<<< TEMBEL::OFFICE$DISK:[NOTES$LIBRARY]VAX.NOTE;1 >>> -< VAX9000 Systems >- ================================================================================ Note 1.0 Welcome ... 4 replies BIGUN::KEOGH "I feel better than James Brown" 5 lines 6-AUG-1991 00:12 -------------------------------------------------------------------------------- Welcome to VAX NOTES This conference is for discussions relating to VAX9000 Systems ================================================================================ Note 1.1 Welcome ... 1 of 4 VAXNET::BEATHJ2 28 lines 20-AUG-1991 18:28 -< VAX 9000 >- -------------------------------------------------------------------------------- Not being a Notes user this is a first time effort. Are there any other 9000 users out there. I was talking to a system manager from the Vic SEC and he was saying that they support 900+ ALL-IN-1 and Oracle users on 3 8820's and 2 6420's in a cluster. They run 20/20, Notes and a few other products we don't. This is similar to our site(Health, Housing and Community Services Canberra), in terms of software(All-in-1 & Oracle) used. We currently have 2 9410's and one 6430 which should be capable of handling these numbers of users. If anyone out there knows the pitfalls that need to be watched in ALL-IN-1(we know Oracle has its problems) can they indicate please. We are aware of a number of potential problem areas, e.g. Shared areas, dafs etc. The 9000's are still being bedded in but i/o performance issues are obvious. Our onsite digit is doing his best, we hope. What other methods will improve the performance. We are disk shadowing and striping with some results but the disk farm tends to dissappear at a great rate of knots as does CPU and Memory Capacity. I'm afraid we can't get rid of ALL-IN-1 or Oracle(we all know what management are like) so we are stuck with both. Ideas are welcome. JB ================================================================================ Note 1.2 Welcome ... 2 of 4 VAXNET::SZEWCZYKR15 26 lines 21-AUG-1991 13:22 -< All-In-1 Requirements >- -------------------------------------------------------------------------------- We are currently looking at implementing All-In-1 at our site. The requirements for running classic All-In-1 and All-In-1 Desktop as determined by enquiring at various running sites as well as benchmark figures from DEC and published information about tuning for All-In-1 appear to be the following: CPU capacity required: About 1 VUP per 15 classic All-In-1 users (logged-in). This can vary between 12 to 22 depending on the source of information. It drops as you go from uniprocessor to multiprocessor operation. All-In-1 Desktop allows 3 times the number of users i.e. 45 users per VUP. Memory: About 1 MB (WSQUO=2048) per user. DEC figures indicate around 0.75 MB. A tuning article I read states 1 to 1.5 MB per logged in user. In addition you require some memory for VMS itself 10-30 MB depending on your system. Disk I/O: It would seem that disk I/O is not usually a problem in an All-In-1 system. One site I contacted quoted around 10 I/O per sec for 100 logged in users from VPA figures. They were offloading some of their wordprocessing to PCs so the figure would normally be higher I guess. At our site we are planning to separate the All-In-1 system from those running Oracle and technical applications. They would be on separate CPUs of our VAXcluster. I would guess that your performance problems would be a result of mixing the two. ================================================================================ Note 1.3 Welcome ... 3 of 4 VAXNET::KARAGIANNA20 17 lines 22-AUG-1991 16:23 -< VAX9000 improving the I/O >- -------------------------------------------------------------------------------- If YOu have identified I/O as the issue, then why do you not look at i/o express - Dynamic Data Caching for improving the I/O throughput of the system. I/O Express is available from Network Computer Software, Sydney 2112322. They are downstairs at the CSO area. I/O Express will use available memory on the VAX9000 (which I'm sure you have plenty of, then get that memory to do some work on your behalf. Arthur Karagiannis ================================================================================ Note 1.4 Welcome ... 4 of 4 PENCEL::DEMO 2 lines 23-AUG-1991 09:12 -------------------------------------------------------------------------------- Did the Depts' choice to use multiple ALL-IN-1 areas instead of 1 make any difference?  ================================================================================ Note 2.0 Bookreader on charcell terminal? 8 replies TEMBEL::BEGGJ17 6 lines 19-AUG-1991 15:35 -------------------------------------------------------------------------------- Can anyone tell me when DECwindows Bookreader will support character-cell terminals and allow pages to be printed (ie other than using screen dump)? Jeremy Begg VSM Software Services ================================================================================ Note 2.1 Bookreader on charcell terminal? 1 of 8 TEMBEL::MURRAYB23 2 lines 19-AUG-1991 15:51 -< Never >- -------------------------------------------------------------------------------- Never of course -- You wouldn't expect Digital to copy IBM and make its documentation accessible would you? ================================================================================ Note 2.2 Bookreader on charcell terminal? 2 of 8 TEMBEL::BUTLERC22 "Chris Butler, NZ Ministry of Forestry" 15 lines 20-AUG-1991 08:05 -< Bookreader 2 ASCII converter >- -------------------------------------------------------------------------------- Hi Jeremy, There is a public domain utility (posted to USENET a while back) which converts bookreader format to ASCII, called (wait for it) BOOKREADER_2_ASCII. It is pretty crude & rude, as page formatting is not good & (obviously) diagrams are missing. It can accessed via anonymous ftp from black.cerritos.edu (130.150.200.21), which is the VMSNET sources archive. Cheers, Chris Butler NZ Forest Research Institute ================================================================================ Note 2.3 Bookreader on charcell terminal? 3 of 8 TEMBEL::LANCASTERR1 "Robert (excuse the typing) Lancaster" 4 lines 20-AUG-1991 09:44 -< DECUS library >- -------------------------------------------------------------------------------- I hope this is not the utility DEC had banned from the DECUS library They seem to get upset if anyone writes usefull software. ================================================================================ Note 2.4 Bookreader on charcell terminal? 4 of 8 VAXNET::KEOGH "I feel better than James Brown" 7 lines 20-AUG-1991 11:28 -------------------------------------------------------------------------------- A couple of points to consider ... Consider the copyright implications of copying/converting/distributing/printing Digital's documentation. Using a copy of the documentation without figures, colour, the use of multiple fonts etc. sometimes can be hazardous to your understanding! ================================================================================ Note 2.5 Bookreader on charcell terminal? 5 of 8 TEMBEL::BEGGJ17 "Jeremy Begg" 26 lines 20-AUG-1991 15:05 -< The point is, I have been told it was coming. >- -------------------------------------------------------------------------------- >Consider the copyright implications of copying/converting/distributing/printing >Digital's documentation. I'm not so sure that's a valid argument, Pat. The documentation is licensed to the system, which in many cases includes utilities to convert document (or other file) formats at will. Surely making a copy for one's own use, on the same system, of something already paid for, is quite legitamate. Anyway, I was told by one of my local Digits that features such as those asked for in my original message were coming. >Using a copy of the documentation without figures, colour, the use of multiple >fonts etc. sometimes can be hazardous to your understanding! I think most people would be happy with just about anything! Sensible use of things like underline, bold and reverse video should cover most situations. Diagrams etc are of course a more difficult problem, but the kind of "books" people want to be able to read on a character cell terminal often don't contain them, for example the entire VMS programming bookshelf. Jeremy Begg VSM Software Services jeremy@vsm.com.au ================================================================================ Note 2.6 Bookreader on charcell terminal? 