1 INFO-VAX	Wed, 10 Aug 2005	Volume 2005 : Issue 444       Contents:( Re: <title>OpenVMS documentation</title>/ Re: Changing Uppercase filenames into Lowercase  DECnet-plus problem 4 Re: HEAT RELATED PROBLEMS; POWER OFF/ON, SUGGESTIONS* HP works to bring Linux to NonStop servers. Re: HP works to bring Linux to NonStop servers. RE: HP works to bring Linux to NonStop servers. Re: HP works to bring Linux to NonStop servers. Re: HP works to bring Linux to NonStop servers. Re: HP works to bring Linux to NonStop servers. Re: HP works to bring Linux to NonStop servers? Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine ? Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine ? Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine ? Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine ? Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine ? Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine  Re: Is OVMS going Bangalore? Re: Is OVMS going Bangalore? Re: Is OVMS going Bangalore?7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 7 Re: Killing a process that has allocated the tape drive 0 Re: Looking for template software for VMS system( Re: looking for TIMES-ROMAN.XDPS$OUTLINEP Re: OpenVMS under the umbrella - or umbrella is supported at the time? time?time+ Re: strange terminal-characteristic problem + Re: strange terminal-characteristic problem + Re: strange terminal-characteristic problem  Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  RE: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN  Re: Telnet over VPN   F ----------------------------------------------------------------------  # Date: Wed, 10 Aug 2005 11:22:45 GMT 5 From: rdeininger@mindspringdot.com (Robert Deininger) 1 Subject: Re: <title>OpenVMS documentation</title> L Message-ID: <rdeininger-1008050722430001@user-uinj4fq.dialup.mindspring.com>  C In article <05080921311013_20200298@antinode.org>, sms@antinode.org  (Steven M. Schweda) wrote:  A >   Am I the only one who is consistently annoyed by the VMS HTML I >documentation in which every page of every manual seems to have the same B >title?  Opening several manuals leaves one's desktop (or browser)A >cluttered with several icons (or tabs), each of which cheerfully I >identifies itself as "OpenVMS documentation", leaving the user who has a D >short-term memory deficit pretty much in the dark about where Waldo >really is.  > B >   I'd say that this annoyance is second only to the _absolutely_& >_screaming_ USELESS search "feature". > 0 >   But at least I'm not paying for maintenance.    G There is an email address provided to contact the documentation folks.  J They do read the email, and they act on the suggestions when they can find a practical way to do so.   C They don't read this newsgroup, AFAIK.  Since there is a documented ? feedback mechanism, I won't bother to pass along your comment.    E In some cases the doc folks are constrained to use corporate-standard F document templates.  Those templates can be changed, but it involves aI long time and a lot of gnashing of teeth.  That annoying page heading may  be part of the template.   ------------------------------    Date: 10 Aug 2005 07:21:48 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) 8 Subject: Re: Changing Uppercase filenames into Lowercase3 Message-ID: <rI2y$yeWEDS8@eisner.encompasserve.org>   J In article <42F95637.50700@triumf.ca>, Fred Bach <music@triumf.ca> writes:  ; >   I already found MAKEUP.COM which breaks down a filename ; >   into its parts (device,directory,name,type,version) and 9 >   uses the RENAME command to rename lowercase names and  >   types to uppercase.  [...] 9 >   The filename remains uppercase because DCL uppercases < >   everything on MAKEUP.COM's RENAME line.  And, the RENAME; >   line in MAKEUP.COM does not seem to work with quotation ' >   marks around the lowercase portion.  [...]  > 8 >   But am I missing something easy?  Is there any other5 >   way to easily rename filenames to lowercase?  Has : >   something just zipped right over my head?  Searches on= >   google.com and Openvms.org didn't produce useful results. 6 >   Can the CONVERT command be used to do what I want?  8    To get DCL to use lowercase filenames you must enter '       $set process/parse_style=extended D    such as many of us now have in our login.com.  Then commands such1    as RENAME will try to honor the case as given.   F    To rename a file and have only the case of the name change you mustG    purge down to one version or rename all versions to a temporary name F    then to the name with only the case changed.  VMS will not make oneA    version of many with a different case than the other versions.   6    And of course, this is assuming you're using ODS-5.   ------------------------------    Date: 10 Aug 2005 09:41:53 -0700 From: mb301@hotmail.com  Subject: DECnet-plus problemB Message-ID: <1123692113.068637.19880@z14g2000cwz.googlegroups.com>  G Have a strange problem with Compaq DECnet-Plus V7.2-1 ECO06 16-SEP-2002    $ set host 1.4 welcome to alpha1 etc...  	 Username:    $ set host alpha1  welcome to alpha1 etc...  	 Username:    $ dir 1.4:: 3 %DIRECT-E-OPENIN, error opening 1.4::*.*;* as input / -RMS-E-FND, ACP file or directory lookup failed = -SYSTEM-F-UNREACHABLE, remote node is not currently reachable    $ dir alpha1::6 %DIRECT-E-OPENIN, error opening alpha1::*.*;* as input/ -RMS-E-FND, ACP file or directory lookup failed = -SYSTEM-F-UNREACHABLE, remote node is not currently reachable   ; Have no idea of what to check can anyone give any pointers?    Mark   ------------------------------    Date: 10 Aug 2005 02:17:02 -0700) From: "Bob Gezelter" <gezelter@rlgsc.com> = Subject: Re: HEAT RELATED PROBLEMS; POWER OFF/ON, SUGGESTIONS A Message-ID: <1123665422.007734.5720@g43g2000cwa.googlegroups.com>    John,   F OpenVMS System Management has been part of my practice portfolio sinceD Version 1.0 (e.g., VAX/VMS, circa 1978). In that entire time, I haveG only had to shut clusters, or for that matter, individual machines down @ for identified reasons (e.g., hardware failures, (some) softwareE changes, power outages, machine room construction). Routine shutdowns " and reboots are simply not needed.  D First step, check if the overall temperature is within the operatingE environment of the hardware involved.  86 to 93 degrees Fahrenheit is E high for traditional machine room environments, but frequently within F the operating range of DEC/COMPAQ/HP servers. The same may not be true: of some of the third party auxilliary communications gear.  F The temperature may be only a secondary problem. More importantly, youE may have some environmental "hot spots" with particularly low airflow C and correspondingly higher temperatures than are being recorded. As E ambient temperature increases, issues of airflow beome more critical. = The airflow that is sufficient at 68 is not sufficient at 90.    A couple of steps:  > - Check that all your in cabinet fans are working. I have seenE situations where one or more fans failed, but the ambient temperature C was low enough that the situation did not become critical. When the A temperature went up, the equipment overheated, not because of the = ambient, but the combination of the reduced airflow at higher  temperature B - Invest a few dollars in some thermometers, possibly the MIN/MAX,B recording, or if you can get it easily, one of the multipoint data= collection systems. Find where your concealed "hot spots" are E - Go to your local home improvement store/discount store and purchase / some pedestal fans and eliminate the hot spots.   G In short, shutdowns are not necessarily the answer, but the problem can F likely be dealt with. Not to mention that the air conditioning problemC needs to be addressed. The measures that I mention are stop gaps to # avoid system failures or shutdowns.   ! I hope tha the above is helpful.    $ - Bob Gezelter, http://www.rlgsc.com   ------------------------------  % Date: Wed, 10 Aug 2005 16:48:00 +0200 " From: Jiri Kulhan <jiri@kulhan.cz>3 Subject: HP works to bring Linux to NonStop servers + Message-ID: <3luid5Fuo5v7U1@individual.net>   B IBM is donating mainframes to universities to train new mainframe  professionals.H HP is doing something similar, but instead of training people with it's I valuable operating systems, they are donating NonStop servers with Linux.   