SWXRC-04 - the multiplatform version@ of the HSZ40 - fortunately that things are gone, too...) und put
 in a new one.   D Of course the new controller was complaining about a bad cache after@ power came back, because he found a foreign signature in memory.  D I've told him to reverse this and shutdown the old controller first.' After that anything was going properly.t  C > If you don't properly follow the operation procedures for the HSx ( > controllers, expect to lose some data.  5 Yes. The manuals say so _several times_!! Read them!!    --
 Uwe Zessin3 (If you want to send mail, please use user "zessin"p/ who lives at "decus.decus.de", not my-deja.com)      Sent via Deja.com  http://www.deja.com/   ------------------------------  $ Date: Sat, 3 Feb 2001 16:23:46 -0500  From: John Santos <JOHN@egh.com>0 Subject: RE: Writes flushed to Disk on Dismount?5 Message-ID: <1010203160700.3117A-200000@Ives.egh.com>s  J   This message is in MIME format.  The first part should be readable text,K   while the remaining parts are likely unreadable without MIME-aware tools.s?   Send mail to mime@docserver.cac.washington.edu for more info.s  ' ------_=_NextPart_001_01C08C77.CB42A234 , Content-Type: TEXT/PLAIN; CHARSET=iso-8859-1. Content-ID: <1010203160700.3117C@Ives.egh.com>  - On Thu, 1 Feb 2001, Bochnik, William J wrote:i  " > I guess what I'm saying is this: >  > D = time of dismount > F = time of failured > S = time of cache flush starta$ > C = time of cache flush completion! > M = time of dismount completionh > 7 > if the cache flush is not forced at time of dismount:g > normal op: > D-M-------------S------C > op with failure: > D-F------    Or more likely:t   D-M---F   F Where you think the dismount has completed successfully, and thereforeG your data has hit the oxide, but it is still in cache for an un-mountede disk when when it fails.  E This is REAL BAD.  In your two scenario's, the failure happens beforedF dismount completion.  (I assume "dismount completion" is the time whenF the dismount program or system service returns a successful completion code to DCL or your program.)n  C Presumably, "DISMOUNT" will return "Fatal I/O failure" or some suchcA error, and you know the disk is broken, but if the failure occurs A minutes or hours later, maybe because you yanked the drive out torC ship it to Peoria, or maybe just because something broke, you don't  know anything went wrong!3   > 3 > if the cache flush is forced at time of dismount:: > normal operation > DS------CM > op with failured
 > DSF----- > N > I both failure situations (whether the cache flushes at time of dismount, orK > at timer) a failure can hit before the cache flush completes - having the L > dismount force the dump decreases the time of vulnerability (especially inF > the perverse case of the timer being set to 65000) but you are stillN > vulnerable (note that the failure in both scenarios hits at the same time onM > the timeline). That's another point that's being overlooked, and the reason=N > why having the second busy disk on the second controller is a bad idea.  I'mN > not sure what can be done to completely bulletproof this.......  Murphy will& > rear  his ugly head sooner or later. >  >  > -----Original Message-----  > From: young_r@eisner.decus.org" > Sent: February 01, 2001 12:22 PM > To: Info-VAX@mvb.saic.com=2 > Subject: Re: Writes flushed to Disk on Dismount? >  > > > In article <87wvba9wqj.fsf@prep.synonet.com>, Paul Repacholi! > <prep@prep.synonet.com> writes:3 >  > > E > > Drive A on C1 is shadowed with drive B on C2. C2 also has drive C0. > > that is having the hell written out of it. > >  > ' > 	Avoid this at all costs.  See below.