<<< OFFICE$DISK:[NOTES$LIBRARY]OPENVMS.NOTE;1 >>> -< OpenVMS Alpha & OpenVMS VAX >- ================================================================================ Note 1.0 Welcome ... No replies OPNOFF::SYSTEM 5 lines 21-AUG-1992 11:59 -------------------------------------------------------------------------------- Welcome to VAX NOTES This conference is for discussions relating to OpenVMS Alpha & OpenVMS VAX ================================================================================ Note 2.0 Mount verification questions 6 replies OPNNET::MCGILLC8 38 lines 24-AUG-1992 11:17 -------------------------------------------------------------------------------- I am interested in learning more about mount verification. I know the SYSGEN params are stored in IOC$GW_[TAPE_]MVTIMEOUT The Driver Dispatch Table has fields DDT$L_MNTVER (address of routine to call at beginning and end of mount verification) DDT$L_MNTV_SSSC (address of routine to call for mount verification for a shadow set state change [reserved to DEC]) DDT$L_MNTV_FOR (address for foreign devices) DDT$L_MNTV_SQD (address for sequential devices) The UCB has flags UCB$V_MNTVERIP (mount verification in progress) UCB$V_MNTVERPND (mount verification is pending and device is busy) (incidentally, what's the difference between these two ?) (The above is primarily from the device support manuals) *BUT* where are the cells that keep track of the time a particular device (when it goes into mount verification) will timeout ? I can't find these for the life of me. I know that when a disk enters mount verification, that changing MVTIMEOUT (although a dynamic param) has no effect on *that* disk. I presume the driver takes a copy of IOC$GW_MVTIMEOUT each time a disk enters mount verification, and then starts decrementing it until zero ? If so, where is this ? If not, how does it work ? Can anyone help ? ================================================================================ Note 2.1 Mount verification questions 1 of 6 OPNNET::DAVISS25 17 lines 24-AUG-1992 17:06 -< mount verification >- -------------------------------------------------------------------------------- I will be discussing mount verification in gory detail at the Cluster High Availability guidelines talk this week. To answer your specific question, Mount verification active countdown time for a given disk is only stored in the internal control IRP that is performing the mount verification operation. For path failover in progress operations, this IRP is alternately used as a fork block on the internal fork and wait queue and is used to issue an IO$_PACKACK function to the disk class driver. THe timeout is stored in one of the overlaid fields in this IRP (I don't remember which off the top of my head...). If you would like a higher bandwidth discussion on the topic, come find me for a chat at one of my booth duty times... Scott ================================================================================ Note 2.2 Mount verification questions 2 of 6 OPNNET::MCMORRAND11 2 lines 25-AUG-1992 09:01 -< set timeout to the limit >- -------------------------------------------------------------------------------- I think the way to go is to set the timeout to be the maximum. It's easier to get your disk working again. ================================================================================ Note 2.3 Mount verification questions 3 of 6 OPNNET::MCGILLC8 24 lines 25-AUG-1992 11:24 -< Yes but !!! >- -------------------------------------------------------------------------------- (like anything) there is a cost in upping this parameter too high. This is that when a disk fails, it goes into mount ver and things stall. We had this just last week at our site. We had MVTIMEOUT set to 12 hours. We thought this would be sufficient to get our engineer out and get the disk going again in all but very bad failures. However, when you reach the point that you decide you want to restore from tape to a spare drive, you can do so, but cannot mount the new disk since it has the same label as the old (stuck) one. You also cannot dismount the old disk. Variants of DISMOUNT, DISMOUNT/ABORT, /OVERRIDE=CHECKS etc. all hang. Lots of processes in RWAST (including about 10 in the system management area trying to look at the disk, do SHOW DEVICE, do DISMOUNT etc. Yes I know the new disk can be temporarily labelled something else, but there really does come a time when you say "I've had enough of this failed disk - *dismount