6 of 8 TEMBEL::KEOGH "I feel better than James Brown" 22 lines 20-AUG-1991 15:21 -------------------------------------------------------------------------------- Jeremy, > I'm not so sure that's a valid argument, Pat. The documentation is > licensed to the system, which in many cases includes utilities to > convert document (or other file) formats at will. Surely making a copy > for one's own use, on the same system, of something already paid for, > is quite legitamate. I am not here to argue the moral rightness (or otherwise) of copyright law. I merely meant that - Digital's documentation is subject to copyright law - It is a breach of copyright law to reproduce copyright materal by any means, including electronic, without the express permission of the copyright holder. At some stage in the future, Digital may authorize you to reproduce some copyright material. Then you would not be in breach of the law! Your point that we provide utilities that could be used to copy this material is just like saying that the owners manual that comes with your photocopier isn't protected by copyright because it comes with a means for reproducing it... ================================================================================ Note 2.7 Bookreader on charcell terminal? 7 of 8 TEMBEL::LANCASTERR1 "Robert (excuse the typing) Lancaster" 4 lines 20-AUG-1991 15:40 -< Copy right law >- -------------------------------------------------------------------------------- I thought coping small sections of a copyrighted document was OK. Not quite sure if 24 consecative lines is too much though. ================================================================================ Note 2.8 Bookreader on charcell terminal? 8 of 8 VAXNET::GORHAMM24 5 lines 22-AUG-1991 09:19 -< Bookreader Reply - Digital Central Engineering >- -------------------------------------------------------------------------------- Bookreader Fans. Digital is working on a solution for character cell support. The solution will involve a degredation in performance. We are also working on printing capabilities for the next bookreader release.  ================================================================================ Note 3.0 Phase II VOLSHAD limits 4 replies TEMBEL::MANDENOG29 "Graham Mandeno - ALCAN" 11 lines 19-AUG-1991 16:23 -------------------------------------------------------------------------------- The Phase II Volume Shadowing manual states (or implies) that there is no limit to the number of shadow sets supported on a VAXcluster. However, an article buried deep in DSIN states that this number is limited to a maximum of 32!! (no reason given) Can some informed Digit please tell me if there is a limit, and if so, what it is, and why? Thanks, Graham Mandeno Alcan Australia Limited ================================================================================ Note 3.1 Phase II VOLSHAD limits 1 of 4 TEMBEL::HARRY27 5 lines 20-AUG-1991 11:16 -< Thats all they have tested >- -------------------------------------------------------------------------------- I believe the answer given ysterday by the relevant digit was that they have only tested 32 so thats all they say they will support !! ================================================================================ Note 3.2 Phase II VOLSHAD limits 2 of 4 TEMBEL::SOMMERVILB19 "Brian Sommerville - BT Aust." 5 lines 20-AUG-1991 11:23 -------------------------------------------------------------------------------- this is also spelled out in the one (of a number) of cover letters shipped with 5.4-x. brian ================================================================================ Note 3.3 Phase II VOLSHAD limits 3 of 4 VAXNET::GORHAMM24 2 lines 22-AUG-1991 09:21 -< Shadow Set Limit - Digital Central Engineering >- -------------------------------------------------------------------------------- The shadow set limitation is based on conservative qualification testing. Good news, it will expand to around 75 with VMS Version 5.5 ================================================================================ Note 3.4 Phase II VOLSHAD limits 4 of 4 TEMBEL::WERTHD2 11 lines 22-AUG-1991 18:09 -< shadowing limits >- -------------------------------------------------------------------------------- I believe the SPD is the final authority ... also the limit is based on expected performance ... so you can have many but may not like the performance. One of the things you want to watch is the number of copies which may be started by any node depending on how the node is connected. Zero is best for NI if any other interconnect is available. Don  ================================================================================ Note 4.0 DECram memory usage 3 replies VAXNET::GRANTA15 4 lines 19-AUG-1991 16:38 -------------------------------------------------------------------------------- Does DECram use paged or nonpaged pool (personally I would prefer nonpaged pool). Thanks, Alistair. ================================================================================ Note 4.1 DECram memory usage 1 of 3 TEMBEL::SOMMERVILB19 "Brian Sommerville - BT Aust." 6 lines 20-AUG-1991 11:25 -------------------------------------------------------------------------------- I don't know but my guess in NON-paged. Doesn't make sense to generate I/O to save I/O...... brian ================================================================================ Note 4.2 DECram memory usage 2 of 3 TEMBEL::GRANTA15 10 lines 20-AUG-1991 13:44 -< Why use paged pool >- -------------------------------------------------------------------------------- >I don't know but my guess in NON-paged. Doesn't make sense to generate >I/O to save I/O...... One place where this would be useful is in a large file/disk which only has a small portion used at one time. The rest could be paged out to save memory, whilst still obtaining the benefits of a ram disk. If a large portion of the disk will be accessed frequently, configure the system so it doesn't page. Alistair. ================================================================================ Note 4.3 DECram memory usage 3 of 3 VAXNET::GORHAMM24 1 line 22-AUG-1991 09:23 -< DECram memory usage - Digital Central Engineering >- -------------------------------------------------------------------------------- The developer says he uses memory from freelist.  ================================================================================ Note 5.0 DECnet on the CIXCD 3 replies TEMBEL::MANDENOG29 "Graham Mandeno - ALCAN" 12 lines 19-AUG-1991 16:57 -------------------------------------------------------------------------------- Having just installed a new, fully XMI-based 6000-420, and configured the local DECnet database, I was delighted to discover that VMS crashed in PADRIVER every time I brought up the network. CSC response was "sorry, DECnet is NOT supported on the CIXCD". Can somebody please tell me: a) why not? and b) when will it be? Thanks, Graham Mandeno Alcan Australia ================================================================================ Note 5.1 DECnet on the CIXCD 1 of 3 VAXNET::ROBINSONM31 3 lines 19-AUG-1991 18:22 -< DECnet on CIXCD OK >- -------------------------------------------------------------------------------- I have an XMI 6510 running DECnet fine. Not a problem! The only CIXCD problem is related to SPM trace which will not run on a combination of CIXCD and 6510 (ie clustered 6510). ================================================================================ Note 5.2 DECnet on the CIXCD 2 of 3 TEMBEL::MANDENOG29 "Graham Mandeno - ALCAN" 3 lines 19-AUG-1991 18:31 -< are you really using the CIXCD? >- -------------------------------------------------------------------------------- Don Werth (VAXcluster Systems Directions) has just confirmed that there IS such a problem with the CIXCD. Are you sure you are running DECnet over the CI, not just over Ethernet? ================================================================================ Note 5.3 DECnet on the CIXCD 3 of 3 VAXNET::GORHAMM24 6 lines 22-AUG-1991 09:26 -< DECnet on CIXCD - Digital Central Engineering >- -------------------------------------------------------------------------------- You are correct, DECnet is not supported on the CIXCD. There are no plans to support DECnet on CIXCD. Will take your (and other requests from this conference) back to Central Engineering for consideration.  ================================================================================ Note 6.0 ALLIN1 (X400) Mail 2 replies VAXNET::GARNERR30 "Whale Oil Beef Hooked" 1 line 19-AUG-1991 18:22 -------------------------------------------------------------------------------- Anyone here using Allin1 Mail (Not ALL-IN-1 Classic) ? ================================================================================ Note 6.1 ALLIN1 (X400) Mail 1 of 2 TEMBEL::HAWLEYD13 "Tektronix EDP Manager" 2 lines 20-AUG-1991 08:14 -< Info please >- -------------------------------------------------------------------------------- We are planning to use this product very soon. Any info would be welcome. ================================================================================ Note 6.2 ALLIN1 (X400) Mail 2 of 2 VAXNET::GARNERR30 "Whale Oil Beef Hooked" 2 lines 20-AUG-1991 13:11 -< Redirect ... >- -------------------------------------------------------------------------------- On reflection, I think my topic is out of place here ... lets redirect to the OFFICE_AUTOMATION conference  ================================================================================ Note 7.0 MEMORY HOGGING IDLE PROCESSES 5 replies TEMBEL::SELDONT5 8 lines 19-AUG-1991 20:52 -------------------------------------------------------------------------------- AT ONE OF THE SESSION THIS AFTERNOON ONE OF SPEAKER SPOKE ABOUT (IN FURTHER RELEASE OF VAX) A FEATURE TO TRIM IDLE PROCESSES MEMORY BACK BEFORE WE RUN OUT (FREELIM I THINK) WHAT IS THE BEST CURRENT METHOD OF RETREIVING THIS MEMORY WITHOUT STOPPING THESE PROCESS THANK YOU TED ================================================================================ Note 7.1 MEMORY HOGGING IDLE PROCESSES 1 of 5 TEMBEL::MANDENOG29 "Graham Mandeno - ALCAN" 18 lines 20-AUG-1991 10:28 -< Use a KAST with $ADJWSL >- -------------------------------------------------------------------------------- We use a bit of privileged code which queues a Kernel-mode AST to the idle process. The AST simply calls SYS$ADJWSL to trim the process' working set list to the desired size, returning the pages to the free (or modified) list. This is a relatively simple technique, and is exactly the one used by SWAPPER's trim routine. One side-effect though, is that when the process resumes activity it needs to fault heavily to extend its working set again, and this can cause excruciating slowness until things get back to normal. One way around this is to call the same routine again, this time setting the working set list size back to the previous value of PPGCNT. I understand the V5.4-3 method calls the MMG$ routines to remove pages from the WS without reducing the WSL size. Hope this helps Graham ================================================================================ Note 7.2 MEMORY HOGGING IDLE PROCESSES 2 of 5 VAXNET::GARNERR30 "Whale Oil Beef Hooked" 12 lines 20-AUG-1991 11:25 -< Other choices >- -------------------------------------------------------------------------------- Another possibility is to increase SWPOUTPAGCNT (spelling ?) to something larger than the normal WSQUOTA. This prevents the Swapper from trimming processes back below quota before swapping them back. As a result the idle processes get swapped out before active processes get trimmed. This assumes that WSQUOTA really does represent an adequate working set size for each user, so you probably want to tune these first. You also need to look at increasing FREELIM - my rule of thumb is to keep 2 WSQUOTAs worth of pages on the Free Page List. This ensures that there are enough pages free to build an inswapped process. You also need to increase the Swap File etc. etc., and keep a close eye on the system when you introduce these changes. ================================================================================ Note 7.3 MEMORY HOGGING IDLE PROCESSES 3 of 5 VAXNET::KEOGH "I feel better than James Brown" 3 lines 20-AUG-1991 11:34 -------------------------------------------------------------------------------- Graham, You're right. 5.4-3 will do the trimming without changing the working set size. This provides a real advantage. ================================================================================ Note 7.4 MEMORY HOGGING IDLE PROCESSES 4 of 5 VAXNET::KEOGH "I feel better than James Brown" 2 lines 20-AUG-1991 11:35 -------------------------------------------------------------------------------- Come and see it all work on the VMS stand. We have a 10Mb VS31000 that is running DECwindows/Motif and is a cluster member, still with adequate memory. ================================================================================ Note 7.5 MEMORY HOGGING IDLE PROCESSES 5 of 5 TEMBEL::GRANTA15 5 lines 22-AUG-1991 10:29 -< Memory trimming utility >- -------------------------------------------------------------------------------- I have placed a utility called MEMTRIM on the Aust. Symposium tape. This will proactively reclaim memory and is great for workstations. This is useful until 5.4-3 arrives. Alistair  ================================================================================ Note 8.0 User written loadable images 7 replies TEMBEL::DUFFJ8 "The Doc is [IN]" 25 lines 20-AUG-1991 08:57 -------------------------------------------------------------------------------- For Central Engineering (and the tee shirt)... If I write a user written loadable image that calls $GETUAI, it assembles and links fine. However, on booting the system I get an exception (the hex code escapes me for the moment). Once the machine is up, I find that my image is not loaded and, upon trying to track down the error message, it is not to be found on ANY message file on the system. I contacted Sydney CSC with this problem and we finally discovered that the cause is the fact that my routine calls $GETUAI. The code that contains $GETUAI has not been loaded at the time VMS tries to load user written system services. Question: Why does VMS not load user written loadable images last? This would allow us to use all system services, etc. If this is done for a reason why isn't it documented? Even better question, why does the act of loading (not executing) my routine cause an exception? Jim Duff (DUFF@DECUS.COM.AU) ================================================================================ Note 8.1 User written loadable images 1 of 7 TEMBEL::MANDENOG29 "Graham Mandeno - ALCAN" 12 lines 20-AUG-1991 10:46 -< $GETUAI is not in the exec >- -------------------------------------------------------------------------------- The answer is simple - $GETUAI is not in part of the loadable executive. The code is in the privileged shareable image SECURESHR (or SECURESHRP ?) and is accessed by the same mechanism as a user-written system service (i.e. it is loaded at image activation time) If you want to call this code from a loadable image you must activate the image yourself (and of course MUST be in process context). The technique is the same as that used by LIB$FIND_IMAGE_SYMBOL, but of course you can't call LIB$xxx because it also is in a shareable image. Good luck - it get's heavy! Graham ================================================================================ Note 8.2 User written loadable images 2 of 7 VAXNET::DUFFJ8 "The Doc is [IN]" 11 lines 20-AUG-1991 11:13 -< You're right. But what about the exception. >- -------------------------------------------------------------------------------- Yeah, sorry. I was trying to recall the problem from 6 months ago. However, it does not answer my question as to why the system coughs up an exception at load time. Surely loading an image should not generate the exception, and if it should the system should come up with an understandable error message, not the dreaded %NONAME-E-NOMSG, Message number xxxxxxxx. Jim Duff (DUFF@DECUS.COM.AU) ================================================================================ Note 8.3 User written loadable images 3 of 7 VAXNET::GARNERR30 "Whale Oil Beef Hooked" 9 lines 20-AUG-1991 11:30 -------------------------------------------------------------------------------- > generate the exception, and if it should the system should come up with > an understandable error message, not the dreaded > > %NONAME-E-NOMSG, Message number xxxxxxxx. > You can't blame VMS for the format of the message, surely! Theres a fair bit of VMS required to format messages correctly, and EXEC_INIT is working in a pretty sparse VMS at the time. Use F$MESSAGE when VMS comes up ... ================================================================================ Note 8.4 User written loadable images 4 of 7 VAXNET::DUFFJ8 "DUFF@DECUS.COM.AU" 10 lines 20-AUG-1991 12:42 -< I did use F$MESSAGE. >- -------------------------------------------------------------------------------- I did use F$MESSAGE. That is the problem. The message code generated (if it is a message code) is NOT contained in ANY message file supplied with VMS. All is I'm saying is that it would be nice if the the load routine generated a meaningful message so we don't have to waste my time and CSC's time finding the cause of the problem. Again, why does the load routine generate an exception on merely LOADING the routine? I would expect one on RUNNING it. Jim Duff. ================================================================================ Note 8.5 User written loadable images 5 of 7 TEMBEL::BEGGJ17 "Jeremy Begg" 16 lines 20-AUG-1991 15:15 -< Error caused by loader filling in inter-image ptrs >- -------------------------------------------------------------------------------- > However, it does not answer my question as to why the system coughs up > an exception at load time. Surely loading an image should not > generate the exception, and if it should the system should come up with > an understandable error message, not the dreaded ... I believe the reason is because the loadable image loader must resolve all inter-image references at load time, and fill in vector tables etc so that the links between images (ie calls etc) become valid. Part of the VMS V5 EXEC reorganisation was to ensure that modules in the executive could be modified without having to modify the whole, much like normal shareable images can be modified without requiring the main program be relinked. Jeremy Begg VSM Software Services jeremy@vsm.com.au ================================================================================ Note 8.6 User written loadable images 6 of 7 TEMBEL::DUFFJ8 "DUFF@DECUS.COM.AU" 6 lines 20-AUG-1991 16:04 -< Reasonable answer. Message needed. >- -------------------------------------------------------------------------------- The reason put forward by Jeremy in the last note sounds good to me. Could a Digit confirm this? If this is indeed the case, then the image loader should be able to output a meaningful error: it could save people days of trial and error. Central Engineering? Jim. ================================================================================ Note 8.7 User written loadable images 7 of 7 VAXNET::DUFFJ8 "DUFF@DECUS.COM.AU" 15 lines 20-AUG-1991 18:17 -< More info found in Internals & Data Structures 5.2 >- -------------------------------------------------------------------------------- Sorry I can't help it. The more I find out about this, the more it seems that the loader routine should be able to resolve references to sharable images. After posting the last note I have had a chance to look at the Internals & Data Structures manual. On page 941 it describes the SYSINIT process loading images specified in SYS$LOADABLE_IMAGES:VMS$SYSTEM_IMAGES.DATA, the file that tells the system which additional images to load (including user written). Now, SYSINIT executes in process context, memory management is turned on, and the I/O routines are loaded. This would seem to indicate that it could page in a sharable image (e.g., the one containing $GETUAI) Jim.  ================================================================================ Note 9.0 why wait for the MPW? 7 replies TEMBEL::MANDENOG29 "Graham Mandeno - ALCAN" 10 lines 20-AUG-1991 11:12 -------------------------------------------------------------------------------- Can someone from Engineering please tell me: What is the purpose of the RWMPB state? I can see no reason for refusing to put more pages on the modified list just because the MPW is busy - if there is plenty of memory on the freelist then there is no shortage of a resource, and if there is not then any process requiring a free page will go into FPG state. Can any harm be caused by setting MPW_WAITLIM and MPW_LOWAITLIM to values very much greater than MPW_HILIMIT? ================================================================================ Note 9.1 why wait for the MPW? 1 of 7 VAXNET::DUFFJ8 "The Doc is [IN]" 5 lines 20-AUG-1991 11:23 -------------------------------------------------------------------------------- I thought RWMPB was caused by a process faulting a page that was on the modified list that the MPW was currently in the process of writing to the backing store? Jim Duff. ================================================================================ Note 9.2 why wait for the MPW? 2 of 7 TEMBEL::MANDENOG29 "Graham Mandeno - ALCAN" 7 lines 20-AUG-1991 13:47 -< that's wrt-in-progress >- -------------------------------------------------------------------------------- Thanks for your reply, Jim, but that's a "write-in-progress" fault. RWMPB is caused when a process wishes to move a page out to the mod list, but the mod list size is currently at or greater than MPW_WAITLIM. A process stays in that state until the mod list gets back down to MPW_LOWAITLIM. Graham. ================================================================================ Note 9.3 why wait for the MPW? 3 of 7 TEMBEL::GARNERR30 "Whale Oil Beef Hooked" 19 lines 20-AUG-1991 16:09 -< I can't see any reason why not ... >- -------------------------------------------------------------------------------- Although I'm not from Central Engineering (or even DEC), I'll get my 2 cents worth in ... I can't see any reason for making MPW_HILIM pretty high. The limits are there for a pretty good reason. If the MP list was allowed to grow unchecked, eventually all available pages would be put on the modified page list, causing the free list to become exhausted. What happens then ? Swapper is woken up to reclaim some space. Swapper, meanwhile, is busy writing pages back from the MP list to disk etc. etc., and the whole thing turns into a mess. A thrashing process *should* have the brakes forcibly applied to it. What I favour doing on VMS systems is setting MPW_WAITLIM to a pretty high figure, around 2 or 3 times MPW_HILIM, and setting LOWAITLIM to 1 or 2 MPW_WRTCLUSTERs lower than HILIM. That way, you only pause for 1 or 2 I/Os to complete before the MPW gets freed up again. I also favour making MPW_LOLIM and MPW_HILIM pretty close together - like about 9000 and 10000 (It's big systems I'm talking about). I don't care whether memory's short or not - I *want* a modified page cache. ================================================================================ Note 9.4 why wait for the MPW? 4 of 7 VAXNET::ROBBINSM13 3 lines 20-AUG-1991 17:48 -< Fine >- -------------------------------------------------------------------------------- On some systems I've seen the HILIMIT (etc) very high. i.e. >30000. There is no reason why to not do this if your system performs well under this scenario. ================================================================================ Note 9.5 why wait for the MPW? 5 of 7 VAXNET::SOMMERVILB19 "Brian Sommerville - BT Aust." 9 lines 20-AUG-1991 18:06 -< whats mpw_iolimit? >- -------------------------------------------------------------------------------- whats you value for MPW_IOLIMIT? We had great success in removing MPMPB from our system from uping MPW_IOLIMIT (5.4-2 nothing less - but that's another story) We now run at #page_files +2 (=7 in our case) brian ================================================================================ Note 9.6 why wait for the MPW? 6 of 7 TEMBEL::GARNERR30 "Whale Oil Beef Hooked" 3 lines 22-AUG-1991 10:13 -< MBAPA MPW_IOLIMIT BWATA >- -------------------------------------------------------------------------------- I never played with MPW_IOLIMIT. I have, however, noticed VPA complaining about a large number of users in RWMPB state, but put it down to VPA being picky. Maybe this could have solved it. ================================================================================ Note 9.7 why wait for the MPW? 7 of 7 VAXNET::SOMMERVILB19 "Brian Sommerville - BT Aust." 1 line 22-AUG-1991 12:41 -------------------------------------------------------------------------------- Yep, VPA *WAS* bleating on our systems too.....  ================================================================================ Note 10.0 DATARAM memory 1 reply TEMBEL::HARRY27 "The Ghost Who Walks ... " 9 lines 20-AUG-1991 11:23 -------------------------------------------------------------------------------- We are about to purchase DATARAM memory for our VAX4300 and VAX3500. We are getting it installed by DEC and putting it under DEC maintenance. Has anyone done the same recently and if so have you had any problems ? ================================================================================ Note 10.1 DATARAM memory 1 of 1 TEMBEL::LINDSAYP21 5 lines 20-AUG-1991 11:39 -< Dataram memory >- -------------------------------------------------------------------------------- We have heaps of Dataram memory installed on Vaxes and Decstations. Digital both installed it and maintain it - without any problems ...so far.  ================================================================================ Note 11.0 DUAL HOST VAXCluster 1 reply TEMBEL::HARRY27 "The Ghost Who Walks ... " 11 lines 20-AUG-1991 11:26 -------------------------------------------------------------------------------- We have just installed a dual host VAXCluster between a VAX3500 and VAX4300 via KFQSA adapaters on the VAX3500. The word is that because of the KFQSA adapaters, all the SCS traffic goes via ETHERNET and not the DSSI bus. Is this traffic significant ? Will heavy ETHERNET traffic affect the cluster ? ================================================================================ Note 11.1 DUAL HOST VAXCluster 1 of 1 VAXNET::AJGAONKARK26 5 lines 21-AUG-1991 17:22 -< No sweat if ave Enet traffic is low >- -------------------------------------------------------------------------------- You understand the problem correctly. All your SCA traffic will go over the Ethernet and not the DSSI link. If the Ethernet gets busy, say over 6Kbits/sec ave load, the delays will increase and the nodes will time out. If the ave traffic on the Enet is below 40% you should not have any problems that affect your cluster.  ================================================================================ Note 12.0 ULTRIX X.29 access to your home machine No replies VAXNET::NASHA20 27 lines 20-AUG-1991 11:39 -------------------------------------------------------------------------------- An additional method of obtaining X.29 access to your home machine has been provided via the new ULTRIX X.25 software. If you wish to call home: at the local prompt on the terminal server type: CONN ULTRIX_X29 login as x25 with password ultrix (note this MUST be in lower case) to connect via X.29 x29login dte-number to run up a test demo in Australia use: x29login 050527532100233 Assistance or answers to questions can be obtained from the X.25 for ULTRIX demo area next to the Networking booth.  ================================================================================ Note 13.0 Reverse LAT printing (I think) 1 reply TEMBEL::HARRY27 "The Ghost Who Walks ... " 16 lines 20-AUG-1991 12:28 -------------------------------------------------------------------------------- We have a service on a terminal server that connects to a clearing house (AUSTRACLEAR). Our users connect to this service and log on remotely to that system. At the end of the day reports are sent to use via the terminal and printed via a printer attached to the printer port of the terminal. We would like to print on one of our line printers, which are also connected to the terminal server. Is it possible to connect the printer port of the terminal to the terminal server and print to the line printer ??? ================================================================================ Note 13.1 Reverse LAT printing (I think) 1 of 1 TEMBEL::BEGGJ17 "Jeremy Begg" 20 lines 20-AUG-1991 15:24 -< Use a Service name and a Dedicated Service port >- -------------------------------------------------------------------------------- Why not just send the reports to the line printer directlty? Anyway, it should be possible. Define the port to which the printer is connected as a service, and set up the port to which the terminal's printer port is connected so that it uses a dedicated connection to that service, ie SET PORT n DEDICATED SERVICE printer where 'n' is the number of the port to which the terminal's printer port is connected, and 'printer' is the name of the service created for the port on which the line printer is connected. Then, when something is sent to the terminal's printer port, the server should send the data stream to the printer. It might be more difficult to break the connection at the end, though. Jeremy Begg VSM Software Service jeremy@vsm.com.au  ================================================================================ Note 14.0 User Profiles No replies TEMBEL::ALDRIDGED14 3 lines 20-AUG-1991 12:32 -------------------------------------------------------------------------------- During DECUS last year, we were told that a "future" version of VMS would have user assignable profiles. No mention this year. Will VMS be giving us profiles (presumably through the UAF), and when?  ================================================================================ Note 15.0 InfoServer 100 4 replies TEMBEL::GREERA14 3 lines 20-AUG-1991 13:32 -------------------------------------------------------------------------------- Does anyone know if the InfoServer 100 can read CD's in the format used by Apple. ================================================================================ Note 15.1 InfoServer 100 1 of 4 TEMBEL::GARNERR30 "Whale Oil Beef Hooked" 3 lines 20-AUG-1991 16:00 -< There's doco on this >- -------------------------------------------------------------------------------- I have seen some doco (can't remember quite where, but I think it's not too hard to find) on which formats the Infoserver understands. One of them is "High Sierra", but I can't remember what the others are. ================================================================================ Note 15.2 InfoServer 100 2 of 4 VAXNET::ROBBINSM13 1 line 20-AUG-1991 17:51 -< Maybe >- -------------------------------------------------------------------------------- Is apple ISO 9660 (?) compliant? If so then yes. ================================================================================ Note 15.3 InfoServer 100 3 of 4 VAXNET::GREERA14 3 lines 20-AUG-1991 17:52 -------------------------------------------------------------------------------- Unforunately I believe it uses it's own format. ================================================================================ Note 15.4 InfoServer 100 4 of 4 VAXNET::GORHAMM24 8 lines 22-AUG-1991 09:33 -< INFOSERVER - DECSTATIONS/VMS Doc - Digital Central Engineering >- -------------------------------------------------------------------------------- A gentleman asked me if DECstations could use the infoserver to access VMS documentation. Today, you can have a VMS system mount the VMS ConOLD disc from an infoserver, then use UCX (the Ultrix/VMS connection) on the VMS system to export the VMS ConOLD to ULTRIX systems. On the ULTRIX system, you NFS mount the exported volume, point Bookreader at the mounted volume, and Bookreader reads the files.  ================================================================================ Note 16.0 DECram vs PDDRIVER 2 replies TEMBEL::WALLACEA21 7 lines 20-AUG-1991 15:37 -------------------------------------------------------------------------------- What advantage does DECram offer over using PDDRIVER and SETSIZE as used in STABACKIT.COM? I've setup a PDA0: device using PDDRIVER, giving 130 IO/sec on a VAXstation 3100. Is DECram just an officially supported version of PDDRIVER at extra cost, or does it provide higher performance? ================================================================================ Note 16.1 DECram vs PDDRIVER 1 of 2 TEMBEL::SOMMERVILB19 "Brian Sommerville - BT Aust." 5 lines 20-AUG-1991 15:43 -------------------------------------------------------------------------------- for one it supported and guaranteed to be there in there in the next release..... brian ================================================================================ Note 16.2 DECram vs PDDRIVER 2 of 2 VAXNET::BEGGJ17 "Jeremy Begg" 11 lines 20-AUG-1991 15:56 -< PDA0 is not very efficient, I believe >- -------------------------------------------------------------------------------- I believe the main problem with using PDA0: is that it uses memory very poorly, doing its transfers using very small buffers. It is probably OK for a single- user machine, but not the best on larger systems. If you want more information, and (WARNING: ADVT!) details on a 3rd-party product, please get in touch. Jeremy Begg VSM Software Services jeremy@vsm.com.au (08) 238 8074  ================================================================================ Note 17.0 SLS Scheduling 3 replies TEMBEL::ROBERTSN29 6 lines 20-AUG-1991 16:00 -------------------------------------------------------------------------------- I have problem with SLS. It schedules jobs at midnight to run during the next day. On some days all of the required jobs make it onto the SLS queue on other days only some make it. The number of jobs that go missing varies from day to day. The work around is to restart SLS or run SBQUEUE.COM myself but I would rather have the automatic scheduling work properly. Any ideas or is just SLS. ================================================================================ Note 17.1 SLS Scheduling 1 of 3 VAXNET::SOMMERVILB19 "Brian Sommerville - BT Aust." 7 lines 20-AUG-1991 18:02 -< clock drift? >- -------------------------------------------------------------------------------- Is this a cluster? If so how far apart do the clocks drift? Ie submitting a job for midnight tomorrow, might have already arrvied on another node. Try tomorrow+ grater than the time drift that you normally see. brian ================================================================================ Note 17.2 SLS Scheduling 2 of 3 TEMBEL::GARNERR30 "Whale Oil Beef Hooked" 2 lines 22-AUG-1991 10:16 -< Avoid boundary conditions >- -------------------------------------------------------------------------------- As a matter of dogma, I avoid doing things on boundaries like midnight - it's asking for trouble. ================================================================================ Note 17.3 SLS Scheduling 3 of 3 VAXNET::OMELEYR22 17 lines 22-AUG-1991 15:19 -< Try looking at the log file for the midnite job >- -------------------------------------------------------------------------------- SLS fires off a job at Midnight that reads all ther *sbk.com files in sls$sysbak... the timing of this job (to my knowledge) is not easily changed, other than locating the command routine in SLS$SYSTEM that does the job and modifying the time (NB: put the altered command routine in SLS$ROOT:[CUSTOM], which is the first part of the SLS$SYSTEM search string). Regarding the failure of all jobs to get started, I'd suspect a resource problem somewhere... for starters are there any "other" SLS jobs executing on the syste around this time ? (files locked.. etc). Secondly have a look at the log file (sorry forget the file name, *clean* mybe) in sls$root:[000000], you should get a good guide to what has gone wrong from the log file. Also ensure that all SBK files are owned by SLS and not someone else. Rob Sydney CSC  ================================================================================ Note 18.0 ScriptPrinter to non-DEC printers 7 replies TEMBEL::WALLACEA21 11 lines 20-AUG-1991 18:02 -------------------------------------------------------------------------------- What is the position of using ScriptPrinter software with non-Digital Postscript printers? Does Digital legalise such use? What modifications are needed to make it work? I have a HP-LaserJet-III with a PostScript cartridge (PacificPage) which works with a COPY Filename.ext Port: command, but just hangs when sending via a queue. I tried using ScriptPrinter, but it also just hangs. Any suggestions? ================================================================================ Note 18.1 ScriptPrinter to non-DEC printers 1 of 7 TEMBEL::BEISSELC15 "Craig Beissel Coles Myer Ltd." 20 lines 20-AUG-1991 18:11 -< The TRUTH (?) about CPS >- -------------------------------------------------------------------------------- See my reply in the Workstations conference. The nitty gritty is that CPS uses the PostScript extensions supported by DEC's printers. If you have a non-DEC PostScript printer on a CPS queue it falls over at the first extension command. I'm affraid you have to »pay« DEC's prices to get the funtionality. Some of the code that uses the extensions is in the device control library SYS$LIBRARY:CPS$LIBRARY.TLB. It is my opinion that DEC could provide a device control library that did not use the extensions, but chooses not to, because then we would all buy an HP Laserjet IIIc or whatever gave the best price/performance at the time. What about it Digits, is this really the case, or is there a »real« technical reason why this could not be done? ================================================================================ Note 18.2 ScriptPrinter to non-DEC printers 2 of 7 TEMBEL::BEISSELC15 "Craig Beissel Coles Myer Ltd." 5 lines 20-AUG-1991 18:16 -< !