1 INFO-VAX	Sun, 10 Dec 2000	Volume 2000 : Issue 689       Contents:' Re: (Change topic) Fast skipping to EOT ' Re: (Change topic) Fast skipping to EOT , Re: ??== Allocating only one block per file., Re: ??== Allocating only one block per file., Re: ??== Allocating only one block per file. Re: Automated RMS tuning script  Re: Cable modem woes update... Re: Cable modem woes update... Re: Cable modem woes update...  DEC Basic: Can not run debugger.$ Re: DEC Basic: Can not run debugger. Kernel Overhead for Direct I/O" Re: Kernel Overhead for Direct I/O" Re: Kernel Overhead for Direct I/O Re: Very weird DCL behavior   F ----------------------------------------------------------------------  % Date: Sun, 10 Dec 2000 09:17:46 +0000  From: Roy Omond <Roy@Omond.net> 0 Subject: Re: (Change topic) Fast skipping to EOT) Message-ID: <3A334A3B.D7048455@Omond.net>    Paul Sture wrote:    > 7 > You mean the following text? Here on Alpha VMS 7.2-1:  >  > SET  >  >   MAGTAPE  >  >     /FAST_SKIP >  >           /FAST_SKIP=option  > 6 >        Allows you to skip by file mark or by record.  A Hmmm... I must be missing an ECO.  It's not on my system (7.2-1).   E > > Thankfully I am no longer at that particular customer site (those > > > who know me within Compaq will know where it was ... don't > > mention the name M*** !).  > > 	 > Slough?    Nope !  	 Roy Omond  Blue Bubble Ltd.   ------------------------------  % Date: Sun, 10 Dec 2000 11:05:04 +0100   From: Paul Sture <paul@sture.ch>0 Subject: Re: (Change topic) Fast skipping to EOT+ Message-ID: <VA.000001cb.02a74997@sture.ch>   : In article <3A334A3B.D7048455@Omond.net>, Roy Omond wrote:! > From: Roy Omond <Roy@Omond.net>  > Newsgroups: comp.os.vms 2 > Subject: Re: (Change topic) Fast skipping to EOT' > Date: Sun, 10 Dec 2000 09:17:46 +0000  >  > Paul Sture wrote:  >  > > 9 > > You mean the following text? Here on Alpha VMS 7.2-1:  > >  > > SET  > > 
 > >   MAGTAPE  > >  > >     /FAST_SKIP > >  > >           /FAST_SKIP=option  > > 8 > >        Allows you to skip by file mark or by record. > C > Hmmm... I must be missing an ECO.  It's not on my system (7.2-1).  > H A brief scan shows that the only ECOs I have don't seem to mention help H corrections. It might be worth mentioning that this was an install from 6 scratch, rather than an upgrade of a previous version.  G And I can confirm it's there on VAX/VMS V7.2 too (which was an upgrade   from V6.1).   G > > > Thankfully I am no longer at that particular customer site (those @ > > > who know me within Compaq will know where it was ... don't > > > mention the name M*** !).  > > >  > > Slough?  >  > Nope ! >  'Twas but a wild stab :-)  ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sun, 10 Dec 2000 07:44:13 -0500 , From: Howard S Shubs <hshubs@mindspring.com>5 Subject: Re: ??== Allocating only one block per file. > Message-ID: <hshubs-9861AF.07441310122000@news.mindspring.com>  B In article <3A32E820.3E6A736A@earthlink.net>, "David J. Dachtera" $ <djesys.nospam@earthlink.net> wrote:  C >The code I mentioned uses LIB$FIND_FILE and LIB$FIND_FILE_END - no H >LIB$SPAWN, no LIB$DO_DCL or anything else. I'm not aware of any similarD >circumstance that would require any kind of (LIB$)SPAWN, although I4 >could see where that might be useful or beneficial.  D Doing a DIRECTORY command would have required a LIB$SPAWN().  Using N LIB$DO_COMMAND() wouldn't have worked if you wanted to continue processing in N your own program, and the rest goes along with what I was saying about trying  to avoid LIB$SPAWN().  --   Howard S ShubsD "Run in circles, scream and shout!"  "I hope you have good backups!"   ------------------------------    Date: 11 Dec 2000 01:08:02 +0800, From: Paul Repacholi <prep@prep.synonet.com>5 Subject: Re: ??== Allocating only one block per file. 0 Message-ID: <87lmtotdal.fsf@k9.prep.synonet.com>  . Howard S Shubs <hshubs@mindspring.com> writes:  F > Doing a DIRECTORY command would have required a LIB$SPAWN().  Using   = Ah well you could be totally outrageuse and do a merged image > activation of directory :| Yes, I know, but it needs no spawn.   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------    Date: 11 Dec 2000 01:12:03 +0800, From: Paul Repacholi <prep@prep.synonet.com>5 Subject: Re: ??== Allocating only one block per file. 0 Message-ID: <87itostd3w.fsf@k9.prep.synonet.com>  . Howard S Shubs <hshubs@mindspring.com> writes:  G > In article <aus-0912001702370001@wvia26.virologie.uni-wuerzburg.de>,  / > aus@vim.uni-wuerzburg.de (Hans M. Aus) wrote:  > M > >My current thoughts are to split the information retrieval into two parts.  > > L > >1) Sometime after midnight, harvest all the daily files, from the last 24- > >hours, into one big indexed, annual file.    9 Beware of the xfer over 'the time' boundary condition. If > you are not carefull that file could appear 1, 0 or 2 times in the records.   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  % Date: Sun, 10 Dec 2000 10:24:47 +0100 2 From: "Thomas H. Pauli" <thomas.pauli@t-online.de>( Subject: Re: Automated RMS tuning script* Message-ID: <3A334BDF.8030301@t-online.de>  
 Dear Stan,  H I still have such a piece running and will send a copy to you on Monday  morning.  
