HECnet

hecnet@lists.dfupdate.se

15556 discussions

DECnet performance problems (on TOPS-10)

by bob＠jfcl.com

but echoes TOPS-10 style (which is also RSTS style) not VMS/RSX style. Ok, I'll bite - what's the difference in echo style between TOPS10/RSTS and VMS/RSX? And which category does TOPS20 belong to? Bob

13 years, 8 months

DECnet performance problems (on TOPS-10)

by bqt＠softjar.se

And just fyi: in RSX, you have different programs for connecting to different systems, when using "rterm". RMT.TSK is for connecting to RSX systems. RVT.TSK is for connecting to VMS. RRS.TSK is for connecting to RSTS/E. HOST.TSK is for connecting to TOPS-20. Johnny On 2011-09-01 12.47, Paul_Koning at Dell.com wrote: The pre-cterm remote terminal protocol is not actually a single protocol, but rather a set of protocols specific to the destination OS. That's why it works well: instead of trying to fit the OS native terminal behavior into a VMS-centric model as is done in Cterm, the rterm protocols simply express a terminal behavior that matches what the OS wants. The reason Cterm is so bad for TOPS-10 is that there is really nothing that matches -- just for starters, TOPS-10 echoes differently. The reason you may see failure for one destination OS type and not others is that you're going through completely different code, because you have a different rterm flavor. There are four flavors (according to the RSTS code I have that implements them): VMS, RSX, RSTS, and TOPS-20. If I remember right, TOPS-10 uses the RSTS flavor, and RT-11 perhaps does as well. The RSTS flavor is line oriented (as is Cterm) but echoes TOPS-10 style (which is also RSTS style) not VMS/RSX style. The VMS and RSX ones are somewhat like Cterm except that they more closely copy the actual set of I/O operations implemented in the terminal drivers of those two. You can find implementations of several of these in DECnet/Linux -- if you're having trouble with the standard one on VMS, you might be able to lift the Linux one and port it to VMS... paul -----Original Message----- From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On Behalf Of gerry77 at mail.com Sent: Wednesday, August 31, 2011 10:13 PM To: hecnet at Update.UU.SE Subject: Re: [HECnet] DECnet performance problems (on TOPS-10) On Wed, 31 Aug 2011 15:18:25 +0200, you wrote: SET HOST from a VMS system implies you use CTERM. CTERM is a horrible protocol, and only really works fine between VMS systems. Using it to connect to anything else usually means you can expect all kind of behavior and performance problems. It might be something else, but I would suspect CTERM as such first and foremost. I think you are quite right: I've just tried a connection using the old RTERM protocol (SET HOST/APP=R) and it works very fine! :-O I'll try anyway to fiddle with a couple of emulator parameters, but... BTW, since here there are many people that run things like RSX and such, I'll take advantage of this message to ask: did you ever had problems using RTERM? I ask this because I've noticed that SET HOST/APP=R on OpenVMS Alpha V8.3 bombs with an ACCVIO whenever I try to connect to RSTS or TOPS-10, but not when I use it with VAX/VMS V4.7 (or OpenVMS V7.2 VAX, for that matter). In other words, I have not so much different versions to try with, but it seems that RTPAD.EXE does not like very old RTERM implementations like ones found in OSes different from VMS, no matter how much vintage it can be. The system showing this problem is not mine, but at a quick check it appears to have every possible patch kit installed. This is the relevant image data: image name: "RTPAD" image file identification: "X-10" image file build identification: "XBC4-0080060000" link date/time: 29-JUN-2006 18:18:54.69 linker identification: "A13-03" And this is the error: %SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=0000517000008AA4, PC=0000517000008AA4, PS=0000001B Improperly handled condition, image exit forced. Signal arguments: Number = 0000000000000005 Name = 000000000000000C 0000000000010000 0000517000008AA4 0000517000008AA4 000000000000001B Register dump: R0 = 0000000000000003 R1 = 0000000000008F50 R2 = 0000000000008134 R3 = 0000000000034804 R4 = 000000007FFCF814 R5 = 0000000000009AEF R6 = 0000000000000000 R7 = 0000000000000001 R8 = 000000000003481B R9 = 000000007FF9DDF0 R10 = 000000007FFA4F28 R11 = 000000007FFCDC18 R12 = 000000007FFCDA98 R13 = 0000000000005170 R14 = 0000000000008E28 R15 = 0000000000008E28 R16 = 0000000002A83089 R17 = 0000000000200000 R18 = FFFFFFFF80903360 R19 = FFFFFFFF820EFA56 R20 = 0000000000000002 R21 = 0000000000000000 R22 = 000000007AD69A54 R23 = FFFFFFFF801877D0 R24 = 0000000000000001 R25 = 0000000000000003 R26 = FFFFFFFF8018780C R27 = 000000007AD69A54 R28 = 0000000000000006 R29 = 000000007AD69A60 SP = 000000007AD69A30 PC = 0000517000008AA4 PS = 300000000000001B Any info about it? Thanks, G.