it from my system now*. Too long a MVTIMEOUT period has its disadvantages, because you just want it to timeout. You could also use the IPC prompt (force CPU to IPL 12 at the console) and enter "C " to cancel mount ver, but I don't know if this will clear it up cluster wide or whether you need to do this on all nodes - can anyone answer this ? ================================================================================ Note 2.4 Mount verification questions 4 of 6 OPNNET::DAVISS25 4 lines 25-AUG-1992 13:23 -------------------------------------------------------------------------------- You must use IPC on each node in the cluster to free up that node... Scott ================================================================================ Note 2.5 Mount verification questions 5 of 6 OPNNET::LITTLEP11 4 lines 27-AUG-1992 15:04 -< IPC ??? >- -------------------------------------------------------------------------------- Could you give me the details for using IPC to dismount a disk, that is in mount verification - first time I've heard of it. Ta ================================================================================ Note 2.6 Mount verification questions 6 of 6 OPNNET::MCGILLC8 46 lines 27-AUG-1992 15:35 -< RTFM >- -------------------------------------------------------------------------------- > Could you give me the details for using IPC to dismount a disk, that is > in mount verification - first time I've heard of it. See the VMS docco - I think it's "Guide to Maintaining A VMS System", but I could be wrong - in any case see the System Management Master Index and look for "IPC". You go to the console, and press ^P (or break - for the little VAX) to get to the "Chevron" prompt (>>>). If this is a VAX you must halt (eg. 8000 series) enter H Then : >>> D/I 14 C >>> C IPC> C DSA45: IPC> EXIT The D/I 14 C means deposit (/internal) 12 (C in hex) into register 14. By the way, IPC is a shortened form of IPL C (ie. 12) The IPC prompt accepts 3 (documented) commands : IPC> Q (recalculates quorum - use when node activity blocked 'cos you've lost quorum) IPC> X (enter XDelta debugger - system must have had it loaded) IPC> C (cancel mount verifucation on the selected drive) Note that if the node is in a cluster, you must complete the whole exercise within the time RECNXINTERVAL seconds, which is 20 by default, so that other nodes in the cluster do not kick you out. In practice, 20 seconds is heaps, don't panic too much. Your system will cancel the mount verification on the selected drive and return to normal. The disk will now be in MV timeout. *WARNING* - There is a bug in the 8600 / 8650 microcode that will make the system crash whenever you leave the IPC prompt (exit). I do not know if this applies to other similar hardware types (eg. 8500 series), but it is certainly OK on 6000 series (6410, etc.). Suggest you try it when next you are about to shutdown your node (ie. users off). Anyway, RTFM !!!!! ================================================================================ Note 3.0 DECdefragger 16 replies OPNNET::POWELLH27 9 lines 24-AUG-1992 15:43 -------------------------------------------------------------------------------- Now that VMS has movefile I understand that a DEC defragger is available. Does anyone out there have it or know anything about it? I am interested in how it compares to the 3rd party offerings and what the pricing is like. Cheers, Harry Powell ================================================================================ Note 3.1 DECdefragger 1 of 16 OPNNET::WADEM15 "Malcolm Wade" 10 lines 24-AUG-1992 16:37 -< DFG on CONDIST >- -------------------------------------------------------------------------------- It's called DFG and is available on CONDIST. You can also get a trial license from DEC. I've been using it for about 4 months now and it's been pretty good. According to my stats the disks are pretty well defrgamented and the system 'feels' better. I'm not sure of the pricing. Malcolm ================================================================================ Note 3.2 DECdefragger 2 of 16 OPNNET::ROSSR12 "Rollo Ross, Uni SA" 8 lines 24-AUG-1992 17:04 -< See Usenet News? >- -------------------------------------------------------------------------------- > Now that VMS has movefile I understand that a DEC defragger is > available. > Does anyone out there have it or know anything about it? It's been discussed in Usenet News (comp.os.vms) a number of times. Do you have