 Greetings,   Thomas   Stan Rose wrote:  I > Many years ago I wrote a DCL script that automated the tuning of an RMS M > file; it optimized bucket size and other parameters. Sometime after I wrote J > the script, someone anonymously posted it to several newsgroups. I don'tN > have it anymore, however, but would like to get a copy. Does anyone remember< > such a DCL script and have a copy that they could send me? >  >  > Thanks very much,  > Stan Rose  >  >  >    ------------------------------  % Date: Sun, 10 Dec 2000 09:37:29 +0100   From: Paul Sture <paul@sture.ch>' Subject: Re: Cable modem woes update... + Message-ID: <VA.000001c9.02571830@sture.ch>   < In article <3A331292.1EC192CE@videotron.ca>, JF Mezei wrote:/ > From: JF Mezei <jfmezei.spamnot@videotron.ca>  > Newsgroups: comp.os.vms ) > Subject: Re: Cable modem woes update... ' > Date: Sun, 10 Dec 2000 00:20:22 -0500  >  > "David J. Dachtera" wrote:L > > On the contrary - having been rather close to the ISP industry this pastK > > summer (I worked for Mark Levy and some of his FSInet customer sites) I H > > became aware that ISPs may have negotiated these "features" with theJ > > manufacturers to help "ensure" their revenue stream, which is based in6 > > part on per IP address (read: per client machine). > M > Well, I can understand wanting to prevent customers from abusing the stuff, O > but that feature is what firces customers to buy routers, at which point they P > will abuse the ISP even more because they will be able to put as many machines' > as they want but pay for a single IP.  > O > If the modem recorded the ether address of only the machines whihc issue DHCP P > requests, then that would be a fair protection against multiple machines using. > the service when only one IP is provisioned. > L > And because of that feature (if that is truly true, it was perhaps just anO > excuse given by the ISP), I have had to monopolise a lot of support resources O > at the ISP  because they are not willing to document it. So the ISP's support I > costs are much higher because of such a silly undocumented restriction, P > especially when the ISP's sales dept continues to tell customers that it is OKM > to have as many machines on your LAN as you want as long as only one issues  > DHCP requests. > O I really don't see, given decent hardware, that it should matter. I have ISDN,  Q but the principle should be the same - my ISDN hub router does the talking to my  M ISP as a _single_ box. It works both with a fixed IP address and dynamically  M assigned addresses issued by the ISP (though probably not both at once). The  P fact that I have 4 systems going into the hub router is irrelevant as far as my J ISP is concerned - the ISP just sees my hub router - 1 address, in effect.  L > You'd think that an ISP with hundreds of thousands of subscribers would doO > everything to reduce support costs by implementing hardware restrictions that , > match its actual policies (or vice-versa). > J > What is also a learning experience is to find out how important it is toP > provide first level support folks with the logs of ALL changes to the network.L > It makes it very hard to debug a problem that props up out of nowhere whenM > from their side, they don't see changes at their end and are thus forced to 0 > blame the problem on the customer's equipment. >    ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sun, 10 Dec 2000 05:11:37 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca> ' Subject: Re: Cable modem woes update... , Message-ID: <3A3356D7.C597BC52@videotron.ca>   Paul Sture wrote:  > The Q > fact that I have 4 systems going into the hub router is irrelevant as far as my L > ISP is concerned - the ISP just sees my hub router - 1 address, in effect.  G My ISP's contract of service stipulates no routers and no servers for a I residential service. It is a ten fold increase in monthly cost to get the  right to have a server.   L Of course, they have no way to know that there is a router, so their rule isI not really enforceable. But I guess that if they wanted to, they probably N could check who the manufacturer of the ethernet device their network sees andF they would see that I have a "NETGEAR" or "CISCO" machine on my net...N However, judging from the type of knowledge their support people have, I doubtK they even know that each manufacturer of ether devices has a range of ether " adresses (does this still apply ?)   ------------------------------  % Date: Sun, 10 Dec 2000 15:56:29 +0100   From: Paul Sture <paul@sture.ch>' Subject: Re: Cable modem woes update... + Message-ID: <VA.000001cd.000f248c@sture.ch>   < In article <3A3356D7.C597BC52@videotron.ca>, JF Mezei wrote:/ > From: JF Mezei <jfmezei.spamnot@videotron.ca>  > Newsgroups: comp.os.vms ) > Subject: Re: Cable modem woes update... ' > Date: Sun, 10 Dec 2000 05:11:37 -0500  >  > Paul Sture wrote:  > > The S > > fact that I have 4 systems going into the hub router is irrelevant as far as my N > > ISP is concerned - the ISP just sees my hub router - 1 address, in effect. > I > My ISP's contract of service stipulates no routers and no servers for a K > residential service. It is a ten fold increase in monthly cost to get the  > right to have a server.  > N Ouch! They are obviously trying to avoid folks hogging bandwidth, as the "no  N servers" implies. What's their position on using a Windows ME system which is H networked to others in the home and can share "one internet connection"?  N > Of course, they have no way to know that there is a router, so their rule isK > not really enforceable. But I guess that if they wanted to, they probably P > could check who the manufacturer of the ethernet device their network sees andH > they would see that I have a "NETGEAR" or "CISCO" machine on my net...  P How would they react to a cheap Linux box acting as a firewall? One system only : talking to the ISP.  Router? What router? That's my PC :-)  P > However, judging from the type of knowledge their support people have, I doubtM > they even know that each manufacturer of ether devices has a range of ether $ > adresses (does this still apply ?) > @ You could always go for a manufacturer who does NIC cards too...  P The real problem is that the ISP clearly wants to avoid someone generating huge R amounts of traffic, but is falling into the trap of thinking that someone who has I multiple systems connectable must be consuming huge amounts of bandwidth.   R Whatever you work out, be careful, as ISPs can just cut you off from the world if ( they think you are abusing the contract.  
 Paul Sture Switzerland    ------------------------------  # Date: Sun, 10 Dec 2000 15:13:03 GMT / From: "Tom Simpson" <simpsont@xxx.mediaone.net> ) Subject: DEC Basic: Can not run debugger. E Message-ID: <34NY5.12678$U6.109915@typhoon.jacksonville.mediaone.net>   F I rebuilt a Alpha DEC 3000 system using VMS 7.2-1 + patches and latestK BASIC compiler release.  I discovered that I can not run the debugger, now.   K No errors during compile and link, but when I try to run the program, I get  the  following message:  - %SYSTEM-F-NODOMAIN, resource domain not found   = I've never seen this before.  Anyone know what this error is?   B It does not matter what program I pick to test.  Even a simple one line program fails.   . I have seen no other problems with the system.   Regards, Tom    ------------------------------  % Date: Sun, 10 Dec 2000 13:14:17 -0500 ( From: Jonas Lindholm <jlindholm@rcn.com>- Subject: Re: DEC Basic: Can not run debugger. ' Message-ID: <3A33C7F9.D26002BF@rcn.com>   ? This indicate that the audit server is not running on the host. % Start the audit server and try again.    /Jonas Lindholm 
 OM Technology      Tom Simpson wrote:  H > I rebuilt a Alpha DEC 3000 system using VMS 7.2-1 + patches and latestM > BASIC compiler release.  I discovered that I can not run the debugger, now.  > M > No errors during compile and link, but when I try to run the program, I get  > the  > following message: > / > %SYSTEM-F-NODOMAIN, resource domain not found  > ? > I've never seen this before.  Anyone know what this error is?  > D > It does not matter what program I pick to test.  Even a simple one > line program fails.  > 0 > I have seen no other problems with the system. > 
 > Regards, > Tom    ------------------------------  % Date: Sun, 10 Dec 2000 10:14:50 -0700 + From: "Barry Treahy, Jr." <treahy@mmaz.com> ' Subject: Kernel Overhead for Direct I/O ( Message-ID: <3A33BA0A.515A2955@mmaz.com>  B I know that Direct I/O consumes kernel stack time, but can someone confirm what is realistic?  F I have a VAX 4000/100 and two weeks ago I yanked all of my old DEC andG third-parts SCSI drives and replaced it with an Infortrends based RAID. F I've got analysis reports running, and I'm seeing on average 182 I/O'sC per second on the hot drive, the I/O queue is less than one but the , kernel time is hanging about 32% on average.  E I know that the new drives would shift the bottle-necks, but I didn't . anticipate this.  Any recommendations on this?   