13 years, 8 months

DECnet performance problems (on TOPS-10)

by Paul_Koning＠Dell.com

The pre-cterm remote terminal protocol is not actually a single protocol, but rather a set of protocols specific to the destination OS. That's why it works well: instead of trying to fit the OS native terminal behavior into a VMS-centric model as is done in Cterm, the rterm protocols simply express a terminal behavior that matches what the OS wants. The reason Cterm is so bad for TOPS-10 is that there is really nothing that matches -- just for starters, TOPS-10 echoes differently. The reason you may see failure for one destination OS type and not others is that you're going through completely different code, because you have a different rterm flavor. There are four flavors (according to the RSTS code I have that implements them): VMS, RSX, RSTS, and TOPS-20. If I remember right, TOPS-10 uses the RSTS flavor, and RT-11 perhaps does as well. The RSTS flavor is line oriented (as is Cterm) but echoes TOPS-10 style (which is also RSTS style) not VMS/RSX style. The VMS and RSX ones are somewhat like Cterm except that they more closely copy the actual set of I/O operations implemented in the terminal drivers of those two. You can find implementations of several of these in DECnet/Linux -- if you're having trouble with the standard one on VMS, you might be able to lift the Linux one and port it to VMS... paul -----Original Message----- From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On Behalf Of gerry77 at mail.com Sent: Wednesday, August 31, 2011 10:13 PM To: hecnet at Update.UU.SE Subject: Re: [HECnet] DECnet performance problems (on TOPS-10) On Wed, 31 Aug 2011 15:18:25 +0200, you wrote: SET HOST from a VMS system implies you use CTERM. CTERM is a horrible protocol, and only really works fine between VMS systems. Using it to connect to anything else usually means you can expect all kind of behavior and performance problems. It might be something else, but I would suspect CTERM as such first and foremost. I think you are quite right: I've just tried a connection using the old RTERM protocol (SET HOST/APP=R) and it works very fine! :-O I'll try anyway to fiddle with a couple of emulator parameters, but... BTW, since here there are many people that run things like RSX and such, I'll take advantage of this message to ask: did you ever had problems using RTERM? I ask this because I've noticed that SET HOST/APP=R on OpenVMS Alpha V8.3 bombs with an ACCVIO whenever I try to connect to RSTS or TOPS-10, but not when I use it with VAX/VMS V4.7 (or OpenVMS V7.2 VAX, for that matter). In other words, I have not so much different versions to try with, but it seems that RTPAD.EXE does not like very old RTERM implementations like ones found in OSes different from VMS, no matter how much vintage it can be. The system showing this problem is not mine, but at a quick check it appears to have every possible patch kit installed. This is the relevant image data: image name: "RTPAD" image file identification: "X-10" image file build identification: "XBC4-0080060000" link date/time: 29-JUN-2006 18:18:54.69 linker identification: "A13-03" And this is the error: %SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=0000517000008AA4, PC=0000517000008AA4, PS=0000001B Improperly handled condition, image exit forced. Signal arguments: Number = 0000000000000005 Name = 000000000000000C 0000000000010000 0000517000008AA4 0000517000008AA4 000000000000001B Register dump: R0 = 0000000000000003 R1 = 0000000000008F50 R2 = 0000000000008134 R3 = 0000000000034804 R4 = 000000007FFCF814 R5 = 0000000000009AEF R6 = 0000000000000000 R7 = 0000000000000001 R8 = 000000000003481B R9 = 000000007FF9DDF0 R10 = 000000007FFA4F28 R11 = 000000007FFCDC18 R12 = 000000007FFCDA98 R13 = 0000000000005170 R14 = 0000000000008E28 R15 = 0000000000008E28 R16 = 0000000002A83089 R17 = 0000000000200000 R18 = FFFFFFFF80903360 R19 = FFFFFFFF820EFA56 R20 = 0000000000000002 R21 = 0000000000000000 R22 = 000000007AD69A54 R23 = FFFFFFFF801877D0 R24 = 0000000000000001 R25 = 0000000000000003 R26 = FFFFFFFF8018780C R27 = 000000007AD69A54 R28 = 0000000000000006 R29 = 000000007AD69A60 SP = 000000007AD69A30 PC = 0000517000008AA4 PS = 300000000000001B Any info about it? Thanks, G.