@ >  > I > > Assumption 1. In the above, the writes to C will force data from C2'sdB > > cache onto the media of B. C1 will hold the data for <timeout> > > seconds. > > @ > > Assumption 2. We have not timed out writes to the shadowset. > > H > > We dismount the shadowset. At this point, we have some set of blocksF > > *x* that are in _a_ cache on a controller. Of this set, the set ofF > > block in C1's Cache *x1* and the blocks in C2's cache *x2* are, byC > > definition - proper subsets of *x*, and *x1* /= *x2*. Hence the C > > blocks on disk are *x*-*x1* and *x*-*x2*, and these will not beT& > > equal. Fail a controller. Or both. > > K > > Now when we bring things back. What do we have? Well, we have differentPD > > disk contents, to a very high probability. We have a significant@ > > chance that the SCB of B, but not A will have gone to media.C > > *BUT, it may not, or may even be the other way around*! Or bothl > > may hit the platter. > > ? > > Case 1 gives B as the 'up to date' drive, and the drives ase > > different.? > > Case 2 gives A as the 'up to date' drive, and the drives ase@ > > different. This is the nasty case, the new master is missing2 > > data that is on B, so the copy will lose data.C > > IF both, or neither ( asuming a previous clean shadowset ) SCBsnD > > go to media, then the shadowset will appear consistant, but willH > > not be. This will result in corruption, inconsistancy, or bugchecks. > > / > > Now, will someone PLEASE shoot this down...t > >  > : > 	Yes.  Be careful what you are doing for the reasons youA > 	outline above.  Note... following applies to certain products.2= > 	RDBMS folks need not worry i.e. archive logs take over andl > shadowsets > 	aren't broken, etc., etc. > : > 		1)  Quiesce/Pause your application.. if you don't, you- > 			have described the scenario where writese > 			don't hit the platter.e >   C I think this is a given, since I think the original description wasoB a dismount of the whole shadow set, not just of a single disk from the set.  D However, I think the scenario applies when you are splitting out one, disk from a shadow set, e.g. to do a backup.    7 > 		2)  Pause AT LEAST CACHE_FLUSH_TIMER seconds BEFORE 7 > 			dismounting a drive to ensure writes do indeed hit  > 			the platter  A Does f$gethsz have an item code that returns CACHE_FLUSH_TIMER?    :-)1   > 		3)  Dismount drive > = > 		4)  unfreeze application(s) to resume writes to remaining  > 			shadowset members >  > F > 	That is why at 10 seconds, (default CACHE_FLUSH_TIMER setting) thisA > 	*normally* isn't an issue.  However, for "performance" reasonseA > 	it is recommended to raise the timer to 65000 seconds.  But ify5 > 	you do that and are using HSZ/HSG .... guess what?  > = > 	That is why this issue ended up out here.  I was wonderingL? > 	what would happen...  we got a good one going this time, eh?r > 	 > 				Robm >    -- e John Santost Evans Griffiths & Hart, Inc. 781-861-0670 ext 539  ' ------_=_NextPart_001_01C08C77.CB42A234 + Content-Type: TEXT/HTML; CHARSET=iso-8859-1 + Content-Transfer-Encoding: QUOTED-PRINTABLEE. Content-ID: <1010203160700.3117D@Ives.egh.com>  2 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
 <HTML>
s <HEAD>
f: <META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
 charset=3Diso-8859-1">
 A <META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
i 5.5.2651.65">
7 <TITLE>RE: Writes flushed to Disk on Dismount?</TITLE>
i </HEAD>
 <BODY>
m 
s: <P><FONT SIZE=3D2>I guess what I'm saying is this:</FONT>
 </P>
0 
10 <P><FONT SIZE=3D2>D =3D time of dismount</FONT>
0 <BR><FONT SIZE=3D2>F =3D time of failure</FONT>
: <BR><FONT SIZE=3D2>S =3D time of cache flush start</FONT>
? <BR><FONT SIZE=3D2>C =3D time of cache flush completion</FONT>
*< <BR><FONT SIZE=3D2>M =3D time of dismount completion</FONT>
 </P>
s 
i@ <P><FONT SIZE=3D2>if the cache flush is not forced at time of =
 dismount:</FONT>
y% <BR><FONT SIZE=3D2>normal op:</FONT>
a3 <BR><FONT SIZE=3D2>D-M-------------S------C</FONT>
 + <BR><FONT SIZE=3D2>op with failure:</FONT>
 $ <BR><FONT SIZE=3D2>D-F------</FONT>
 </P>
h 
-< <P><FONT SIZE=3D2>if the cache flush is forced at time of =
 dismount:</FONT>
0+ <BR><FONT SIZE=3D2>normal operation</FONT>
a% <BR><FONT SIZE=3D2>DS------CM</FONT>
a* <BR><FONT SIZE=3D2>op with failure</FONT>
# <BR><FONT SIZE=3D2>DSF-----</FONT>
m </P>
t 
eI <P><FONT SIZE=3D2>I both failure situations (whether the cache flushes =
 G at time of dismount, or at timer) a failure can hit before the cache =
 J flush completes - having the dismount force the dump decreases the time =
G of vulnerability (especially in the perverse case of the timer being =
aG set to 65000) but you are still vulnerable (note that the failure in =