access to these articles? If not, give me a yell. Rollo ================================================================================ Note 3.3 DECdefragger 3 of 16 OPNNET::MCINTOSHR2 6 lines 24-AUG-1992 17:41 -< $1 for education >- -------------------------------------------------------------------------------- If you're an EDU site you can get DFO for $1. The other defragger vendors claim that DFO's movefile mechanism is no different from theirs, movefile just has the Digital stamp of approval and is included in VMS. ================================================================================ Note 3.4 DECdefragger 4 of 16 OPNNET::BEISSELC15 4 lines 25-AUG-1992 08:27 -< Bug with movefile in a cluster >- -------------------------------------------------------------------------------- I'm not sure of the versions, but I believe that in VMS v5.5 the Movefile function DOES NOT work correctly in a VAXcluster (locks etc.). Can any Digits out there confirm or deny? ================================================================================ Note 3.5 DECdefragger 5 of 16 OPNNET::ROSSR12 "Rollo Ross, Uni SA" 42 lines 25-AUG-1992 08:41 -< ex >- -------------------------------------------------------------------------------- > I'm not sure of the versions, but I believe that in VMS v5.5 the > Movefile function DOES NOT work correctly in a VAXcluster (locks etc.). This was mentioned in News a while ago: Newsgroups: comp.os.vms Subject: Re: DEC DFO (DEC File Optimizer) - V1.0 "DECs Disk Defragmenter" Message-ID: <1992Mar26.095517.12414@rdg.dec.com> From: verbist@movies.enet.dec.com (Guy Verbist) Date: 26 Mar 92 09:43:23 GMT DWING@UH01.Colorado.EDU (Dan Wing) writes... >As DEC has admitted two bugs (that I've heard of) in their MoveFile primitive, >I'd be wary of using DFO, or using one of the third-party products that are >using VMS V5.5's MoveFile. (The bugs are possible corruption of ACLs, which >I haven't seen a patch for, and the other is failure to erase blocks after >they've been moved for a file with erase-on-delete - which does have a patch). > Perhaps I can clarify matters a little . .. 1) There is a *very* small timing window involving a system crash occurring while you are trying to move a multi-header file, with an ACL, when the file system is under heavy stress. When (if) this window occurs then the ACL may become corrupt, and is marked as such. This window has *never* been obs ervedin practice. There is no chance of any actual data being lost. 2) I believe there is a CSC patch available for the erase-on-delete blocks not being erased. It is certainly fixed in any release post v5.5 (i.e. v5.5-1) Hope this helps. guy +-----------------------------------------------------------------------------+ | verbist@movies.enet.dec.com | Free the Files 11 ! | | Guy Verbist | VMS Engineering, Livingston, UK, EH54 6AG | | | | Opinions, where given, are mine. Any attempt to attribute them to another | | party is an infringement of copyright. | +-----------------------------------------------------------------------------+ ================================================================================ Note 3.6 DECdefragger 6 of 16 OPNNET::MCMORRAND11 5 lines 25-AUG-1992 09:05 -------------------------------------------------------------------------------- As for pricing, the prices we were quoted were cheaper than 3rd party. We have just started using it with VMS V5.5-1 without any trouble. ================================================================================ Note 3.7 DECdefragger 7 of 16 OPNNET::CAMERONJ12 6 lines 25-AUG-1992 14:46 -------------------------------------------------------------------------------- Been using it myself on my workstation (3100) with an RZ25, nightly, seems to work very well. No data integrity issues. Just don't run it when you want to do other work on the disk. James (DEC) ================================================================================ Note 3.8 DECdefragger 8 of 16 OPNNET::GRINTONG4 "Geoff Grinton, SECV" 6 lines 25-AUG-1992 19:14 -< You can play without a licence >- -------------------------------------------------------------------------------- THere was a DEC advertisement in a Computerworld recently suggesting that you can install DFO without the licence. The functionality you get is that you can use the SHOW function to examine your fragmentation, etc, but not do anything real. It provides a useful look at the product. ================================================================================ Note 3.9 DECdefragger 9 of 16 OPNNET::MCCORMACKP9 7 lines 26-AUG-1992 18:32 -< Its cheap sometimes >- -------------------------------------------------------------------------------- We bought DFG for our 4300 Vaxserver. It was less than $1000. I could not believe it. Its definitely a different story though on non-server systems. We run it 3 hours per night from 1:00 to 4:00 am. No integrity problems in 6 weeks. Paul McCormack BHP Petroleum ================================================================================ Note 3.10 DECdefragger 10 of 16 OPNNET::CAMERONJ12 9 lines 27-AUG-1992 09:42 -------------------------------------------------------------------------------- Re: Note 3.9 by OPNNET::MCCORMACKP9 > It was less than $1000. It's always something that internal Digits tend to ignore; getting a license for products for internal use is so easy that we forget that it costs so much externally. James ================================================================================ Note 3.11 DECdefragger 11 of 16 OPNNET::GARNERR30 "Robin Garner" 4 lines 27-AUG-1992 10:14 -< Curious >- -------------------------------------------------------------------------------- > It's always something that internal Digits tend to ignore; getting a > license for products for internal use is so easy that we forget that it What's an *external* digit ... ? ================================================================================ Note 3.12 DECdefragger 12 of 16 OPNNET::CAMERONJ12 9 lines 27-AUG-1992 10:19 -------------------------------------------------------------------------------- Re: Note 3.11 by OPNNET::GARNERR30 > What's an *external* digit ... ? Don't know. Maybe a kick-out digit? Or a right-sized digit? Would you call it right-justification ala FMS? ;-) ================================================================================ Note 3.13 DECdefragger 13 of 16 OPNNET::DUMBRELLG22 1 line 27-AUG-1992 13:54 -< *external*? >- -------------------------------------------------------------------------------- An *external* digit? Why that must be one that's been extracted... ================================================================================ Note 3.14 DECdefragger 14 of 16 OPNNET::MCMORRAND11 2 lines 27-AUG-1992 15:14 -------------------------------------------------------------------------------- I guess you've all heard the one about what's transparent and lies on the floor? ================================================================================ Note 3.15 DECdefragger 15 of 16 OPNNET::CAMERONJ12 6 lines 27-AUG-1992 16:08 -------------------------------------------------------------------------------- Re: Note 3.14 by OPNNET::MCMORRAND11 > I guess you've all heard the one about what's transparent and lies on > the floor? No, what? ================================================================================ Note 3.16 DECdefragger 16 of 16 OPNNET::TONKINP2 11 lines 27-AUG-1992 18:28 -< PerfectDisk good as any >- -------------------------------------------------------------------------------- We have PerfectDisk from Software Intelligence which you can use with or without the movefile primitive. We use the product without movefile and have found no problems with it. We also purchased this product as a corporate licence at a time when the Digital had not released theirs and would not budge on corporate licences. The only feature Digital offers that is different is using VPA to determine HOTfiles for file placement. If you want to use it with this you will have to pay the bucks. If you wish to discuss it further please don't hesitate to call. Peter Tonkin Coles Myer ================================================================================ Note 4.0 $ENQ goes into MWAIT 2 replies OPNNET::DUMBRELLG22 9 lines 24-AUG-1992 17:44 -------------------------------------------------------------------------------- I have a pplication that at time does many $ENQ's (creations, not converts). It's multithreaded and so the $ENQ is asynchronous. I've noticed that it spends some time in MWAIT while it's doing this. I assume that this is the directory lookup operation. I understand why it has to do the dir function, but going into MWAIT doesn't sound like a well-behaved system service to me!?! Seems it's effectively stalling my application while it goes to query the dir entry. ================================================================================ Note 