W http://news.com.com/HP+works+to+bring+Linux+to+NonStop+servers/2100-7344_3-5825765.html   = What is the sense of this? Are they going to abandon NonStop?     B The good part of the news is that HP know it's worth to give some F computers to universities. Hope that somebody can persuade them to do 7 the same with operating systems (OpenVMS, for example).    J.   ------------------------------  % Date: Wed, 10 Aug 2005 11:15:35 -0400 ( From: Bill Todd <billtodd@metrocast.net>7 Subject: Re: HP works to bring Linux to NonStop servers R Message-ID: <yeqdnZ2dnZ3vAGf1nZ2dnYWHZ9-dnZ2dRVn-zp2dnZ0@metrocastcablevision.com>   Jiri Kulhan wrote:D > IBM is donating mainframes to universities to train new mainframe  > professionals.J > HP is doing something similar, but instead of training people with it's K > valuable operating systems, they are donating NonStop servers with Linux.  > Z > http://news.com.com/HP+works+to+bring+Linux+to+NonStop+servers/2100-7344_3-5825765.html  >  > ? > What is the sense of this? Are they going to abandon NonStop?   G Why, surely not!  Just look at their commitment to the continuation of  F high-end, proprietary technology like PA-RISC (and Alpha, even though I that was nominally Compaq's fault), and Tru64 (HP killed that, after the  & acquisition was announced), and MPE...  F To think that a high-tech stalwart like HP would ever abandon (or, in E cases like VMS, ignore) its high-end treasures and embrace commodity  & solutions is simply absurd - isn't it?   - bill   ------------------------------  % Date: Wed, 10 Aug 2005 11:49:28 -0400 ' From: "Main, Kerry" <Kerry.Main@hp.com> 7 Subject: RE: HP works to bring Linux to NonStop servers R Message-ID: <FD827B33AB0D9C4E92EACEEFEE2BA2FB6B1CDB@tayexc19.americas.cpqcorp.net>   > -----Original Message-----. > From: Jiri Kulhan [mailto:jiri@kulhan.cz]=20  > Sent: August 10, 2005 10:48 AM > To: Info-VAX@Mvb.Saic.Com 5 > Subject: HP works to bring Linux to NonStop servers  >=20F > IBM is donating mainframes to universities to train new mainframe=20 > professionals.B > HP is doing something similar, but instead of training people=20 > with it's=20B > valuable operating systems, they are donating NonStop servers=20
 > with Linux.  >=20@ > http://news.com.com/HP+works+to+bring+Linux+to+NonStop+servers > /2100-7344_3-5825765.html  >=20? > What is the sense of this? Are they going to abandon NonStop?  >=20 >=20F > The good part of the news is that HP know it's worth to give some=20J > computers to universities. Hope that somebody can persuade them to do=209 > the same with operating systems (OpenVMS, for example).  >=20 > J. >=20   Do you mean like the following? . http://h71000.www7.hp.com/news/ospp_turin.html  ? Also, check out Turin (and other Cust's) feedback on the latest  Integrity video:@ http://h71000.www7.hp.com/openvms/integrity/integrity_video.htmlJ http://h71000.www7.hp.com/openvms/integrity/integrity_video_script.html=20  H Marco Mezzalama, Vice Chancellor Professor of Computer Engineering Turin	 Polytech:   E "Our mission at the Politecnico Turin is to provide our students with C the highest level of instruction and access to the best technology. C Currently we have HP AlphaServers running OpenVMS with business and   administrative critical data.=20  @ As we expect our application to expand in the next year the easyE integration of Integrity servers with our Alpha cluster will ensure a * secure, smooth and reliable transition.=20  H We have successfully ported security application to OpenVMS on IntegrityH servers and realized a 30% performance gain in cryptographic algorithms.C In the next year the Politecnico will increase greatly its size and G scope. So also the computing infrastructure must be extremely flexible, E scalability, cost effective and preserving today's investment. We are D confident that HP can give us in the future as in the past all these values."   Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Wed, 10 Aug 2005 18:26:28 +0200 " From: Jiri Kulhan <jiri@kulhan.cz>7 Subject: Re: HP works to bring Linux to NonStop servers , Message-ID: <3luo5qF14dil6U1@individual.net>   >>D >>The good part of the news is that HP know it's worth to give some H >>computers to universities. Hope that somebody can persuade them to do 9 >>the same with operating systems (OpenVMS, for example).  >> >>J. >> >  > ! > Do you mean like the following?   0 > http://h71000.www7.hp.com/news/ospp_turin.html  C Well, something like this. But this is only one university, do you   really think it's enough?   9 Do you have some numbers about the CSLG and ELP programs?    J.   ------------------------------  % Date: Wed, 10 Aug 2005 12:36:27 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 7 Subject: Re: HP works to bring Linux to NonStop servers , Message-ID: <42FA2D02.F6199526@teksavvy.com>  H It is "neat" that HP would donate a Tandem system to a university so the students can play with Linux.   N Consdider the major software providers for the Tandem platform such as Logica,M Base20 etc. Having universities get kids to play on Tandem/Linux combinations K would create potential employees who would be good, once really educated on Q serious business software, to help port from NSK to Linux the real aopplications.   K However, don't expect any of the stuff the university does to ever run in a  mission critical environment.     I Now, I really do not know what HP's plans with regards to NSK vs Linux on M fault tolerant IA64s are. Nor do I know what customers and software providers 6 really think of Linux in a fault tolerant environment.   ------------------------------    Date: 10 Aug 2005 09:50:17 -0700 From: bob@instantwhip.com 7 Subject: Re: HP works to bring Linux to NonStop servers C Message-ID: <1123692617.778159.321400@g47g2000cwa.googlegroups.com>   8 one look at the cert counts for linux should answer that question ...   ------------------------------  # Date: Wed, 10 Aug 2005 17:17:49 GMT ( From: Alan Greig <greigaln@netscape.net>7 Subject: Re: HP works to bring Linux to NonStop servers : Message-ID: <1FqKe.4964$3N4.493@fe1.news.blueyonder.co.uk>   Main, Kerry wrote:    J > We have successfully ported security application to OpenVMS on IntegrityJ > servers and realized a 30% performance gain in cryptographic algorithms.  C And that's supposed to be good? I always got *far* more than a 30%  F performance improvement across the board any time I replaced a VAX or H Alpha. Would have been a pointless upgrade otherwise. And unless we are E told the specs of the Alpha and Itanium processors involved this 30%  # figure tells me absolutely nothing.   E > In the next year the Politecnico will increase greatly its size and I > scope. So also the computing infrastructure must be extremely flexible, G > scalability, cost effective and preserving today's investment. We are F > confident that HP can give us in the future as in the past all these
 > values."  I I am fairly confident that HP/Compaq/DEC's past actions can help predict   the future as well...      --  