uH both scenarios hits at the same time on the timeline). That's another =
F point that's being overlooked, and the reason why having the second =
G busy disk on the second controller is a bad idea.&nbsp; I'm not sure =
eF what can be done to completely bulletproof this.......&nbsp; Murphy =
: will rear&nbsp; his ugly head sooner or later.</FONT></P>
 <BR>
o 
n4 <P><FONT SIZE=3D2>-----Original Message-----</FONT>
9 <BR><FONT SIZE=3D2>From: young_r@eisner.decus.org</FONT>
m; <BR><FONT SIZE=3D2>Sent: February 01, 2001 12:22 PM</FONT>
z4 <BR><FONT SIZE=3D2>To: Info-VAX@mvb.saic.com</FONT>
< <BR><FONT SIZE=3D2>Subject: Re: Writes flushed to Disk on =
 Dismount?</FONT>
" </P>
  <BR>
e 
eH <P><FONT SIZE=3D2>In article &lt;87wvba9wqj.fsf@prep.synonet.com&gt;, =
< Paul Repacholi &lt;prep@prep.synonet.com&gt; writes:</FONT>
 </P>
  
, <P><FONT SIZE=3D2>&gt; </FONT>
 H <BR><FONT SIZE=3D2>&gt; Drive A on C1 is shadowed with drive B on C2. =
 C2 also has drive C</FONT>
3A <BR><FONT SIZE=3D2>&gt; that is having the hell written out of =
i it.</FONT>
e  <BR><FONT SIZE=3D2>&gt; </FONT>
 </P>
y 
yJ <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>Avoid this =
& at all costs.&nbsp; See below.</FONT>
 </P>
e <BR>
d 
oJ <P><FONT SIZE=3D2>&gt; Assumption 1. In the above, the writes to C will =
 force data from C2's</FONT>
F <BR><FONT SIZE=3D2>&gt; cache onto the media of B. C1 will hold the =
  data for &lt;timeout&gt;</FONT>
( <BR><FONT SIZE=3D2>&gt; seconds.</FONT>
  <BR><FONT SIZE=3D2>&gt; </FONT>
H <BR><FONT SIZE=3D2>&gt; Assumption 2. We have not timed out writes to =
 the shadowset.</FONT>
  <BR><FONT SIZE=3D2>&gt; </FONT>
G <BR><FONT SIZE=3D2>&gt; We dismount the shadowset. At this point, we =
e have some set of blocks</FONT>
-H <BR><FONT SIZE=3D2>&gt; *x* that are in _a_ cache on a controller. Of =
 this set, the set of</FONT>
J <BR><FONT SIZE=3D2>&gt; block in C1's Cache *x1* and the blocks in C2's =
 cache *x2* are, by</FONT>
G <BR><FONT SIZE=3D2>&gt; definition - proper subsets of *x*, and *x1* =
n /=3D *x2*. Hence the</FONT>
H <BR><FONT SIZE=3D2>&gt; blocks on disk are *x*-*x1* and *x*-*x2*, and =
 these will not be</FONT>
 B <BR><FONT SIZE=3D2>&gt; equal. Fail a controller. Or both.</FONT>
  <BR><FONT SIZE=3D2>&gt; </FONT>
J <BR><FONT SIZE=3D2>&gt; Now when we bring things back. What do we have? =
 Well, we have different</FONT>
 H <BR><FONT SIZE=3D2>&gt; disk contents, to a very high probability. We =
 have a significant</FONT>
H <BR><FONT SIZE=3D2>&gt; chance that the SCB of B, but not A will have =
 gone to media.</FONT>
I <BR><FONT SIZE=3D2>&gt; *BUT, it may not, or may even be the other way =
0 around*! Or both</FONT>
4 <BR><FONT SIZE=3D2>&gt; may hit the platter.</FONT>
  <BR><FONT SIZE=3D2>&gt; </FONT>
H <BR><FONT SIZE=3D2>&gt; Case 1 gives B as the 'up to date' drive, and =
 the drives as</FONT>
a* <BR><FONT SIZE=3D2>&gt; different.</FONT>
H <BR><FONT SIZE=3D2>&gt; Case 2 gives A as the 'up to date' drive, and =
 the drives as</FONT>
wE <BR><FONT SIZE=3D2>&gt; different. This is the nasty case, the new =
u master is missing</FONT>
 C <BR><FONT SIZE=3D2>&gt; data that is on B, so the copy will lose =
o
 data.</FONT>
 I <BR><FONT SIZE=3D2>&gt; IF both, or neither ( asuming a previous clean =
o shadowset ) SCBs</FONT>
F <BR><FONT SIZE=3D2>&gt; go to media, then the shadowset will appear =
 consistant, but will</FONT>
B <BR><FONT SIZE=3D2>&gt; not be. This will result in corruption, =
$ inconsistancy, or bugchecks.</FONT>
  <BR><FONT SIZE=3D2>&gt; </FONT>
> <BR><FONT SIZE=3D2>&gt; Now, will someone PLEASE shoot this =
 down...</FONT>
i  <BR><FONT SIZE=3D2>&gt; </FONT>
 </P>
i 
 J <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>Yes.&nbsp; =
9 Be careful what you are doing for the reasons you</FONT>
tH <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>outline =
C above.&nbsp; Note... following applies to certain products.</FONT>
tF <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>RDBMS =
G folks need not worry i.e. archive logs take over and shadowsets</FONT>
 G <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>aren't =
S broken, etc., etc.</FONT>
 </P>