13 years, 8 months

Anyone with a spare MMJ cable + RS232 adapter?

by schwa.phil＠gmail.com

On 11-08-30 05:25 PM, Sampsa Laine wrote: MMJ - MMJ RX, TX, GND 6' (2m) long enough? Qty 4 Done. I actually wired all 6 lines as crossover cables because it was easier. Hack them up at will. Send me a snail-mail address, and I'll give you the total after I pop them in the post tomorrow. How fast a service are you willing to spring for? Phil

13 years, 8 months

DECUS archive

by mark＠wickensonline.co.uk

Guys, My AlphaServer 1000A node SLAVE now has the DECUS archive from D I G I A T E R on the FAL$SERVER account, so you can access it directly from SLAVE:: For example: $ dir slave:: Directory SLAVE::DECUS$:[DECUS] 00BVMSLT.DIR;1 01AVMSLT.DIR;1 02AVMSLT.DIR;1 02BVMSLT.DIR;1 03AVMSLT.DIR;1 03BVMSLT.DIR;1 04AVMSLT.DIR;1 04BVMSLT.DIR;1 05AVMSLT.DIR;1 05BVMSLT.DIR;1 06AVMSLT.DIR;1 07AVMSLT.DIR;1 90AVAXLT.DIR;1 91AVAXLT.DIR;1 91BVAXLT.DIR;1 92AVAXLT.DIR;1 92BVAXLT.DIR;1 93AVMSLT.DIR;1 93BVMSLT.DIR;1 94AVMSLT.DIR;1 94BVMSLT.DIR;1 95AVMSLT.DIR;1 96BVMSLT.DIR;1 97AVMSLT.DIR;1 97BVMSLT.DIR;1 98AVMSLT.DIR;1 98BVMSLT.DIR;1 99AVMSLT.DIR;1 99BVMSLT.DIR;1 CIRCUIT_SIMULATOR.DIR;1 DECUSC.DIR;1 DECUSERVE.DIR;1 DECUSLIB.DIR;1 DECUSUTL.DIR;1 DECWINDOWS.DIR;1 EXTRA.DIR;1 FREEWAREV10.DIR;1 FREEWAREV20.DIR;1 FREEWAREV30.DIR;1 FREEWAREV40.DIR;1 FREEWAREV50.DIR;1 FREEWAREV60.DIR;1 FREEWAREV70.DIR;1 FREEWAREV80.DIR;1 INFO-VAX.DIR;1 ISO9660.DIR;1 LT86A.DIR;1 LT87A.DIR;1 LT89A.DIR;1 LT89B1.DIR;1 LT89B2.DIR;1 LT90A.DIR;1 LT90B1.DIR;1 LT90B2.DIR;1 LT91A.DIR;1 LT91B.DIR;1 LT92A.DIR;1 LT92B.DIR;1 LT93A.DIR;1 LT93B.DIR;1 LT94A.DIR;1 LT94B.DIR;1 LT95A.DIR;1 LT95B.DIR;1 MN.DIR;1 NCSA.DIR;1 NET.DIR;1 PCSI_KITS.DIR;1 PERL.DIR;1 $ dir slave::[.91AVAXLT] Directory SLAVE::DECUS$:[DECUS.91AVAXLT] 000README.TXT;1 0README.ALSO;1 AAAREADME.91A_LT;1 AAAREADME.91A_VAX;1 ABSTRACT.DOC;1 ABSVLT91.TXT;1 BUILDREAD.COM;1 BUILDREADL.COM;1 BUILDREADV.COM;1 CATALOG.DIR;1 COMPRESSIONS.TXT;1 COPYTAPES.COM;1 COPYTREES.COM;1 DECOMPRESS_ALL.COM;1 DEFAULT.EXE;1 DELETE_ALL.COM;1 GRAND_TOTS.COM;1 LOGIN.COM;1 OLD_AAAREADME.DIR;1 OLD_B.COM;1 REVIEWS.DIR;1 SUBMIT.TXT;1 TAPECOPY.HISTORY;1 TCTREE.LIS;1 TCTREEPROC.LIS;1 TEMPLATE.RNO;1 TOOLS.DIR;1 TREE.SEQ;1 TREESIZE.COM;1 TREETAPE.COM;1 VAXLT91ATPE.DOC;1 VLT91A.TXT;1 VLT91A.XREF;1 SLAVE isn't currently on 24/7 - normally I run it from midday to midnight BST, approximately. Feedback welcome! Regards, Mark.