 Alan Greig   ------------------------------  % Date: Wed, 10 Aug 2005 10:15:34 -0400 ( From: Bill Todd <billtodd@metrocast.net>H Subject: Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine= Message-ID: <F-ydnWobzoGakWffRVn-sg@metrocastcablevision.com>    JF Mezei wrote:  > Bill Todd wrote: > ; >>Without AMD's push, *all* of x86 would be on its way out.  >  >  > I am not convinced of that.    To be blunt:  so what?  /   While this may have been the original plan at M > Intel to justify spending billions on that IA64 contraption, at once point, O > well before AMD started to be more than a token competitor under leash, Intel K > realised that IA64 wouldn't cut it as "industry standard" high volume and L > abandonned plans for it to replace the 8086. This was before June 25 2001.  I AMD 'started to be more than a token competitor under leash' with its K5  I (in-house-developed) and K6 (development acquired from NexGen) chips; in  F fact, its first at least partially-independent x86-compatible designs I date back to 1991 for the Am386 and 1993 for the Am486, though while the  H circuits in those chips were AMD's at least their early microcode still H owed major debts to Intel).  K5 (1995) was disappointing, but K6 (early ? 1997) and its K6-II and K6-III follow-ons were very legitimate  F competitors even to Intel's top-of-the-line products.  By 1999 and K7 H (Athlon) AMD had become a full-fledged threat - which quickly blossomed D into actual superiority as Athlon clock rates beat Intel's to 1 GHz.  E By contrast, Intel didn't even know that it couldn't fit Merced on a  G feasible die in a 250 nm. process until mid-1996.  It didn't know just  H how miserable Merced's performance would be for another couple of years F after that, and even then clung to the hope that McKinley would atone I for all Merced's sins until close to the time of the Alphacide (remember  G Rob's triumphant revelation that Intel engineers had told their Compaq  E colleagues in February, 2001 that McKinley could clock at 1.4 GHz or   possibly even higher?).   G Had AMD never shown the kind of independent competitiveness that began  G with K5/K6 (or arguably even earlier with Am386/486), Intel would have  B been happy to follow its strategy (as expressed internally to Bob E Colwell of the Intel x86 group back in 1994) of allowing its own x86  F development to wind down over the late '90s - since it would have had H all the time in the world for Itanic, however disappointing its ramp-up  may have been, to catch up.   H So AMD's vigorous *32-bit* competition is what kept x86 moving ahead at B a pace which Itanic could not catch (leaving aside Itanic's other F deficiencies in power and chip area compared to comparably-performing F x86 parts).  AMD's much more recent 64-bit competition is merely what E will leave Itanic no mid-to-high-range niche in which to survive and  I nurse its wounds, hoping someday to re-emerge and complete Intel's "take  G over the world with an architecture which makes compatible competition   virtually impossible" strategy.    - bill   ------------------------------    Date: 10 Aug 2005 08:36:14 -0700 From: icerq4a@spray.seH Subject: Re: InformationWeek: Intel Says Itanium Chip Is Doing Just FineB Message-ID: <1123688174.108981.13050@g44g2000cwa.googlegroups.com>   Bill Todd wrote: > icerq4a@spray.se wrote:  >  > ...  > 5 > > Intel has never said 2006 for the common chipset.  > I > While I haven't seen an Intel press release on the subject (it was only E > a few weeks ago that they started talking to the press about CSI at G > all), Intel was *reported* to have been shooting to introduce CSI for F > x86 by the end of 2006 (and originally for Itanic as well, back when' > Tukwila was scheduled for late 2006).   C Intel began talking about a common chipset (before the CSI name was @ known)some 2(?) years back and that was about to happen in 2007.F Tukwila and CSI has never been anything other than 2007, now in second half of 2007 though.   > D > > In the very recent roadmaps I have seen Intel still say 2007 for > > Tukwila. > F > Funny - they don't on those I've seen:  they go 2005, 2006, and thenJ > 'future' (no explicit date provided).  See some recent Inquirer pictures > of them, for example.   G No, they went from Future (which some believed to be 2006, but that was B never Intels intention) to 2007. I don't think you can show me any7 public evidence that Tukwila was scheduled before 2007.    ------------------------------    Date: 10 Aug 2005 08:42:30 -0700 From: icerq4a@spray.seH Subject: Re: InformationWeek: Intel Says Itanium Chip Is Doing Just FineB Message-ID: <1123688550.504367.28490@z14g2000cwz.googlegroups.com>   Bill Todd wrote: > icerq4a@spray.se wrote:  >  > ...  > J > > There is nothing wrong in the McKinley core as I see it. It is doing a > > very good job. > H > At consuming inordinate amounts of power and cache for the performance > it provides, anyway.  C I agree with power. The caches in McKinley/Madison also drew fairly F much power, but they have fixed that in Montecito now with a clockless asynch design.  5 >   The Itanium microarchitecture designer have had a = > > tougher job than a lot of other microarchitecure designer  > G > And have had over a decade now to get it right (as much time as Alpha F > had between its conception and EV6, in fact).  Not to mention havingH > off-loaded a significant portion of the job onto the compiler writers.  F OK, I was not so precise. I was relating to the PA-RISC teams with SamE Naffziger. I think I remember that they were starting to do Itanium 2 
 work in 2000.    ------------------------------  % Date: Wed, 10 Aug 2005 12:19:17 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> H Subject: Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine, Message-ID: <42FA28FD.546086A1@teksavvy.com>   icerq4a@spray.se wrote: I > No, they went from Future (which some believed to be 2006, but that was D > never Intels intention) to 2007. I don't think you can show me any9 > public evidence that Tukwila was scheduled before 2007.   G Agreed. Starting early in 2004, Intel started to send out "pulling out" N messages about IA64 and 2007 was always the key date where the 8086 would cost" the same and share common chipset.   ------------------------------  % Date: Wed, 10 Aug 2005 12:44:43 -0400 ' From: Dave Froble <davef@tsoft-inc.com> H Subject: Re: InformationWeek: Intel Says Itanium Chip Is Doing Just Fine0 Message-ID: <11fkbg1c0raru8d@corp.supernews.com>   icerq4a@spray.se wrote:   % Pulling one statement out of context:   3 > The Itanium microarchitecture designer have had a ; > tougher job than a lot of other microarchitecure designer   E Why is that?  Could it be the decision to continue with a dumb idea?  G What's so hard about saying "Alpha and Power and such OOO designs have  / proved to be feasible and fast, lets go there"?    --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------    Date: 10 Aug 2005 10:08:44 -0700 From: icerq4a@spray.seH Subject: Re: InformationWeek: Intel Says Itanium Chip Is Doing Just FineC Message-ID: <1123693724.101677.231270@z14g2000cwz.googlegroups.com>    Dave Froble wrote: > icerq4a@spray.se wrote:  > ' > Pulling one statement out of context:  > 5 > > The Itanium microarchitecture designer have had a = > > tougher job than a lot of other microarchitecure designer  > F > Why is that?  Could it be the decision to continue with a dumb idea?  G No, because they were breaking new grounds. If the majority of the CPUs G would have been EPIC/VLIW type of devices, a new and before unknown OOO 0 thinking design would also have had a tough job.  H > What's so hard about saying "Alpha and Power and such OOO designs have1 > proved to be feasible and fast, lets go there"?   C I don't know. ;) OOO is good, but there is atleast interesting that C people try other alternatives. :) EPIC has atleast been able to get 2 itself together now, with comparable speed of OOO.   ------------------------------    Date: 10 Aug 2005 08:52:04 -0700 From: ZJJJ.JJJZ@gmail.com % Subject: Re: Is OVMS going Bangalore? C Message-ID: <1123689124.613857.294980@g49g2000cwa.googlegroups.com>   F 3 out of the first 4 replies were from the people close to the action.  G Does it mean that they have an assurance from some corner OR is it just  their paranoia?    ------------------------------  % Date: Wed, 10 Aug 2005 12:21:51 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> % Subject: Re: Is OVMS going Bangalore? , Message-ID: <42FA2998.F6E52184@teksavvy.com>   ZJJJ.JJJZ@gmail.com wrote:H > 3 out of the first 4 replies were from the people close to the action. > I > Does it mean that they have an assurance from some corner OR is it just  > their paranoia?   N No. it just means that they have not been given any information to support theG rumours of the move to India etc. Just like they had not been given any I information about the murder of Alpha before it was publically announced.    ------------------------------  # Date: Wed, 10 Aug 2005 16:03:23 GMT 6 From: "Kenneth Farmer" <kfarmer@NOSPAM.spyderbyte.com>% Subject: Re: Is OVMS going Bangalore? ? Message-ID: <fzpKe.107801$3j2.3820571@twister.southeast.rr.com>   6 "Alan Greig" <greigaln@netscape.net> wrote in message 4 news:AK8Ke.1011$3N4.261@fe1.news.blueyonder.co.uk... >  > Dan O'Reilly wrote:  > + >> At 01:44 PM 8/9/2005, John Reagan wrote:  >>4 >>> Yes, we've all read the article.  It isn't true. >>K >> Well, for what it's worth, the Colorado Springs Gazette had, a couple of E >> days ago, an article on this.  It mentioned not only the Inquirer   >> article, L >> but also a "secret HP memo", whose existence was confirmed supposedly by  >> aG >> senior HP official, stating much of the same thing that the Inquirer  >> said. > P > http://www.gazette.com/display_search.php?id=1309554&table=story_archive&sec=1     People,   D When you see something like this please let me know or submit it to  OpenVMS.org for posting.  " http://www.openvms.org/contrib.php   Thanks.  :)    Ken   % _____________________________________ " Kenneth R. Farmer <>< 336-736-73763 www.OpenVMS.org | dba.OpenVMS.org | dcl.OpenVMS.org  HP OpenVMS News and Info     ------------------------------  % Date: Wed, 10 Aug 2005 10:27:36 +0200  From: S <soterroatyahoodotcom>@ Subject: Re: Killing a process that has allocated the tape drive& Message-ID: <42f9ba79$1@news1.ethz.ch>   Larry Kilgallen wrote:F  > I would gather, then, that your ideal world is centered on paintingG  > the roses red.  There is still code which has provided an address to G  > hardware, which has the ability to DMA to the address.  Renaming the I  > code to be something else accomplishes nothing for the stated problem.   F Well well well, you didn't get my point. So I said: the driver itself F cannot decide the DMA at will that way. Also it's easier to stop (and G its process to kill) as it's less privileged - it's just a tape driver   why would it be kernel anyway.@ Now if the DMA was already started, that's indeed another story.   Dave Froble wrote:H > returning control to the process, would be to designate the memory in J > question as no longer belonging to the process, but be placed into some J > type of 'unusable system memory' where the OS would wait for the device I > timeout, transfer, or whatever, and release the memory and device when   > and as appropriate.   F Note: if the process stays in the memory it's not only its slot which I takes space but also the allocated memory. We are talking about physical  F memory (for DMA use) not something swappable, so it can be a sensible   loss for some particular system.  D On a normal hardware implementation the faulty device itself can be G disconnected from the bus so its controller can DMA at will in the air  D until it bumps off. Maybe even then, whenever the device feels like D being able to talk again 'normally' its data paths can be connected 1 again to the bus, welcome back online mr. device.   G It's a matter of the software quality if that runs fine or not. If the  D aim is to keep the hardware management (and all extra purpose) code H minimal then of course such features can never be implemented. CP/M was I a heck of a stable system, but why aren't people using it anymore? Maybe   they needed extra features.    S    ------------------------------    Date: 10 Aug 2005 07:05:30 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) @ Subject: Re: Killing a process that has allocated the tape drive3 Message-ID: <$HI+qsAQjIgg@eisner.encompasserve.org>   Z In article <11fi86e137l1h0f@corp.supernews.com>, Dave Froble <davef@tsoft-inc.com> writes: > J >  From the perspective of an in-progress DMA transfer, it seems the main H > concern is the section of memory in which the device may at some time H > write some data.  One possible solution when it's desired to interupt B > the DMA transfer, such as for killing the process, or even just H > returning control to the process, would be to designate the memory in J > question as no longer belonging to the process, but be placed into some J > type of 'unusable system memory' where the OS would wait for the device I > timeout, transfer, or whatever, and release the memory and device when   > and as appropriate.   H    Yep, that could be done, if the process rundown routines had a way ofE    digging through the I/O data base and finding out what pages those G    were.  A lot of work just to cure a percieved problem that goes away F    by itself after milliseconds in most cases and worst case about 20 D    minutes on a properly operating system with one of those painfull    little tape drives.   ------------------------------    Date: 10 Aug 2005 07:07:47 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) @ Subject: Re: Killing a process that has allocated the tape drive3 Message-ID: <echIFX06NniU@eisner.encompasserve.org>   Z In article <11fipe75d9e17b7@corp.supernews.com>, Dave Froble <davef@tsoft-inc.com> writes: > J > But if that hung process is also holding onto your terminal session you H > might become annoyed.  Yes, on a PC with a terminal emulator, you can 5 > just start up another session.  Times have changed.   G    Now there's an area that might be addressed with less work.  Perhaps D    the kernel could go ahead and $DASSGN (implies $CANCEL) the otherG    channels in parallel instead of working through them serially.  Then E    the RWAST process wouldn't be tying up resources it wasn't waiting     on.   ------------------------------    Date: 10 Aug 2005 07:13:17 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) @ Subject: Re: Killing a process that has allocated the tape drive3 Message-ID: <0ihYk$zBSR9n@eisner.encompasserve.org>   G In article <42f9ba79$1@news1.ethz.ch>, S <soterroatyahoodotcom> writes:  > H > Well well well, you didn't get my point. So I said: the driver itself H > cannot decide the DMA at will that way. Also it's easier to stop (and I > its process to kill) as it's less privileged - it's just a tape driver    > why would it be kernel anyway.B > Now if the DMA was already started, that's indeed another story.  G    The tape driver is kernel because only the kernel can manipulate I/O C    space.  This enables one to implement things some of us feel are D    important, like security.  I can't imagine using a system where IG    don't have security on my tape files.  Oh, wait a minute, its called     UNIX.  D    I know of no reports of anyone getting RWAST on a tape drive whenE    an I/O operation hasn't been started.  That leaves the question of C    whether the hardware was designed to do certain operations, like 0    seek, without a DMA.  Back to earlier post...   ------------------------------  # Date: Wed, 10 Aug 2005 14:07:51 GMT 7 From: John Malmberg <malmberg@dskwld.zko.dec.compaq.hp> @ Subject: Re: Killing a process that has allocated the tape drive2 Message-ID: <XSnKe.10140$nD3.637@news.cpqcorp.net>  H I will try to clear up some things in this thread.  Keep in mind that I + do not normally work with the tape drivers.   I The tape drive in the original post was named $1$MGA which means it is a   Fibre Channel device.   D This means that the maximum time that a process should be in a hung F state because of a Tape problem is the value of TAPE_MVTMEOUT + (Time I expected for tape operation * error retry count).  TAPE_MVTIMEOUT is set  I to 10 minutes on my DS10, but I do not think that value is checked until  6 the error retries and recovery attempts are exhausted.  F If a process is in this hung state for more than that amount of time, H then that is a Fibre/SCSI-Tape driver bug somewhere, even if there is a I problem with the hardware that was the initial cause.  This is something  D that logging a call to HP Software Support may be needed to resolve.  H In the original posting case the DMA transfers are under the control of G the Fibre Channel Adapter hardware, and until it confirms that the I/O  G has been canceled the memory needs to remain locked by the tape driver.   H As the original poster has not responded to this thread, we do not know G what version of OpenVMS they are running, nor do we know what ECOs are  C present, or even if after TAPE_MVTIMEOUT + ??? elapsed the process   cleaned it self up.    