n 
i0 <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
E &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>1)&nbsp; =
d: Quiesce/Pause your application.. if you don't, you</FONT>
1 <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
(- &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
 A &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>have =
 + described the scenario where writes</FONT>
q1 <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
h- &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
dJ &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>don't hit the =
 platter.</FONT>
 </P>
m 
 0 <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
E &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>2)&nbsp; =
e7 Pause AT LEAST CACHE_FLUSH_TIMER seconds BEFORE</FONT>
o1 <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
p- &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
 J &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>dismounting a =
, drive to ensure writes do indeed hit</FONT>
1 <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
 - &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
4@ &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>the =
 platter</FONT>
i </P>
e 
g0 <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
E &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>3)&nbsp; =
  Dismount drive</FONT>
 </P>
i 
 0 <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
E &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>4)&nbsp; =
 = unfreeze application(s) to resume writes to remaining</FONT>
i1 <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
t- &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
aF &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>shadowset =
 members</FONT>
s </P>
i <BR>
. 
 G <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>That is =
tC why at 10 seconds, (default CACHE_FLUSH_TIMER setting) this</FONT>
t7 <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT =
a9 SIZE=3D2>*normally* isn't an issue.&nbsp; However, for =
 ' &quot;performance&quot; reasons</FONT>
 F <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>it is =
E recommended to raise the timer to 65000 seconds.&nbsp; But if</FONT>
 G <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>you do =
i3 that and are using HSZ/HSG .... guess what?</FONT>
  </P>
a 
,G <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>That is =
0? why this issue ended up out here.&nbsp; I was wondering</FONT>
:E <BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>what =
pD would happen...&nbsp; we got a good one going this time, eh?</FONT>
 </P>
t 
i0 <P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
- &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
-- &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
kE &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT SIZE=3D2>Rob</FONT>
o </P>
T 
w </BODY>
 </HTML>6) ------_=_NextPart_001_01C08C77.CB42A234--c   ------------------------------  $ Date: Sat, 3 Feb 2001 16:38:30 -0500  From: John Santos <JOHN@egh.com>0 Subject: RE: Writes flushed to Disk on Dismount?5 Message-ID: <1010203162846.3117E-100000@Ives.egh.com>b   On 2 Feb 2001, Rob Young wrote:o   >  > 	Scenario....o > ? > 	You wish to make things fast.  You hear through grapevine oro: > 	CETS sessions that setting CACHE_FLUSH_TIMER to 65000+ B > 	"increases performance."  So blindingly, you make that setting.H > 	In your environment, you quiesce your database product/app , dismountA > 	a contributing shadowset, back up that shadowset and return itm$ > 	to be resynched.  The problem is: > 2 > 			1)  On dismount, the HSG does not flush cache0 > 				writes to disk.  Conversely, the HSJ does. > : > 			2)  Because writes aren't flushed, there are a number. > 				that aren't on that shadowset member you > 				just dismounted.  J Shouldn't matter, if there are no failures!  When BACKUP reads the removedE member, it will get data that hasn't hit the disk yet from CACHE, notnJ from the disk.  Remember *HOST*-based volume shadowing...  The HSG doesn'tH know this disk is part of a shadow set.  