@#$% DECservers >- -------------------------------------------------------------------------------- Sorry, my reply is Note 5.1 in the OFFICE_AUTOMATION conference. By the way, is everyone getting SICK of these !@#$% DECserver 90s throwing you out with a Protocol Error just BEFORE you save your note? ================================================================================ Note 18.3 ScriptPrinter to non-DEC printers 3 of 7 VAXNET::WONGS3 4 lines 20-AUG-1991 18:41 -< Scriptprinter - some do and some don't >- -------------------------------------------------------------------------------- It is *possible* to use Scriptprinter with non-Digital printers. I've had success with a DataProducts LZR1260 but not the Apple Lazerwriter IINTX. This is using v2.0 Scriptprinter. ================================================================================ Note 18.4 ScriptPrinter to non-DEC printers 4 of 7 VAXNET::GORHAMM24 7 lines 23-AUG-1991 08:48 -< CPS with non-Digital printers - Central Engineering Reply >- -------------------------------------------------------------------------------- Yes, you can use CPS with non-Digital printers, as long as you have the CPS license. The problem you describe could be caused by different extensions to Postscript used by different vendors, the default fonts on the printers, and/or the concentions on use of control characters. Digital Software Service may be able to help you solve the problem. ================================================================================ Note 18.5 ScriptPrinter to non-DEC printers 5 of 7 VAXNET::SARGENTD25 16 lines 23-AUG-1991 11:19 -< More on ScriptPrinter to non-DEC printers >- -------------------------------------------------------------------------------- I heard mentioned during a session today that a VMS print symbiont for Apple Laserwriters was available. I would appreciate more details on this if anyone has more information. Since the Apple Laserwriter conforms closely to the Adobe "standard" I would have thought this symbiont would be more successful driving non-Digital printers. We are currently running one LN03R and *lots* of DP1260s using the ScriptPrinter symbiont with mixed success. Problems include queues stopping for no apparent reason and loss of whole print jobs. Again if anyone knows more about the Laserwriter symbiont could they please reply. Daryl Sargent ACT Electricity & Water ================================================================================ Note 18.6 ScriptPrinter to non-DEC printers 6 of 7 TEMBEL::MCINTOSHR2 10 lines 23-AUG-1991 12:26 -< HP LaserJet III with HP Postscript is OK >- -------------------------------------------------------------------------------- Re: .0 "I have a HP-LaserJet-III with a PostScript cartridge (PacificPage) which works with a COPY Filename.ext Port: command, but just hangs when sending via a queue." We print to a HP LaserJet III with HP Postscript cartridge using the standard LAT symbiont without any other software (CPS etc.) without any problems. Maybe it's something in your PacificPage cartridge? ================================================================================ Note 18.7 ScriptPrinter to non-DEC printers 7 of 7 VAXNET::MCKENZIED17 "David McKenzie Digital Box Hill " 2 lines 23-AUG-1991 12:27 -< done via pathworks >- -------------------------------------------------------------------------------- You can drive an Apple Laserwriter from VMS via pathworks. i.e the Laserwriter becomes accesable to VMS users  ================================================================================ Note 19.0 rms bashing 1 reply TEMBEL::PUGMIREC17 18 lines 21-AUG-1991 13:41 -------------------------------------------------------------------------------- How about some RMS bashing! I found with a simple program, reading a large file, the speed of the program was proportional to the number of records, not the size of them. speed = k*nrecords This isn't surprizing for records of size 1-->80 bytes, but it still holds true for much larger records, 100-->5000 bytes. The program was written in basic, doing simple move/put operations to a sequential file. Is there some clever RMS feature I should have enabled to make this work at a reasonable speed, or is this a straight qio limitation, in which case is DEC planning any improvements to this area. ================================================================================ Note 19.1 rms bashing 1 of 1 TEMBEL::GARNERR30 "Whale Oil Beef Hooked" 15 lines 22-AUG-1991 10:34 -< Features coming out of its *#%&#@ >- -------------------------------------------------------------------------------- > Is there some clever RMS feature I should have enabled to > make this work at a reasonable speed, or is this a straight > qio limitation, in which case is DEC planning any improvements > to this area. RMS has so many features for speeding things up you wouldn't believe ... and since you're writing the program, you have access to lots that most system managers can't use. The simplest to use for straight sequential reads is the Multiblock count. Setting this to a large number (e.g. 127) should speed up reading a sequential file considerably. You can experiment with it using SET RMS/BLOCK, and then code it into the program when you've found a nice value. The RAB field is RAB$B_MBC. You could also play with read-ahead and multi buffers, and doing asynchronous reads etc. etc.  ================================================================================ Note 20.0 Introducing ACMS 4 replies VAXNET::GREEND4 1 line 22-AUG-1991 10:15 -------------------------------------------------------------------------------- Some shared experience on introducing ACMS would be useful. ================================================================================ Note 20.1 Introducing ACMS 1 of 4 VAXNET::GARNERR30 "Whale Oil Beef Hooked" 6 lines 22-AUG-1991 13:27 -< a title for your reply: >- -------------------------------------------------------------------------------- Can you narrow your question down a little ? I worked on a big ACMS site for 4 years, and if I were to tell you about it in detail, I'd be typing all day (my feet are killing me already). How many applications ? How many users ? Who is writing them ? Are you using ACMS for performance reasons only ? ... ================================================================================ Note 20.2 Introducing ACMS 2 of 4 VAXNET::GREEND4 24 lines 22-AUG-1991 18:10 -< ACMS - Anybody out there >- -------------------------------------------------------------------------------- We are large VAX shop with 2 clusters and a network spanning Australia. Currently we have the usual mix of in-house applications, packages and Digital products such as All-in-1. We are about to embark on some large-scale application development using Decforms, Rdb and ACMS. These will be supporting some hundreds of users and cater for periods of high transaction rates. The questions I have relate to the introduction of ACMS into the development environment, the design decisions in choosing ACMS candidate applications, program design considerations and the integration of ACMS menuing/security with other existing applications. Now my feet are killing me and it's getting near time to be fed. Perhaps we could continue some dialogue or correspondence outside of this notes session. I don't know where you are (NZ or Aust) If I leave a contact perhaps you could drop me a line (preferably one without the Whale Oil on it) Regards, Darrell Green Western Mining Corporation Melbourne (03) 685 6206 ================================================================================ Note 20.3 Introducing ACMS 3 of 4 PENCEL::DEMO 0 lines 23-AUG-1991 09:14 -< I'd try asking in the IS or L&T conference >- -------------------------------------------------------------------------------- ================================================================================ Note 20.4 Introducing ACMS 4 of 4 VAXNET::GARNERR30 "Whale Oil Beef Hooked" 5 lines 23-AUG-1991 13:17 -< Who was that masked man ... >- -------------------------------------------------------------------------------- Robin Garner, Canberra (06) 243 6480  ================================================================================ Note 21.0 4000 alternate boot 3 replies VAXNET::ROBERTSN29 5 lines 22-AUG-1991 10:22 -------------------------------------------------------------------------------- At an unattended site is it possible to boot from an alternate disk on a 4000 series machine? This application has redundancy built into all components, it would be better to have this last possibilty covered as well. ================================================================================ Note 21.1 4000 alternate boot 1 of 3 VAXNET::MCKENZIED17 3 lines 22-AUG-1991 13:21 -< Use VCS >- -------------------------------------------------------------------------------- I would suggest that VCS will do this David McKenzie ================================================================================ Note 21.2 4000 alternate boot 2 of 3 VAXNET::GORHAMM24 3 lines 23-AUG-1991 08:49 -< Alternate disk on 4000 - Central Engineering Reply >- -------------------------------------------------------------------------------- Check with your local office - the folks back at Spitbrook said no conditionally, but they really needed to know your configuration in detail to more adequately answer your question. ================================================================================ Note 21.3 4000 alternate boot 3 of 3 TEMBEL::KEOGH "I feel better than James Brown" 8 lines 23-AUG-1991 10:35 -------------------------------------------------------------------------------- David is right ... for once :-) ! Use VCS to - detect that the system is down and initiate a boot automatically or - allow you to connect as the console from any terminal on your wide area net, even a dialup (there is only one problem with doing system management via laptop+modem+celular phone, you can't keep the sand out of the keyboard).  ================================================================================ Note 22.0 MOVEFILE VMS feature 2 replies TEMBEL::RITCHIEJ16 8 lines 22-AUG-1991 10:37 -------------------------------------------------------------------------------- Will the MOVEFILE feature coming in VMS V5.5 be used as the basis for a disk defragmentation tool (utility, program etc...) from Digital in the future ??? - JR ================================================================================ Note 22.1 MOVEFILE VMS feature 1 of 2 TEMBEL::GORHAMM24 1 line 22-AUG-1991 11:21 -< MOVEFILE/DEFRAGMENTATION TOOL - Digital Central Engineering >- -------------------------------------------------------------------------------- Right on. Work is currently underway on a defrag tool. ================================================================================ Note 22.2 MOVEFILE VMS feature 2 of 2 VAXNET::KARAGIANNA20 13 lines 22-AUG-1991 16:31 -< MOVEFILE PRIMITIVES FOR ALL TO USE. >- -------------------------------------------------------------------------------- These MFP will be available to all people which have written disk defrag software. The primitives will be public knowledge and the Diskeeper product has already allowed for these primitives as they are a field test site for VMS. I hope you don't expect that because the MFP's are going to be in VMS to relate to getting a free defragger as part of VMS. The socalled defragger will have a price tag. Arthur Karagiannis Network Computer Software  ================================================================================ Note 23.0 How do I generate an AUDIT event? 2 replies TEMBEL::BEGGJ17 "Jeremy Begg" 23 lines 22-AUG-1991 10:57 -------------------------------------------------------------------------------- I am using a 3rd-party TCP/IP product on my Vax (TGV MultiNet), which marks Telnet and Rlogin connections as LOCAL, unlike UCX which marks them as REMOTE. For the system I am considering, it would be nice to know when a user comes in via Telnet/Rlogin instead of LAT, but the only way I can achieve this at present is to enable security auditing for ALL interactive logins, and look at the terminal source. I would like to have Telnet/Rlogin connections audited separately, and as MultiNet provides exit points for doing things like audit trails etc, I thought I could try writing my own security audit records, much in the same way as the system provides mechanisms for user-written accounting messages. I could get MultiNet to maintain its own audit trail (there are examples provided which do this) but I woulr prefer to record them in the existing VMS security audit. Can someone please let me know if this is possible? Ideally the audit message would be an INTERACTIVE LOGIN with a different subtype for each kind of terminal connection. Jeremy Begg VSM Software Services jeremy@vsm.com.au P.S. Is this a candidate for a T-shirt? :-) ================================================================================ Note 23.1 How do I generate an AUDIT event? 1 of 2 VAXNET::GARNERR30 "Whale Oil Beef Hooked" 5 lines 22-AUG-1991 12:05 -< A partial solution >- -------------------------------------------------------------------------------- Use $SNDJBC (or $ REPLY) to send a message to the security operator, and enable OPERATOR.LOG for security messages (the default). This is the best I can suggest -- It doesn't go to the security audit log, but it does at least get recorded in a standard VMS place, and can go to an operator's terminal. ================================================================================ Note 23.2 How do I generate an AUDIT event? 2 of 2 TEMBEL::DUFFJ8 "DUFF@DECUS.COM.AU" 8 lines 22-AUG-1991 12:41 -< Try appendix A of Analyse/Audit manual >- -------------------------------------------------------------------------------- In appendix A of the Analyse/Audit manual the format of the messages sent to the audit server mailbox are described. I can see nothing stopping you from defining your own facility and alarm codes, attaching to the mailbox, formatting a message and sending it to the mailbox. Hope this helps. Jim.  ================================================================================ Note 24.0 DSH32 sync line speeds No replies VAXNET::GREERA14 4 lines 22-AUG-1991 11:02 -------------------------------------------------------------------------------- On the DSH32 with 2 Sync lines. What is the maximum speed these can be driven at?  ================================================================================ Note 25.0 Index file extent size No replies VAXNET::COUCHD27 15 lines 22-AUG-1991 12:30 -------------------------------------------------------------------------------- If a disk is initialised with default parameters, the index file is initially quite small. Later, as files are created on the disk, the index file is extended by 1000 blocks at a time to accommodate the extra headers. Because the index file itself can have only a single header, it can have only about 50 extents, giving a maximum size of about 50,000 blocks. That's about 50,000 headers, or 50,000 files if your disk is defragmented. We have been finding, with RA92 and larger disks, that this limit is easily exceeded, and we get HEADERFULL errors when we try to create more files. I know that the fix for this is (1) Initialise the disk with a large /HEADERS value (2) Compress the disk when this problem occurs. My question is whether there is any way to change the 1000 block extent size for the index file.  ================================================================================ Note 26.0 CIXCD No replies VAXNET::MCKENZIED17 2 lines 22-AUG-1991 12:32 -------------------------------------------------------------------------------- I believe that the reason that DECnet over CIXCD is not supported is that the CIXCD can use both CI paths requiring a new driver  ================================================================================ Note 27.0 Data Compression and Backup Recovery Algorithms 3 replies VAXNET::GARNERR30 "Whale Oil Beef Hooked" 3 lines 22-AUG-1991 12:47 -------------------------------------------------------------------------------- I remember hearing that the compression on the TA90E invalidated the algorithm used by backup for data recovery (i.e. redundant groups etc). Can anybody tell me what the problem is ? (academic interest only). ================================================================================ Note 27.1 Data Compression and Backup Recovery Algorithms 1 of 3 VAXNET::OMELEYR22 18 lines 22-AUG-1991 15:37 -< Compaction moves Backup 1 step further from the tape data >- -------------------------------------------------------------------------------- I believe that the key issue is that the data is being fiddled with after BACKUP has processed it and set up the CRC and redundency groups. Hence your backup is then relying on the integrity of the TA90E to restore the compacted data to allow backup to perform it's processing. The