John! malmberg@dskwld.zko.dec.compaq.hp  Personal Opinion Only    ------------------------------  % Date: Wed, 10 Aug 2005 11:19:03 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> @ Subject: Re: Killing a process that has allocated the tape drive, Message-ID: <42FA1AE6.987C6A1C@teksavvy.com>   S wrote:G > Well well well, you didn't get my point. So I said: the driver itself G > cannot decide the DMA at will that way. Also it's easier to stop (and H > its process to kill) as it's less privileged - it's just a tape driver  > why would it be kernel anyway.B > Now if the DMA was already started, that's indeed another story.    I That is the whole point here. Driver tells tape drive "find me the end of ? tape, when done, deposit the information at memory location X".   K Tape drive then take 45 minutes to locate end of tape on a TK50. Meanwhile, * user gets impatient and stops his process.  H If the tape drive doesn't have a "cancel immediatly", then the operatingM system knows that there is pending IO on that device which will deposit stuff K in the dying process ' memory. So VMS keeps process on life support so that " the memory location remains valid.  M What I find odd is that tape devices wouldn't have had a "RESET" command that J the driver could issue to stop the previous IO.  Seems rather odd that VMS< seems to be the only OS that had problems with tape drives.   G > Note: if the process stays in the memory it's not only its slot which J > takes space but also the allocated memory. We are talking about physicalG > memory (for DMA use) not something swappable, so it can be a sensible " > loss for some particular system.  M The amount of memory needed to keep the process on life support while waiting N for the tape IO to complete is not very big. The problem is that the system isH deprived of a functioning tape drive until reboot.  If you have a narrow* window for backups, this is a real killer.    E > On a normal hardware implementation the faulty device itself can be H > disconnected from the bus so its controller can DMA at will in the air > until it bumps off.     K Actually, perhaps VMS needs a SET DEVICE/RESET  to cancel any pending IO on ' the device and zap it to initial state. M Doing so would then release the RWASTed process since it woudl no longer have  any pending IO.    ------------------------------    Date: 10 Aug 2005 10:18:36 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen) @ Subject: Re: Killing a process that has allocated the tape drive3 Message-ID: <eqRPFnh7jhkd@eisner.encompasserve.org>   G In article <42f9ba79$1@news1.ethz.ch>, S <soterroatyahoodotcom> writes:  > Larry Kilgallen wrote:H >  > I would gather, then, that your ideal world is centered on paintingI >  > the roses red.  There is still code which has provided an address to I >  > hardware, which has the ability to DMA to the address.  Renaming the K >  > code to be something else accomplishes nothing for the stated problem.  > H > Well well well, you didn't get my point. So I said: the driver itself H > cannot decide the DMA at will that way. Also it's easier to stop (and I > its process to kill) as it's less privileged - it's just a tape driver    > why would it be kernel anyway.B > Now if the DMA was already started, that's indeed another story.  E That is the situation we are talking about -- the command has already D been handed to the hardware.  If the "driver" you envision would not8 interface to the hardware, what purpose would it serve ?  D The only thing a VMS direct IO device driver does in process context( is validate arguments and return status.   ------------------------------  % Date: Wed, 10 Aug 2005 11:39:42 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> @ Subject: Re: Killing a process that has allocated the tape drive, Message-ID: <42FA1FBD.EFAD4DA7@teksavvy.com>   Bob Koehler wrote:I >    were.  A lot of work just to cure a percieved problem that goes away G >    by itself after milliseconds in most cases and worst case about 20 F >    minutes on a properly operating system with one of those painfull >    little tape drives.    M Sorry, but the number of times that the "RWASTed tape issue" has been brought N up here means that it isn't such a rare occurance. By the time someone posts a@ question here, the tape has been stuck for more than 20 minutes.  L BACKUP seems to do OK with tape handling. But mounting tapes as ODS2 devicesJ (which triggers MUAACP) is the problematic issue here and that leads me toK believe that the problem is fixable since BACKUP doesn't have problems with F CTRL-Y, but  MUAACP does, especially if you try to do a DIR on a tape.   ------------------------------  % Date: Wed, 10 Aug 2005 11:51:54 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> @ Subject: Re: Killing a process that has allocated the tape drive, Message-ID: <42FA2297.3E076A31@teksavvy.com>   John Malmberg wrote:E > This means that the maximum time that a process should be in a hung G > state because of a Tape problem is the value of TAPE_MVTMEOUT + (Time 4 > expected for tape operation * error retry count).     K Every tried to do a DIR on a TK50 ?  And then doing CTRL-Y while the DIR is  huffing and puffing ?   )  TAPE_MVTIMOUT has nothing to do with it.    ------------------------------    Date: 10 Aug 2005 12:14:21 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) @ Subject: Re: Killing a process that has allocated the tape drive3 Message-ID: <qmluszwywZVZ@eisner.encompasserve.org>   \ In article <42FA1AE6.987C6A1C@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes:  M > Actually, perhaps VMS needs a SET DEVICE/RESET  to cancel any pending IO on ) > the device and zap it to initial state. O > Doing so would then release the RWASTed process since it woudl no longer have  > any pending IO.   @    VMS is already trying to do this when you rundown an image orH    process.  Both of those rundowns include $DASSGN on all open channels7    which in turn does a $CANCEL on all outstanding I/O.    ------------------------------    Date: 10 Aug 2005 05:06:25 -0700 From: sam@ratex.dk9 Subject: Re: Looking for template software for VMS system C Message-ID: <1123675584.979443.313980@z14g2000cwz.googlegroups.com>    Thomas Wirt skrev: > 7 > I'm looking for a product which will print the static 5 > preprinted template and the variable information on 4 > a single pass.  If such a product is available, my5 > assumption is that the preprinted template would be  > linked to queue.  8 I am using Printkit from Notrhlake www.nls.com for that.. You can download the software and try it for a 30 days trial period.    Sam    ------------------------------  + Date: Wed, 10 Aug 2005 11:48:09 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)1 Subject: Re: looking for TIMES-ROMAN.XDPS$OUTLINE $ Message-ID: <ddcphp$2i7$2@online.de>  D In article <ddcpb4$2i7$1@online.de>, helbig@astro.multiCLOTHESvax.de3 (Phillip Helbig---remove CLOTHES to reply) writes:    F > In article <ddcmom$uc3$1@online.de>, helbig@astro.multiCLOTHESvax.de5 > (Phillip Helbig---remove CLOTHES to reply) writes:   > 8 > > GS tells me, when converting from PostScript to PDF: > > c > > Can't find (or can't open) font file SYS$COMMON:[SYSFONT.XDPS.OUTLINE]TIMES-ROMAN.XDPS$OUTLINE. 8 > > Unable to load default font Times-Roman!  Giving up.# > > Error: /invalidfont in findfont  > > H > > [VMS$COMMON.SYSFONT.XDPS.OUTLINE] exists on my system, but is empty. > > I > > Presumably, all I have to do is install the font in this directory.    > > Where can I get it?  > J > Is this one of the files which was deleted during an upgrade around the A > 7.3-timeframe since support for display PostScript was dropped?  > : > How would a newer version of GhostScript deal with this?  I Another alternative would be to somehow edit the PostScript file, but it  * is 3183 blocks and relatively complicated.   ------------------------------  % Date: Wed, 10 Aug 2005 15:34:21 +0530 ! From: gokrix <gokrix@hotmail.com> Y Subject: Re: OpenVMS under the umbrella - or umbrella is supported at the time? time?time B Message-ID: <1123668241.c08b8509228876dec35f1367f511385c@teranews>   JF Mezei wrote: $ > Sue: marketing gimmick suggestion: > P > Get a new batch of flashing balls, make sure they are slightly bigger than the > original batch..   > & > Then send them to the VMS customers. > J > You can then make the marketing claim "VMS customers have bigger balls"  > 