It should be keeping its cachesG on a per-drive basis, and satisfying reads for cached, unwritten blockse" from the cache, not from the disk.  5 > 			3)  Your backups aren't worth plug nickel... bute. > 				you are still okay if your service level/ > 				agreement contains language to the effect-% > 				"backups are write-only"  (HA!)   G Your backups are fine, assuming the cache (or HSG) doesn't fail betweene: the time you dismounted and the time you write the backup.  @ What worries me is if you dismount the disk, then pull it out to@ transport it without waiting for the cache to flush, or power it down for some reason.I  E (I've seen dead batteries in HSx's and UPS that interrupted.  All your! need is both at the same time...)1  = > 	Hence this whole discussion as the subject line ties into.o > 	 > 				Rob2 >  >  > In article <4B279CF3578CD211B945009027178017046D039E@swilnts809.wil.fusa.com>, "Everhart, Glenn (FUSA)" <GlennEverhart@FirstUSA.com> writes:: > > The disk driver will see everything thru the HSG. Thus; > > there is no problem on shadow recovery or anywhere elseo; > > as long as the HSG either guarantees cache gets to diskt; > > before power can fail (with the aid I presume of a ups)r= > > or saves it in non-volatile memory. I've heard of batteryhA > > backup being used for this; can't say about HSG specifically.t > > - > > Shadow recovery is, repeat, not an issue.u > > = > > Remember what the cpu sees as a disk might be a partitiono> > > of a disk or some other controller construct. I expect you= > > do have to tell folks not to pull drives out of the HSG'sp! > > set without telling it first.s > >  > > -----Original Message-----D > > From: young_r@eisner.decus.org [mailto:young_r@eisner.decus.org]+ > > Sent: Friday, February 02, 2001 9:56 AMk > > To: Info-VAX@Mvb.Saic.Como4 > > Subject: RE: Writes flushed to Disk on Dismount? > >  > > M > > In article <RNGYhGR0S$k5@eisner.decus.org>, young_r@eisner.decus.org (Rob  > > Young) writes:K > >> In article <A8854B7F33E5D3119FDA00508B6A8C6334B1D0@ASX235.asx.com.au>,o2 > > Malcolm Wade <Malcolm.Wade@asx.com.au> writes:! > >>>> -----Original Message----- G > >>>> From: young_r@eisner.decus.org [mailto:young_r@eisner.decus.org] . > >>>> Sent: Friday, February 02, 2001 6:40 AM  > >>>> To: Info-VAX@Mvb.Saic.Com7 > >>>> Subject: Re: Writes flushed to Disk on Dismount?  > >>>> c > >>>  > >>> stuff deleted  > >>>  > >>>> e > >>>> 	One wrinkle to help ...h > >>>>  F > >>>> 	Remember to do a dismount, file streams are *closed* therefore7 > >>>> 	at the time of the dismount that drive is idle.  > >>>> e2 > >>>> 	Also, flush timer is 10 seconds by default > >>>> m
 > >>>> 	So...a > >>>> d > >>>> 	$ wait 00:00:10r! > >>>> 	$ dismount hsg80a_drive1:r > >>>> yG > >>>> 	This doesn't meet all needs but prevents the scripted blunders.nH > >>>> 	Also , during shutdown many seconds transpire prior to dismount, > >>>> 	etc. > >>> I > >>> Surely it doesn't matter about the dismount?  If the drive has beenjO > >>> dismounted, at the HSG controller level it's gone idle so after the cache K > >>> flush timer value expires the cache is flushed.  If you re-mount that 	 > > driveaO > >>> (shadow set member or otherwise) before that time you'll get cached data?w > >>>  > >>  D > >> 	Supposedly... if you have mirrored cache across dual-redundantA > >> 	pairs, yes.  But if you are running without mirrored cache,>? > >> 	does anyone know for sure in dual-redundant mode that thet? > >> 	controller that last had the drive mounted ALWAYS returnss4 > >> 	to that controller (with NO PREFERRED PATH)??? > >> a > > B > > 	Haven't read ahead to see if someone added this wrinkle.. theC > > 	main reason a Dismount NOT flushing writes is a BIG problem isuC > > 	shadowset consistency.  Typically, you dismount the shadowset,*D > > 	back it up and return it to be resynched.  If all the writes in> > > 	progress aren't on that disk... you might have a problem > > > 	on restore!  But that is why when anyone asks if I testedC > > 	the backups, I reply:  "Not really... we do write-only backups ; > > 	here."   