13 years, 8 months

Changing console from serial to graphics on AS800 from VMS?

by sampsa＠mac.com

I HAVE a DECserver - 200, with DB25 ports lol :) Sampsa On 31 Aug 2011, at 14:26, Mark Wickens wrote: On 31/08/11 13:54, hvlems at zonnet.nl wrote: The converter from DB9 (PC style) to MMJ is called an H8571-J The -J is very important since other H8571's will do different things. Hans ------Origineel bericht------ Van: Sampsa Laine Afzender: owner-hecnet at Update.UU.SE Aan: hecnet at Update.UU.SE Beantwoorden: hecnet at Update.UU.SE Onderwerp: [HECnet] Changing console from serial to graphics on AS800 from VMS? Verzonden: 31 augustus 2011 14:24 Is it possible to change the OPA0: (on reboot, so it sticks) from serial to graphics on an AS800 through VMS? Sampsa Verzonden vanaf mijn draadloze BlackBerry -toestel Sampsa, consider investing in a DECserver... then you can have all your consoles available over ethernet.

13 years, 8 months

DECnet performance problems (on TOPS-10)

by hvlems＠zonnet.nl

What I thought was this: all DECnet parameters are set to default values. The other DECnet nodes probably also. Which is good because messing with most parameters won't improve performance. The next step is that DECnet is well implemented across DEC o/s'es. TOPS is no exception to that rule so it's not the first place to start investigating. Though do try alt.sys.pdp10 or alt.sys.folklore. So what remains is the host system. Suspect #1 is pcap. Is the package recent? I can't think of parameters to change here. On linux you might check network security parameters and whether pcap needs tweaking of linux paramters. ------Origineel bericht------ Van: gerry77 at mail.com Afzender: owner-hecnet at Update.UU.SE Aan: hecnet at Update.UU.SE Beantwoorden: hecnet at Update.UU.SE Onderwerp: Re: [HECnet] DECnet performance problems (on TOPS-10) Verzonden: 31 augustus 2011 15:05 On Wed, 31 Aug 2011 06:54:12 +0000, you wrote: The paramters you posted are aal defaults except for the node name aand address, right? Yes. Anyway I tried also with some modified parameters we use on other nodes, namely executor block size and buffer size, but nothing changed. The "problem" is that most of those parameters are "embedded" into the monitor, so to change them the monitor has to be rebuilt every time. Non particularly diffucult, but quite time-consuming. Moreover they have names like %RTBSZ and %DLBSZ, and is not always so easy to discover which NCP/DECnet parameter corresponds to which internal monitor variable... :) - have a look at the exec, line and circuit counters In my previous message I forgot to mention that all the error-related counters are at zero, so I thought not to paste them here. - examine the host IP characteristics and (if present) those of the winpcap package installed on the host Specifically, what characteristics are you thinking about? I have no idea about whatever I could check at host and pcap (on Linux) level. Thanks for your time, :) G. Verzonden vanaf mijn draadloze BlackBerry -toestel

13 years, 8 months

DECnet performance problems (on TOPS-10)

by hvlems＠zonnet.nl

You have a DECnet key (not a PAK!) for DECnet on a pre V5.0 VMS system, rigjt? ------Origineel bericht------ Van: gerry77 at mail.com Afzender: owner-hecnet at Update.UU.SE Aan: hecnet at Update.UU.SE Beantwoorden: hecnet at Update.UU.SE Onderwerp: Re: [HECnet] DECnet performance problems (on TOPS-10) Verzonden: 31 augustus 2011 15:05 On Wed, 31 Aug 2011 10:58:53 +0200, you wrote: On PDP:s (RSX11-Mplus) and VAX/VMS, C-TERM (eg. SET HOST) is usually sort of "sluggy". Oh well, it depends! :) Usually our DECnet connections are very smooth, at least between VMS nodes; VAX, Alpha and Itanium, no matter which VMS version. In Italy we say that it glides (so well that) it seems greasy. :P LAT (eg SET TERM/LAT) is a protocol developed to cope with terminals character by character communication at first hand! I don't know if LAT is available to your 10-system, though, but if it is, use that for terminal traffic as far as possible. Yes, LAT is available, but I have other problems with it. As soon as I'll be able to tackle it, I'll try also that path. TOPS-10 is a though "game"... Besides, what version of VMS, on what hardware? I've heard birds singing of problems combining modern VMS-DECnet with older implementations in other machines... By now I have tried VMS 7.2 on VAX and 8.3 on Alpha. Now I'm waiting to try VMS 4.7 on the same Ethernet segment as the TOPS-10 node, then I will report on results! :) Thanks, G. Verzonden vanaf mijn draadloze BlackBerry -toestel