 > :-) :-) :-)  > M > (This, along with condoms with the VMS logo on it, along with the marketing & > slogan "always up when you need it".   This is a load of bollocks.    :D   Thanks,  --GS   ------------------------------    Date: 10 Aug 2005 07:15:03 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) 4 Subject: Re: strange terminal-characteristic problem3 Message-ID: <Bi27rbWAondb@eisner.encompasserve.org>   \ In article <42F9575A.B10FD4D3@teksavvy.com>, JF Mezei <jfmezei.spamnot@teksavvy.com> writes: > QUESTION:  > J > MC SYSGEN SHOW/DEV reveals that OPA is handled by the "OPERATOR" driver. > L > While for most drivers, you can do a DIR SYS$LOADABLE_IMAGES: and find the3 > relevant file, this is not the case for OPERATOR.  > G > How does one go about finding out in which file this driver resides ?   D   The drivers for OP and MB (mailbox) devices are built into the VMS=   kernel.  You can find the file name on the source listings.    ------------------------------  % Date: Wed, 10 Aug 2005 11:28:37 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 4 Subject: Re: strange terminal-characteristic problem, Message-ID: <42FA1D24.EDAE838A@teksavvy.com>  / Phillip Helbig---remove CLOTHES to reply wrote: H > I also tried JF's suggestion of setting OPA0: to /NOEIGHTBIT.  Doesn't > solve the problem either.   N With /EIGHTBIT turned on, can you type accented characters properly ? Like the  or  or  ?   I If you type  in a an environment where the 8th bit gets chopped off, the  character comes back as i   D > I thus conclude that it is somehow specific to the VAXstation 4000
 > series.   K What is puzzling is that the second serial port seems to work, and when you 2 telnet out from the OPA0:, it then works fine too.  M Have you tried SET TERM/PASSTHROUGH ? (or is it /PASSALL which is still valid M ?)  This removes a layer of interpretation of data from the terminal driver.  E (will affect DCL line editing, but may give EDT and TPU full control)     3 Also, what is the TTY_TYPAHD buffer size ? (SYSGEN)    ------------------------------  % Date: Wed, 10 Aug 2005 11:30:16 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 4 Subject: Re: strange terminal-characteristic problem, Message-ID: <42FA1D87.9503CF16@teksavvy.com>  @ Another thing. Does the process have sufficient bufquo in UAF ?   L Do you have any FMS applications available ? Might be interesting to see how
 that behaves.   9 Have you tried PHONE ? Does it also screw up the screen ?    ------------------------------  % Date: Wed, 10 Aug 2005 12:56:23 +0100 % From: David Gray <police@spamcop.net>  Subject: Telnet over VPN8 Message-ID: <rlojf154d7b6685gaq6vs542adba5i4dqs@4ax.com>   Hi,   C We have a small sub office in Connecticut which is connected to our D LAN in the UK using business DSL down a secure VPN tunnel.  Three orE four times a day the users in CT who are connecting using SmarTerm to B our VMS (7.3-2) system get disconnected and have to log back in.    F Is there anything I can do to make TELNET more resilient without goingB to the expense of leased line?  Is there anyway of controlling the timeout periods?    C The only other traffic going down the tunnel is Outlook to Exchange C server emails which is more forgiving than TELNET when it comes lag D time.  I appreciate that the tunnel is still just using the internet= and therefore sharing the pipe with millions of other users.    D Has anyone used terminal emulation in a browser?  Even if this still= uses TELNET is this web based approach any more forgiving?      
 Equipment 	 =========    CT   ----
 Win 2K PC 
 Smarterm V11   Netgear 10/100 managed switch * Business broadband U/S 38k D/S 1.5mb - 3mb! Symantec SGS420 (secure gateway)     UK ----  ' BT Business broadband U/S 2mb D/S 256k  ! Symantec SGS460 (secure gateway)   10/100 managed switch  ES45 Alpha server 
 VMS 7.3-2  HP TCP/IP Services V5.4   D Can't move to SDSL yet as neither of the local exchanges are enabled yet.    	 Regards,   	Dave.     ------------------------------  # Date: Wed, 10 Aug 2005 12:06:28 GMT ( From: Alan Greig <greigaln@netscape.net> Subject: Re: Telnet over VPN< Message-ID: <85mKe.45959$FG3.8927@fe3.news.blueyonder.co.uk>   David Gray wrote:  > Hi,  >  > Equipment  > =========  >  > CT   > ---- > Win 2K PC  > Smarterm V11    > Netgear 10/100 managed switch , > Business broadband U/S 38k D/S 1.5mb - 3mb  * Do you really mean the uplink is only 38k?  # > Symantec SGS420 (secure gateway)   >  > UK > ---- > ) > BT Business broadband U/S 2mb D/S 256k    I There seems to be a huge mismatch here. Could you reconfirm these speeds  K please. I'm surprised anything works at all if these are the actual speeds.        --  
 Alan Greig   ------------------------------    Date: 10 Aug 2005 07:23:48 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)  Subject: Re: Telnet over VPN3 Message-ID: <cMdtat7d4fcD@eisner.encompasserve.org>   ` In article <rlojf154d7b6685gaq6vs542adba5i4dqs@4ax.com>, David Gray <police@spamcop.net> writes: > Hi,  > E > We have a small sub office in Connecticut which is connected to our F > LAN in the UK using business DSL down a secure VPN tunnel.  Three orG > four times a day the users in CT who are connecting using SmarTerm to D > our VMS (7.3-2) system get disconnected and have to log back in.   >   C    Any possibility you have an idle process killer (such as Hitman)     running?    ------------------------------  % Date: Wed, 10 Aug 2005 13:44:07 +0100 % From: David Gray <police@spamcop.net>  Subject: Re: Telnet over VPN8 Message-ID: <4gtjf1pqo52dhf970h14f20ih535d3e6l5@4ax.com>  D Sorry that was a typo.  The DSL provider in CT tells me the upstream
 speed 384k.     B The DSL speeds for the UK end are correct.  Apparently that is theE best that DSL can provide.  SDSL seems to give the same up & down but - as yet it is not available at either location    Dave.   D On Wed, 10 Aug 2005 12:06:28 GMT, Alan Greig <greigaln@netscape.net> wrote:   >  >  >David Gray wrote: >> Hi,   >>
 >> Equipment   >> ========= >>   >> CT  >> ---- 
 >> Win 2K PC   >> Smarterm V11 ! >> Netgear 10/100 managed switch  - >> Business broadband U/S 38k D/S 1.5mb - 3mb  > + >Do you really mean the uplink is only 38k?  > $ >> Symantec SGS420 (secure gateway)  >>   >> UK  >> ----  >>  * >> BT Business broadband U/S 2mb D/S 256k  > J >There seems to be a huge mismatch here. Could you reconfirm these speeds L >please. I'm surprised anything works at all if these are the actual speeds.   ------------------------------    Date: 10 Aug 2005 05:34:03 -0700 From: dooleys@snowy.net.au Subject: Re: Telnet over VPNC Message-ID: <1123677243.203740.309620@o13g2000cwo.googlegroups.com>   - is "three or four times a day" every 2 hours? = if so check Matt's replies to the thread in this group titled B Why does idle telnet session get disconnected - posted June 8 2004 Phil   ------------------------------  % Date: Wed, 10 Aug 2005 13:48:44 +0100 % From: David Gray <police@spamcop.net>  Subject: Re: Telnet over VPN8 Message-ID: <hltjf15e6lrph755tc4dl2f619mg93ffm9@4ax.com>  F The firmware on both the SGS boxes is updated daily via the Liveupdate feature of the box.       ) On Wed, 10 Aug 2005 11:57:05 +0000 (UTC), C helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to 