Some folks get it and laugh ... others don't.m > >  > > 				Robr >  >    -- r John Santosa Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  " Date: Sat, 3 Feb 2001 23:13:04 GMT6 From: moroney@world.std.spaamtrap.com (Michael Moroney0 Subject: Re: Writes flushed to Disk on Dismount?& Message-ID: <G87F5t.AwA@world.std.com>  < "Koska, John C. (LNG-MBC)" <John.C.Koska@bender.com> writes:  I For whatever it's worth, I looked through the DKDRIVER source to see whatoG is done on a dismount (IO$_AVAILABLE disk command).  Nothing is done asaE far as SCSI commands to the disk other than some synchronization done.G for shared SCSI bus systems, and a STOP UNIT if the command is actually K IO$_UNLOAD not IO$_AVAILABLE, and the disk is removable.  I assume DGDRIVERf is similar.    -Mike.   ------------------------------  % Date: Sat, 03 Feb 2001 15:37:02 -0500 : From: "Koska, John C. (LNG-MBC)" <John.C.Koska@bender.com>0 Subject: RE: Writes flushed to Disk on Dismount?K Message-ID: <3D35AD137AAAD411A6BA0008C7B1B12D6DBDA3@MBCALBEXC03.BENDER.COM>e   > -----Original Message-----5 > From: Paul Repacholi [mailto:prep@prep.synonet.com]s+ > Sent: Saturday, February 03, 2001 8:09 AMa > To: Info-VAX@mvb.saic.comt2 > Subject: Re: Writes flushed to Disk on Dismount? > > > "Koska, John C. (LNG-MBC)" <John.C.Koska@bender.com> writes: > @ > > I have been trying to follow this thread, since it is great  > interest to me.t! > > Please excuse as I jump in...C > > < > > If an Operations or electrician guy drops power to your  > HGS80 and disk@ > > arrays and you have cache batteries, are you not ok as long  > as the cache? > > batteries have power?  Which is quite a few hours (over 20 m > hours as I recall > > > from my site's requirement of having the systems down for  > over 24 hours), to > > keep the cache intact. > G > So, how much are those cute lil suckers worth now. You will take REAL  > good care of them I hope.d  n: They may not be worth as much as when we first got'em, but: that tends to be the nature of computer hardware overtime.  > I believe we spent a little over $100K US dollars for a dozen C of them and another $13K US dollars for the 256 meg cache upgrades.o  ; They are all on a service contract for 2 hour response over3; 7 by 24 by 365.  So we get fairly good care over time, evenDA if they are not the same HSG80s we started out with; not that we -> have had a lot of failures outside of our little power fiasco.   :) jck   ------------------------------  % Date: Sat, 03 Feb 2001 21:20:19 -0500R, From: "Glenn C. Everhart" <Everhart@GCE.com>0 Subject: Re: Writes flushed to Disk on Dismount?' Message-ID: <3A7C7613.70E611E0@GCE.com>>  > The designers of these boxes know very well that if they cache9 any blocks, they must maintain cache consistency if there2> are paths to a piece of storage from more than one controller.C This is done via separate paths in some designs I saw presentationst on. A non issue...    LBohan@dbc.spam_less..com wrote: > D > On 2 Feb 2001 11:13:02 -0500, young_r@eisner.decus.org (Rob Young) > wrote: > < > >Hate when this happens... but yes... the writes are still9 > >in the cache so when you remount it you are okay.  TheF6 > >only wrinkle in this whole thing then , what if you? > >aren't running mirrored-cache (many of us don't as shadowseti- > >members are split across controller pairs)r > 2 > I thought running mirrored cache was an absolute' > requirement for the HSG...  It's not?
   ------------------------------    Date: 04 Feb 2001 10:30:54 +0800, From: Paul Repacholi <prep@prep.synonet.com>0 Subject: Re: Writes flushed to Disk on Dismount?- Message-ID: <87r91ffach.fsf@prep.synonet.com>e  " John Santos <JOHN@egh.com> writes:  G > (I've seen dead batteries in HSx's and UPS that interrupted.  All you # > need is both at the same time...)    Or a HSG bugcheck?   -- (< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.i@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------    Date: 04 Feb 2001 10:35:50 +0800, From: Paul Repacholi <prep@prep.synonet.com>0 Subject: Re: Writes flushed to Disk on Dismount?- Message-ID: <87n1c3fa49.fsf@prep.synonet.com>   < "Koska, John C. (LNG-MBC)" <John.C.Koska@bender.com> writes:  I > > So, how much are those cute lil suckers worth now. You will take REALT > > good care of them I hope.