Regards,   Barry    --  ? Barry Treahy, Jr  *  Midwest Microwave  *  Vice President & CIO   A E-mail: Treahy@mmaz.com * Phone: 480/314-1320 * FAX: 480/661-7028    ------------------------------    Date: 10 Dec 2000 12:55:34 -05009 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) + Subject: Re: Kernel Overhead for Direct I/O + Message-ID: <9V4ERGPB$hEq@eisner.decus.org>   V In article <3A33BA0A.515A2955@mmaz.com>, "Barry Treahy, Jr." <treahy@mmaz.com> writes:D > I know that Direct I/O consumes kernel stack time, but can someone > confirm what is realistic? > H > I have a VAX 4000/100 and two weeks ago I yanked all of my old DEC andI > third-parts SCSI drives and replaced it with an Infortrends based RAID. H > I've got analysis reports running, and I'm seeing on average 182 I/O'sE > per second on the hot drive, the I/O queue is less than one but the . > kernel time is hanging about 32% on average. > G > I know that the new drives would shift the bottle-necks, but I didn't 0 > anticipate this.  Any recommendations on this?  G VAX does not have support for FAST_IO that Alpha has to alleviate this.    ------------------------------  % Date: Sun, 10 Dec 2000 11:14:12 -0700 + From: "Barry Treahy, Jr." <treahy@mmaz.com> + Subject: Re: Kernel Overhead for Direct I/O ( Message-ID: <3A33C7F4.A66EF2FB@mmaz.com>  ( Kilgallen@eisner.decus.org.nospam wrote:   > > I > > I know that the new drives would shift the bottle-necks, but I didn't 2 > > anticipate this.  Any recommendations on this? >0I > VAX does not have support for FAST_IO that Alpha has to alleviate this.e  Y Yes, but I cannot move to Alpha; its a legacy system, TDMS in particular, and now that CA=Y and Oracle own the parts to CDD, DBMS, DTR, it isn't an option at this time since its tooo expensive...   Barry      --  ? Barry Treahy, Jr  *  Midwest Microwave  *  Vice President & CIO   A E-mail: Treahy@mmaz.com * Phone: 480/314-1320 * FAX: 480/661-7028:   ------------------------------  # Date: Sun, 10 Dec 2000 10:47:31 GMTa! From: Ian Parker <parker@gol.com>a$ Subject: Re: Very weird DCL behavior& Message-ID: <lOmQaGAx8uM6EwVu@gol.com>  ? In article <o19X5.8166$Cn4.198087@weber.videotron.net>, Syltrem0 <syltrem@videotron.ca> writesn >Hi! >oK >I use a DCL procedure (what else!) to do my nightly backups. I have 4 tape4C >drives, and I backup parts of my data (that could be one disk, onefM >directory, one oracle db - but it doesn`t really matter). The main procedure L >has a list of things to do which it dispatches (via a spawn/nowait command)" >to the next available tape drive. >.F >When one backup is finiished, the spawned procedure notifies the mainJ >process that it has finished so that another job can be sent to that tape >drive (am I clear enough?). > J >Upon reading the mailbox, the main procedure sometimes (once every 100 or/ >200 times it does the READ command) hits error-+ >%RMS-E-RSA, record stream currently active:E >when (or right after) reading the mailbox. I do not know why this ispJ >happening, and the weird thing about this is that the .LOG shows commands* >that do not exist in the .COM file, like: >n >$       Set NoOno. >$       Read Backup_Mbx Mbx_Rec /Time_Out=255< >$       If .not. $(Backup_Status<NL><NL>0,Mbx_Rec)<NL>_Disk+ >%RMS-E-RSA, record stream currently activea >$ Error_Trap: >$       Exit_Status = $STATUS >i >Note:K >a) The 2nd line in this .LOG excerpt does not belong to the .COM. Note the " ><NL> characters in the line, too.J >b) The error message does not follow immediately the READ command, but it >sure is related to itM >c) The SET NOON seems to have no effect as the procedure jumps into my errort >trap. >y' >The .COM contains the following lines:  >$       Set NoOnh. >$       Read Backup_Mbx Mbx_Rec /Time_Out=2553 >$       If .not. $STATUS then Goto Backup_One_Diske >$       Set Ona# >$       Unit = f$extr(0,1,Mbx_Rec)t- >$       Backup_Status = f$extr(1,10,Mbx_Rec)  >hL >It seems like the DLC processor has lost its pointer in reading the command >file. >n >Can someone help? >  >Thanks so much! >  >Syltrem >  >e >*  F I save multiple savesets on a single DLT tape by copying saveset filesH from disk to tape, one by one without rewinding.  Restores are painfullyC slow because I copy the specific saveset file back to disk and thensD restore it (for various reasons I don't restore directly from tape).  7 Would this qualifier speed things up in this situation?>   -- a
 Ian Parker   ------------------------------   End of INFO-VAX 2000.689 ************************