 reply) wrote:   D >In article <rlojf154d7b6685gaq6vs542adba5i4dqs@4ax.com>, David Gray ><police@spamcop.net> writes:  > F >> We have a small sub office in Connecticut which is connected to ourG >> LAN in the UK using business DSL down a secure VPN tunnel.  Three or H >> four times a day the users in CT who are connecting using SmarTerm toE >> our VMS (7.3-2) system get disconnected and have to log back in.    > I >No VPN, but I routinely log in via telnet from one VMS system to another D >with DSL being the internet connection at both ends.  I was used toB >telnet connections staying open, even if idle, as long as the DSLE >connection was up.  When I tried another router, I noticed that they D >started timing out.  The vendor claims that this can be fixed by anH >upgrade to even more recent (i.e. very recent---I already upgraded just! >a couple of weeks ago) firmware.  > H >Not sure if it applies in your case, but perhaps this is something you E >can configure in your DSL router and/or fix with a firmware upgrade.    ------------------------------  % Date: Wed, 10 Aug 2005 13:50:18 +0100 % From: David Gray <police@spamcop.net>  Subject: Re: Telnet over VPN8 Message-ID: <dutjf153lik9bukfo3e6o3gu2n8e9iuh6k@4ax.com>  4 There's no process killer running on the VMS system.  F On 10 Aug 2005 07:23:48 -0500, koehler@eisner.nospam.encompasserve.org (Bob Koehler) wrote:  a >In article <rlojf154d7b6685gaq6vs542adba5i4dqs@4ax.com>, David Gray <police@spamcop.net> writes:  >> Hi,   >>  F >> We have a small sub office in Connecticut which is connected to ourG >> LAN in the UK using business DSL down a secure VPN tunnel.  Three or H >> four times a day the users in CT who are connecting using SmarTerm toE >> our VMS (7.3-2) system get disconnected and have to log back in.    >>   > D >   Any possibility you have an idle process killer (such as Hitman) >   running?   ------------------------------  % Date: Wed, 10 Aug 2005 13:50:37 +0100 % From: David Gray <police@spamcop.net>  Subject: Re: Telnet over VPN8 Message-ID: <e0ujf1d5sm0vke6u2sgqshs1t34npaii0g@4ax.com>  ) Thanks, I'll have a look for the thread.    : On 10 Aug 2005 05:34:03 -0700, dooleys@snowy.net.au wrote:  . >is "three or four times a day" every 2 hours?> >if so check Matt's replies to the thread in this group titledC >Why does idle telnet session get disconnected - posted June 8 2004  >Phil    ------------------------------  % Date: Wed, 10 Aug 2005 08:51:42 -0500 * From: Michael Clark <MClark@Nemschoff.com> Subject: RE: Telnet over VPNA Message-ID: <3A0A94E82B68C64D9412D3CA3B32D6A6C73FAD@EMAILSERVER3>    David,  J I believe Telnet has a session disconnnect variable that can be increase I believe.  SSH has the sameJ thing for most operating systems.  Not sure what version of TCPIP you need to support usingI this option.  Basically this increases the amount of time the VMS machine  will wait for the telnet6 session to recover before giving up and disconnecting.  L This is how to change the keepalive values as well if you are having the two
 hour problem.   L http://groups-beta.google.com/group/vmsnet.networks.tcp-ip.multinet/browse_fL rm/thread/370aa493a4a6da9a/878f9b988cca1a90?tvc=1&q=telnet+OpenVMS+timeout&h l=en#878f9b988cca1a90      Michael    -----Original Message-----, From: David Gray [mailto:police@spamcop.net]( Sent: Wednesday, August 10, 2005 7:51 AM To: Info-VAX@Mvb.Saic.Com  Subject: Re: Telnet over VPN      ) Thanks, I'll have a look for the thread.    : On 10 Aug 2005 05:34:03 -0700, dooleys@snowy.net.au wrote:  . >is "three or four times a day" every 2 hours?> >if so check Matt's replies to the thread in this group titledC >Why does idle telnet session get disconnected - posted June 8 2004  >Phil     A CONFIDENTIALITY NOTE: This electronic transmission, including all L attachments, is directed in confidence solely to the person(s) to whom it isL addressed, or an authorized recipient, and may not otherwise be distributed,L copied or disclosed. The contents of the transmission may also be subject toJ intellectual property rights and all such rights are expressly claimed andG are not waived. If you have received this transmission in error, please H notify the sender immediately by return electronic transmission and thenH immediately delete this transmission, including all attachments, without* copying, distributing or disclosing same.    ------------------------------  % Date: Wed, 10 Aug 2005 16:24:35 +0100 % From: David Gray <police@spamcop.net>  Subject: Re: Telnet over VPN8 Message-ID: <jr6kf15aq1002v4l99bor6otnc31apn7kn@4ax.com>  D On Wed, 10 Aug 2005 12:56:53 GMT, Alan Greig <greigaln@netscape.net> wrote:   >  >  >David Gray wrote: > G >> Sorry that was a typo.  The DSL provider in CT tells me the upstream  >> speed 384k.   >>  E >> The DSL speeds for the UK end are correct.  Apparently that is the H >> best that DSL can provide.  SDSL seems to give the same up & down but0 >> as yet it is not available at either location > J >Ok. Telnet is normally fairly resilient. Can you give us ping times from F >the PCs to the Alpha? Without other tools available I'd just leave a G >"continue until stopped" (ping -t on windows)  ping running. It won't  H >hurt your bandwidth but will tell us what sort of round trip times and J >packet loss you are experiencing. Can you also try another telnet client D >on the PC? Possibly just Windows own basic client from the command J >prompt. It won't handle the VT emulation but will give us something else  >to compare.  F Both good ideas.  I'm getting the users to run DOS batch files on eachC PC which will run all day long and output to a log file.  I've been D pinging all the PCs from the alpha for the last few days, is this of	 any use?    F Also getting them to open up a DOS TELNET session and leave it runningC all day long as you suggest to see if it also times out.  Will post  the results in a few days.     > I >It is possible to start playing with timeouts and re-transmit values in  E >the IP stacks but the defaults are pretty generous and these aren't  I >things you should normally touch. How does a telnet session "feel"? Are  J >characters echoed back almost immediately or is there a 30 second lag or  >something?   @ Has anyone had any experiences with the web based VT emulation?    ------------------------------  % Date: Wed, 10 Aug 2005 16:35:48 +0100 % From: David Gray <police@spamcop.net>  Subject: Re: Telnet over VPN8 Message-ID: <gc7kf1hriaf404666qkacrl74tst4i4iup@4ax.com>  D Thanks for the info.  This looks like it's more for Multinet than HP; TCP/IP, I will have a dig around through the help though.     F The timeouts occur even when the users are beavering away (or at least  they say so) with their work.     B There also does not appear to be a timeout setting on the Symantec Gateway Security 420 box.       Dave.   1 On Wed, 10 Aug 2005 08:51:42 -0500, Michael Clark  <MClark@Nemschoff.com> wrote:    >David,  > K >I believe Telnet has a session disconnnect variable that can be increase I  >believe.  SSH has the same K >thing for most operating systems.  Not sure what version of TCPIP you need  >to support using J >this option.  Basically this increases the amount of time the VMS machine >will wait for the telnet 7 >session to recover before giving up and disconnecting.  > M >This is how to change the keepalive values as well if you are having the two  >hour problem. > M >http://groups-beta.google.com/group/vmsnet.networks.tcp-ip.multinet/browse_f M >rm/thread/370aa493a4a6da9a/878f9b988cca1a90?tvc=1&q=telnet+OpenVMS+timeout&h  >l=en#878f9b988cca1a90 >  >  >Michael >  >-----Original Message----- - >From: David Gray [mailto:police@spamcop.net] ) >Sent: Wednesday, August 10, 2005 7:51 AM  >To: Info-VAX@Mvb.Saic.Com >Subject: Re: Telnet over VPN  >  >  > * >Thanks, I'll have a look for the thread.  > ; >On 10 Aug 2005 05:34:03 -0700, dooleys@snowy.net.au wrote:  > / >>is "three or four times a day" every 2 hours? ? >>if so check Matt's replies to the thread in this group titled D >>Why does idle telnet session get disconnected - posted June 8 2004 >>Phil >  > B >CONFIDENTIALITY NOTE: This electronic transmission, including allM >attachments, is directed in confidence solely to the person(s) to whom it is M >addressed, or an authorized recipient, and may not otherwise be distributed, M >copied or disclosed. The contents of the transmission may also be subject to K >intellectual property rights and all such rights are expressly claimed and H >are not waived. If you have received this transmission in error, pleaseI >notify the sender immediately by return electronic transmission and then I >immediately delete this transmission, including all attachments, without + >copying, distributing or disclosing same.   >    ------------------------------  % Date: Wed, 10 Aug 2005 11:36:27 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com>  Subject: Re: Telnet over VPN+ Message-ID: <42FA1EFA.8617CC3@teksavvy.com>    David Gray wrote: H > Is there anything I can do to make TELNET more resilient without goingD > to the expense of leased line?  