 >  >< > They may not be worth as much as when we first got'em, but< > that tends to be the nature of computer hardware overtime. > @ > I believe we spent a little over $100K US dollars for a dozen E > of them and another $13K US dollars for the 256 meg cache upgrades.g  F No, no, not the HSG, the batteries? $10? $20? ( other than from a well know vendor that is. )  < So that's how many dollars for a failure per dollar or NiCd?  L I just can NOt believe, after all the 'we go to all this.... quallification'C stuff ( yes guys who do it, I know you do ) thats why it is SO muchd better and cost more.&  F Then they dump a load of crap like this. BTW, and one for a sweepstake! on a new version of HSG firmware?o   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.t@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  $ Date: Sat, 3 Feb 2001 12:07:55 -0700& From: "Lee Gillie" <Lee@NOSPAModp.com>? Subject: Re: [Q] How to Highlight a text in fortran 77 on VMS ? 0 Message-ID: <95hkl6$7c8$1@personal.myavista.com>   Julien  D I've not done Fortran for years, but if I were starting on a project@ in VMS, and wanted broad support for a variety of terminals, andE planned to do cursor positioning, and use textual display attributes,&E I would certainly give SMG$ routines a first consideration.  That waytA you don't have to manipulate escape codes, and really should not, C unless necessary. And like LIB$ routines, comes stock on just about  every flavor of VMS.  , From the VMS/DCL prompt, enter this command:     help RTL_Routines SMG$  F Here is the complete documentation for SMG, which includes an overview? to get you started.  Unfortunately it is not dense with Fortranh	 examples:l@     http://www.openvms.compaq.com:8000/72final/5935/5935pro.html  # Have fun - best regards, Lee Gillieh --F ______________________________________________________________________F Lee Gillie, CCP                                Remove NOSPAM to E-MailF Online Data Processing, Inc. - 3501 N. Haven -  Spokane, WA 99207-8500  7 "TheBigfoot" <thebigfoot@softhome.net> wrote in message'/ news:tKWe6.3026$Aq4.5994543@nnrp5.proxad.net... K > Everything is in the text, so I wonder how to highlight a text in fortranh 77L > ... Someone gave me recently a way to insert a text anywhere on the screenL > thanks to LIB$SET_CURSOR(x,y) and I'd like to use a similar command ... atH > school I couldn't find the RTL LIB$CURSOR so of course the routine I'm > looking for was missing too.K > I'm not really good at manipulating escape codes so if someone could giveB me& > an example I would be very thankful. >d >     Thanks already, Julien >  > -- >/" >                          _\\|//_" >                          ( . . )( > -----------------------ooO-(_)-Ooo---- >>& > TheBifgoot <thebigfoot@softhome.net> > http://www.thebigfoot.fr.st  >A >s   ------------------------------  % Date: Sat, 03 Feb 2001 14:23:42 -0500 , From: Howard S Shubs <hshubs@mindspring.com>? Subject: Re: [Q] How to Highlight a text in fortran 77 on VMS ? > Message-ID: <hshubs-DE2760.14234203022001@news.mindspring.com>  C In article <tKWe6.3026$Aq4.5994543@nnrp5.proxad.net>, "TheBigfoot" i  <thebigfoot@softhome.net> wrote:  M >I'm not really good at manipulating escape codes so if someone could give med% >an example I would be very thankful.a  H http://vt100.net/docs/vt220-rm/    ...for the VT programming informationA http://www.openvms.compaq.com:8000/index.html   for the VMS docs.b  '    subroutine put_cursor (line, column)  cs' c  Position the cursor at LINE, COLUMN.O cS c  Arguments/ c  LINE: I*4 r/o    line to place the cursor ono1 c  COLUMN: I*4 r/o  column to place the cursor on; cs    implicit none   cs c  Arguments cs    integer*4 line     integer*4 columns   cs c  Functions cOG    integer*4 sys$fao             !Formatted Ascii Output system servicen   cn c  Local variables cn9    integer*2 output_length       !length of output string=7    integer*4 status              !returned status value=2    character*1 esc               !ESCape character@    character*40 output           !arbitrary length output string   cs
 c  Initializen c;A    esc = char(27)                !initialize the escape charactern   cFL c  Format the output string.  I don't like or particularly trust FORMAT moreN c  than I have to.  This avoids messy details about the length of the numbers  c  involved. Details are at