compaction process also compacts the redundency group, so backup see's this recovery means through the TA90E's interpretation of the compacted data... Feedback on the TA90E's compaction reliability is extremely good, similarly with the TA91's. Backup is designed for maximum data save/restore integrity (CRC, Groups), compaction appears to reduce this to an extent (at least in theory...). Perhaps follow this up with the CSC after DECUS. Rob Sydney CSC ================================================================================ Note 27.2 Data Compression and Backup Recovery Algorithms 2 of 3 TEMBEL::BEGGJ17 "Jeremy Begg" 15 lines 23-AUG-1991 09:00 -------------------------------------------------------------------------------- > Feedback on the TA90E's compaction reliability is extremely good, > similarly with the TA91's. Backup is designed for maximum data > save/restore integrity (CRC, Groups), compaction appears to reduce this > to an extent (at least in theory...). Is the TA90 (and 91) drive exactly the same as the IBM mechanism? If so, I would expect them to be very reliable. My experience of IBM systems people has been that they would not be at all happy with a tape drive which was sub- standard, and this type of cartridge drive would be installed at the majority of IBM mainframe sites. Jeremy Begg VSM Software Services jeremy@vsm.com.au ================================================================================ Note 27.3 Data Compression and Backup Recovery Algorithms 3 of 3 TEMBEL::KEOGH "I feel better than James Brown" 13 lines 23-AUG-1991 10:41 -------------------------------------------------------------------------------- The previous answers are correct. Backups's group encoding scheme relies on a knowlege of the phyiscal layout of records on tape. The aim is to recover from one or more records being bad. However, with data compression, the algorithms that backup uses are less useful because the layout on the tape is different. However, the good news is that the TA90 is a reliable backup medium. I would give a best guess ranking something like the following 1) 800 and 1600 BPI tape 2) 8mm and 6250 BPI tape 3) TLZ04 and TA90 4) TF85 (subject to more experience)  ================================================================================ Note 28.0 POSIX Compliance 3 replies VAXNET::WILMOREA31 3 lines 22-AUG-1991 13:57 -------------------------------------------------------------------------------- The new POSIX compliance to be introduced to VMS - will this come as part of the standard VMS license or will additional licensing be required for these features ? ================================================================================ Note 28.1 POSIX Compliance 1 of 3 TEMBEL::BEGGJ17 "Jeremy Begg" 5 lines 22-AUG-1991 14:56 -< POSIX will be separate >- -------------------------------------------------------------------------------- I believe that Mark Gorham said POSIX will not be distributed with VMS. Jeremy Begg VSM Software Services jeremy@vsm.com.au ================================================================================ Note 28.2 POSIX Compliance 2 of 3 VAXNET::OMELEYR22 6 lines 22-AUG-1991 15:23 -< Yep.. it's seperate >- -------------------------------------------------------------------------------- My understanding is that it'll be a LP for VMS. Some of our other products (like the realtime product VAXELN V4.3) have POSIX (1003.1, 1003.4) support bundled. Rob Sydney CSC ================================================================================ Note 28.3 POSIX Compliance 3 of 3 VAXNET::GORHAMM24 15 lines 23-AUG-1991 07:56 -< POSIX/Packaging - Digital Central Engineering >- -------------------------------------------------------------------------------- POSIX 1003.1, .2, and .4 will be delivered as a separate license. We are doing this for time-to-market reasons - VMS is moving forward on three major fronts: open, production system, and performance. There are many advances in all three fronts we will need to bring out, and feel if we offer many of the open features (POSIX and DECwindows MOTIF) separately we can both speed our delivery of the features, and also deliver the features to those who want it (and not to those who don't). Look for Digital to package open components together in the future, similar to the way workstations are packaged with VMS, DECnet, VAXcluster Software, and DECwindows. Our intent is to give you the option of buying a-la carte, but also having a few packages (like open) that are competitively priced and deliver components like POSIX, DECwindows MOTIF, TCP/IP, NFS, etc. We KNOW that price, including service price, will be a key factor here.  ================================================================================ Note 29.0 EWS & Floppy disks on these VAXstations 1 reply TEMBEL::BEGGJ17 "Jeremy Begg" 6 lines 23-AUG-1991 09:21 -------------------------------------------------------------------------------- Does EWS (ELN Workstation Software) allow the floppy disk to be accessed? I suspect not. Jeremy Begg VSM Software Services jeremy@vsm.com.au ================================================================================ Note 29.1 EWS & Floppy disks on these VAXstations 1 of 1 TEMBEL::KEOGH "I feel better than James Brown" 5 lines 23-AUG-1991 10:44 -------------------------------------------------------------------------------- It doesn't. However, if you use an VAXELN development system you could build a target system that had support for and supported device. It would require some knowlege of VAXeln and more memory. What do you want to use it for?  ================================================================================ Note 30.0 Modify queue entry by jobname? No replies TEMBEL::BURGESSI9 4 lines 23-AUG-1991 11:26 -------------------------------------------------------------------------------- One of the features of VMS 5.5 (or 5.4-3?) mentioned was SHOW ENTRY /JOBNAME=nnnn Is there a plan for the more useful SET ENTRY/JOBNAME=nnnn ?  ================================================================================ Note 31.0 Dual-Headed display on a 3100-76 SPX No replies TEMBEL::GRANTA15 4 lines 23-AUG-1991 11:30 -------------------------------------------------------------------------------- Is it possible to (is there a patch to the driver?) to run a dual-headed display on a VAXstation 3100-76 SPX? Thanks, Alistair  ================================================================================ Note 32.0 Accounting for print forms 1 reply TEMBEL::BURGESSI9 8 lines 23-AUG-1991 11:34 -------------------------------------------------------------------------------- I have lots of print queues, many of them simply so the users can be billed for the special stationery they are using -- the accounting entry holds the queuename, but not the form name. Is there any other way to work around the problem? Any plan by Digital to put that vital piece of information into the accounting record? ================================================================================ Note 32.1 Accounting for print forms 1 of 1 VAXNET::BEGGJ17 "Jeremy Begg" 3 lines 23-AUG-1991 14:45 -< Use a user-written symbiont >- -------------------------------------------------------------------------------- One way would be to write your own symbiont (preferably using the PSM$ routines) which writes a user accounting record at the end of the job. Of course, this will result in two records per job, but the information will be recorded.  ================================================================================ Note 33.0 Name/order # of VAX Performance Doc. No replies VAXNET::AJGAONKARK26 11 lines 23-AUG-1991 12:39 -------------------------------------------------------------------------------- THe document that describes the details fo the performance data presented in the conference is: VAX Performance Summary Doc #: EC-N0495-41 It can be ordered from your normal documentation contact. Regards, Kami  ================================================================================ Note 34.0 VMS features first, standards second?? No replies TEMBEL::MCINTOSHR2 20 lines 23-AUG-1991 12:40 -------------------------------------------------------------------------------- The "official" line on VMS vs. Ultrix I have heard is that with Ultrix development the emphasis will be on standards not necessarily features, and with VMS it will be features first and standards second. This sounds OK to me but... When I ask for network wide authentification of users, instead of the current per system UAF authentification, I am told that we will have to wait for the OSF solution. Now I realise that will be necessary for a multi-vendor environment, but if my major requirement is VMS only, then I want a tool NOW that give me network-wide authentification across my VAXs. So returning to my first paragraph I want Digital to live up to the rhetoric about VMS having features first and bring out such a product. It could possibly be a layered product (or SIP) so that when OSF finally does come out with this one, then Digital could offer customers a choice of VMS only or Open (priced accordingly?).