Is there anyway of controlling the > timeout periods?  D There is a way to make telnet connectiosn disconnectable. When usersN reconnect, they are given the option to reconnect to their previous session inK progress. Not sure of the specifics though, it depends on whcih TCPIP stack 	 you have.   I Does your router provide some sort of syslog ? If so, you may want to log N "events" from your router. If your DSL provider resets your connections duringE the day, there is something to complain about from your DSL provider.   L As an example, my previous ISP would give me about 3 session resets per day.C Figured it was line quality etc etc and never really bothered since M complaining to the previous ISP was risky business. (they have since gone out J of business!).  When I switched to the new ISP, the session resets stoppedM right away, same DSL physical line.  And at one point, the ISP started having J an overloaded router and I had session resets for about a month until they upgraded their router.   ------------------------------  % Date: Wed, 10 Aug 2005 11:48:30 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com>  Subject: Re: Telnet over VPN, Message-ID: <42FA21CC.3D8C33C9@teksavvy.com>   David Gray wrote: D > The DSL speeds for the UK end are correct.  Apparently that is the > best that DSL can provide.    N No. DSL is good for up to 8mbps to you. Not sure how fast from you to network. But some have it at 1mbps.  N Here, I have 3mbs from network, 800kbps to network. "Premium" service is 4mbs.  J The telco decides how fast they want it. And it generally is a question ofL competition from the cable company. When the cable company announces a speedO increase and the telco sees drop is subscribers, the telco then ups its speed.    D The problem is that every speed increase in DSL applies to a smallerK percentage of subscribers. The further you are from the DLSAM (at the CO or : street concentrator), the lower the max speed is for DSL.   I The real reason neither the cable nor DSL companies are providing maximum N possible speed is just marketing. They only need to be much better than dialupF to get customers. Once enough customers have moved to high speed, then9 competition kicks in and they are forced to raise speeds.    ------------------------------  % Date: Wed, 10 Aug 2005 17:52:37 +0100 % From: David Gray <police@spamcop.net>  Subject: Re: Telnet over VPN8 Message-ID: <grbkf1to693cbo0ni31accnkojbphvpmfu@4ax.com>  , On Wed, 10 Aug 2005 11:36:27 -0400, JF Mezei% <jfmezei.spamnot@teksavvy.com> wrote:    >David Gray wrote:I >> Is there anything I can do to make TELNET more resilient without going E >> to the expense of leased line?  Is there anyway of controlling the  >> timeout periods?  > E >There is a way to make telnet connectiosn disconnectable. When users O >reconnect, they are given the option to reconnect to their previous session in L >progress. Not sure of the specifics though, it depends on whcih TCPIP stack
 >you have.   I'm using HP TCP/IP services.   ;   HP TCP/IP Services for OpenVMS Alpha Version V5.4 - ECO 4 7   on a AlphaServer ES45 Model 2B running OpenVMS V7.3-2    F Are you saying that after a disconnection the user is given the option@ of reconnecting to the lost sesssion?  I have noticed that thereE processes hang around for a while after disconnection so reconnecting ? would be helpful.  I guess I need to have a look through the VT  emulator settings.      J >Does your router provide some sort of syslog ? If so, you may want to logO >"events" from your router. If your DSL provider resets your connections during F >the day, there is something to complain about from your DSL provider. > M >As an example, my previous ISP would give me about 3 session resets per day. D >Figured it was line quality etc etc and never really bothered sinceN >complaining to the previous ISP was risky business. (they have since gone outK >of business!).  When I switched to the new ISP, the session resets stopped N >right away, same DSL physical line.  And at one point, the ISP started havingK >an overloaded router and I had session resets for about a month until they  >upgraded their router.   E That's interesting.  The connection drops off at about 2pm local time @ every day and from then on the quality of the connection is muchF worse.   Have started the business of complaining (loudly) to the ISP.   ------------------------------    Date: 10 Aug 2005 09:55:49 -0700# From: "WhoDat?" <whohe@whoever.com>  Subject: Re: Telnet over VPNB Message-ID: <1123692949.448010.85560@g14g2000cwa.googlegroups.com>   David Gray wrote: F > On Wed, 10 Aug 2005 12:56:53 GMT, Alan Greig <greigaln@netscape.net> > wrote: > A > Has anyone had any experiences with the web based VT emulation?   . Compared to your direct VPN-TELNET connection:  ? You can connect to a remote Windows or Citrix server via client C software and run a VT emulator on that server. That doesn't work as G well as direct VPN-telnet because the mapped PF keys keep getting reset E by either the local PC or one of the layers between you and the host.   G There are Web-host based emulators where you browse to a web-server and G open the emulator at the host end. This is similar to the above example E except you only need a browser on the client side rather than special  client software.  E The above two options both require you to connect to a non-VMS server F at the host end to run the emulator to get to the VMS system. I've notD seen pure VMS based software to do this, but maybe someone else has?  F Some emulators allow you to open a VT session in a browser window, butG the VT emulator still runs on the PC and there is no real difference in   the way it connects to the host.  E I've also seen what are basically browser based "screen-scapers" that G let you map your VT screens to a browser interface. Those require a bit G of work and the test-drives I've had with them weren't very satisfying.   A Attachmate and WRQ were probably the leaders in this area and now D they've merged into AttachmateWRQ; I have no experience with the new company.   ------------------------------  % Date: Wed, 10 Aug 2005 13:20:33 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com>  Subject: Re: Telnet over VPN, Message-ID: <42FA3755.CB326653@teksavvy.com>   David Gray wrote: G > processes hang around for a while after disconnection so reconnecting A > would be helpful.  I guess I need to have a look through the VT  > emulator settings.  M disconnectable sessions are a server side thing, not a VT emulator thing. The L server maintains the session in a comatise state (disconnected) for a numberD of minutes after the link to the terminal has been lost. If the userM reconnects within those minutes, after entering password he is prompted if he < wants to reconnect to a previous session or start a new one.  G > That's interesting.  The connection drops off at about 2pm local time B > every day and from then on the quality of the connection is muchH > worse.   Have started the business of complaining (loudly) to the ISP.    E Not sure about the large scale ISP/Telcos in the USA. But the smaller N alternative ISPs in Canada (who buy a service from the big Telco to run DSL toJ customers and carry packets via ATM to their premises) have easy access toI their router's logs to see when and why a PPPoE session was dropped.  The N drones who answer the phone at the large telco ISPs probably don't have access to those logs.  H If your DSL service doesn't use the pesky PPPoE aberration, DSL servicesH shouldn't be dropping unless your modem actually loses sync or they have' serious problems at the central office.   C You can do an easy test of telnet's resilience during "good" hours.   N With a user logged in, ask him to pull the ethernet out of his PC, and see how2 long it takes for his session to drop at your end.  N Now, in a VPN environment, the router may also be playing a role in this if itM does more than encapsulate packets and actually maintains individual sessions  (as is the case with NAT).    < This is why having access to the router's log is good thing.   ------------------------------   End of INFO-VAX 2005.444 ************************