K c  http://www.openvms.compaq.com:8000/72final/4527/4527pro_035.html#jun_245n c;I c  The cursor position sequence used here is documented in section 4.7 ons9 c  page http://vt100.net/docs/vt220-rm/chapter4.html#S4.7t c<;    status = sys$fao ('!AS[!UL;!ULH', output_length, output,&$    1  esc, %val(line), %val(column))J    if (.not.status) call exit (status) !always check returned status value  	    return&    end  F note: I haven't actually compiled this, so there may be typos or such. -- T Howard S ShubsD "Run in circles, scream and shout!"  "I hope you have good backups!"   ------------------------------  % Date: Sat, 03 Feb 2001 14:57:16 -0500s2 From: rdeininger@mindspring.com (Robert Deininger)? Subject: Re: [Q] How to Highlight a text in fortran 77 on VMS ?/L Message-ID: <rdeininger-0302011457170001@user-2ivebes.dialup.mindspring.com>  c In article <tKWe6.3026$Aq4.5994543@nnrp5.proxad.net>, "TheBigfoot" <thebigfoot@softhome.net> wrote:s  N > Everything is in the text, so I wonder how to highlight a text in fortran 77L > ... Someone gave me recently a way to insert a text anywhere on the screenL > thanks to LIB$SET_CURSOR(x,y) and I'd like to use a similar command ... atH > school I couldn't find the RTL LIB$CURSOR so of course the routine I'm > looking for was missing too.N > I'm not really good at manipulating escape codes so if someone could give me& > an example I would be very thankful. >  >     Thanks already, Julien   It appears you should stop fiddling with escape codes and stuff, and just use the Screen Management routines.  Look up the SMG$ section in the Run-Time Library documentation.  Examples are included, IIRC.  c As your screen manipulations get more complex, SMG is more and more worth the small learning curve.;   -- n Robert Deininger rdeininger@mindspring.com    ------------------------------  # Date: Sat, 03 Feb 2001 20:19:21 GMT/, From: "TheBigfoot" <thebigfoot@softhome.net>? Subject: Re: [Q] How to Highlight a text in fortran 77 on VMS ?a2 Message-ID: <dJZe6.246$7o.959573@nnrp1.proxad.net>  H Thank you everybody for your help, it warms one's heart to see that some& people are willing to help the others.   Best regards, Julien   --                             _\\|//_                           ( . . )& -----------------------ooO-(_)-Ooo----  $ TheBifgoot <thebigfoot@softhome.net> http://www.thebigfoot.fr.ste1 "Lee Gillie" <Lee@NOSPAModp.com> wrote in messagen* news:95hkl6$7c8$1@personal.myavista.com... > Julien >dF > I've not done Fortran for years, but if I were starting on a projectB > in VMS, and wanted broad support for a variety of terminals, andG > planned to do cursor positioning, and use textual display attributes,lG > I would certainly give SMG$ routines a first consideration.  That way C > you don't have to manipulate escape codes, and really should not,sE > unless necessary. And like LIB$ routines, comes stock on just abouta > every flavor of VMS. >w. > From the VMS/DCL prompt, enter this command: >     help RTL_Routines SMG$ >bH > Here is the complete documentation for SMG, which includes an overviewA > to get you started.  Unfortunately it is not dense with Fortrane > examples:aB >     http://www.openvms.compaq.com:8000/72final/5935/5935pro.html >t% > Have fun - best regards, Lee Gilliet > --H > ______________________________________________________________________H > Lee Gillie, CCP                                Remove NOSPAM to E-MailH > Online Data Processing, Inc. - 3501 N. Haven -  Spokane, WA 99207-8500 > 9 > "TheBigfoot" <thebigfoot@softhome.net> wrote in messagee1 > news:tKWe6.3026$Aq4.5994543@nnrp5.proxad.net...dE > > Everything is in the text, so I wonder how to highlight a text in  fortran  > 77G > > ... Someone gave me recently a way to insert a text anywhere on thet screenK > > thanks to LIB$SET_CURSOR(x,y) and I'd like to use a similar command ...  atJ > > school I couldn't find the RTL LIB$CURSOR so of course the routine I'm  > > looking for was missing too.H > > I'm not really good at manipulating escape codes so if someone could give > me( > > an example I would be very thankful. > >h > >     Thanks already, Julien > >n > > -- > >s$ > >                          _\\|//_$ > >                          ( . . )* > > -----------------------ooO-(_)-Ooo---- > >t( > > TheBifgoot <thebigfoot@softhome.net> > > http://www.thebigfoot.fr.ste > >d > >a >r >  >    ------------------------------   Date: 3 Feb 2001 22:03:30 GMT 7 From: Thomas.Hahnemann@nospam_s-t.de (Thomas Hahnemann) ? Subject: Re: [Q] How to Highlight a text in fortran 77 on VMS ? 0 Message-ID: <Oozvf8elmJpy-pn2-ldN8f3H10BTJ@Tom2>  = Very good are the Screen Management Services( SMG routines ). * You can open virtual displays ("windows"),( read single chars from keybord and more.  7 If you really want to to it in FORTRAN here is a simple. example:  #         subroutine ANSI$GotoXY(x,y)k#         character       CTR$CSI*(*) 0         parameter       (CTR$CSI= char(27)//'[')!         integer         x,y,ix,iy0!         character       sx*3,sy*3n 1       format(a,$)          ix = max(1,min(80,x))l         iy = max(1,min(25,y))8         write(sx,'(i3.3)') ix          write(sy,'(i3.3)') iya9         write(*,1) ' '//CTR$CSI//sy//';'//sx//'H'       !>         endg             program test         print'(a,$)',' '"         call ANSI$GotoXY( 35, 10 )!         print'(a)','+Hello World'@         end      Thomas Hahnemann   ------------------------------   End of INFO-VAX 2001.069 ************************