13 years, 8 months

Changing console from serial to graphics on AS800 from VMS?

by mark＠wickensonline.co.uk

On 31/08/11 13:54, hvlems at zonnet.nl wrote: The converter from DB9 (PC style) to MMJ is called an H8571-J The -J is very important since other H8571's will do different things. Hans ------Origineel bericht------ Van: Sampsa Laine Afzender: owner-hecnet at Update.UU.SE Aan: hecnet at Update.UU.SE Beantwoorden: hecnet at Update.UU.SE Onderwerp: [HECnet] Changing console from serial to graphics on AS800 from VMS? Verzonden: 31 augustus 2011 14:24 Is it possible to change the OPA0: (on reboot, so it sticks) from serial to graphics on an AS800 through VMS? Sampsa Verzonden vanaf mijn draadloze BlackBerry -toestel Sampsa, consider investing in a DECserver... then you can have all your consoles available over ethernet.

13 years, 8 months

DECnet performance problems (on TOPS-10)

by bqt＠softjar.se

(I'll top-post this one...) SET HOST from a VMS system implies you use CTERM. CTERM is a horrible protocol, and only really works fine between VMS systems. Using it to connect to anything else usually means you can expect all kind of behavior and performance problems. It might be something else, but I would suspect CTERM as such first and foremost. Johnny On 2011-08-31 02:20, gerry77 at mail.com wrote: Dear All, I'm in need of some suggestions to improve DECnet performance of a TOPS-10 monitor running on KLH10. I'd like a lot to hear from you even if you do not know or run that specific system or configuration. When used locally at the CTY (Console terminal), the system response is very fast and smooth, so I'd tend to exclude emulator problems and such. Instead, when used from a remotely connected terminal (e.g. from VMS via SET HOST), it appears very sluggish: single character echo is very poor and output from the system fits and starts all the time. Output from programs is very fast for some hundreds lines, then stops briefly, then again appears almost instantly, and so on, like in bursts. Anyway, while the overall bulk data performance can be considered absolutely acceptable, the char-by-char responsiveness, e.g. typing interactive commands at the monitor prompt, is horrible: I type "D", "I", "R", and nothing happens, then I type "E" and "DIRE" suddenly appears... The TOPS-10 system has plenty of free core (both virtual and physical), free disk space, and free CPU, so it shouldn't be an OS issue in itself. Probably the most likely cause of this problem lies in some DECnet parameter that needs some tweaking, but I do not know which one... Have you ever had any experience with some ancient Digital OS having poor DECnet performance as described? And how did you solve that, or at least improve it? Here are some (maybe) interesting bits: Request # 1; Show Executor Node Characteristics Completed Executor Node = 1.1010 (DIECI) Identification = DECnet-10 Version 4.0 Management Version = 4.0.0 Loop Count = 1 Loop Length = 127 Loop With = Mixed Incoming Timer = 30 Outgoing Timer = 60 NSP Version = 4.0.0 Maximum Links = 65535 Delay Factor = 48 Delay Weight = 10 Inactivity Timer = 120 Retransmit Factor = 10 Routing Version = 2.0.0 Type = Nonrouting IV Routing Timer = 600 Broadcast Routing Timer = 40 Maximum Address = 1023 Maximum Circuits = 20 Maximum Cost = 100 Maximum Hops = 16 Maximum Visits = 20 Maximum Broadcast Nonrouters = 64 Maximum Broadcast Routers = 32 Maximum Buffers = 80 Buffer Size = 576 Segment Buffer Size = 576 Request # 5; Show Line Characteristics Completed Line = ETH-0 Receive Buffers = 6 Controller = Normal Protocol = Ethernet Hardware Address = AA 00 04 00 F2 07 Receive buffer size = 576 Request # 7; Show Circuit Characteristics Completed Circuit = ETH-0 Cost = 1 Maximum Routers = 16 Router Priority = 5 Hello Timer = 90 Type = Ethernet I'll be grateful for any suggestion! :) G. -- Johnny Billquist || "I'm on a bus || on a psychedelic trip email: bqt at softjar.se || Reading murder books pdp is alive! || tryin' to stay hip" - B. Idol

13 years, 8 months

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

HECnet