4.1 $ENQ goes into MWAIT 1 of 2 OPNNET::DAVISS25 14 lines 25-AUG-1992 09:39 -------------------------------------------------------------------------------- I do not know of any mutex wait associated with $ENQ operations. The common stall states associated with asynchronous lock requests (.ie processing prior to the request being queued) are RWSCS and RWCLU. RWSCS is waiting for low level SCS messages that queue the lock request and can be caused by SCS resource problems or network connection problems. RWCLU is cluster state transition lock rebuild. For non-asynchronous lock requests, LEF is obviously common. So, what else is this application doing besides locking and have you actually verified that the MWAITed processes PC is part of the ENQ? Scott ================================================================================ Note 4.2 $ENQ goes into MWAIT 2 of 2 OPNNET::CAMERONJ12 7 lines 25-AUG-1992 14:47 -------------------------------------------------------------------------------- Don't forget that RWSCS, RWCLU and co. appear as MWAIT when you are doing a SHOW PROCESS/CONTINUOUS SHOW SYSTEM will show a more detailed cause of the MWAIT. James (DEC) ================================================================================ Note 5.0 POSIX documentation available 5 replies OPNNET::KEOGH 15 lines 25-AUG-1992 09:05 -------------------------------------------------------------------------------- There is a set of VMS POSIX documentation at the OpenVMS stand. This set includes the IEEE standard 1003.1, and the draft standards 1003.1a, 1003.2, 1003.2a and 1003.4. Please feel free to come by and check them out. The documentation set will be available by auction. The doc kit will be available for collection at the OpenVMS stand on Friday afternoon. The proceeds of the auction will go to SIDS. To bid, enter a reply to this note with the amount and your full name. Only cash and personal cheques accepted. Bids close Friday 10:00am. Patrick Keogh ================================================================================ Note 5.1 POSIX documentation available 1 of 5 OPNNET::FRYA22 5 lines 25-AUG-1992 10:03 -< $10 for POSIX Documentation >- -------------------------------------------------------------------------------- Well I'll go first! Let's start low at $10. Andrew Fry. ================================================================================ Note 5.2 POSIX documentation available 2 of 5 OPNNET::CAMERONJ12 0 lines 25-AUG-1992 14:49 -< $11 >- -------------------------------------------------------------------------------- ================================================================================ Note 5.3 POSIX documentation available 3 of 5 OPNNET::GILLARDI20 2 lines 26-AUG-1992 11:15 -< $20 >- -------------------------------------------------------------------------------- Ok I'll bid $20 ================================================================================ Note 5.4 POSIX documentation available 4 of 5 OPNNET::PENINGTONA27 1 line 26-AUG-1992 13:03 -< $30 >- -------------------------------------------------------------------------------- I'll try $30 ================================================================================ Note 5.5 POSIX documentation available 5 of 5 OPNNET::GILLARDI20 2 lines 27-AUG-1992 09:52 -< $50 >- -------------------------------------------------------------------------------- I'll bid gain at $50 ================================================================================ Note 6.0 Remote VAX time access 11 replies OPNNET::HAWLEYD13 "Dave Hawley, IT Manager, Tektron" 20 lines 25-AUG-1992 17:08 -------------------------------------------------------------------------------- I have a simple problem to do with DECnet: We make a DLM connection to our US and European sites using low-speed X.25 (P.S.I.); so effectively we "join our DECnets". This is so we can exchange Message Router info and email, so the connection is usually on and off in under 60 seconds. However, I would really like to be able to determine from the REMOTE VAX node what the VAX system time is; either in ASCII or binary format, I don't care. Accuracy to within a few seconds would be quite acceptable (it's a simple self-correcting world clock function). The limitations are that most of the European sites have PSI security enabled, so I can't do the classic "copy a DCL command procedure to the remote node and run it" routine. In fact, I can't run anything on their machine from mine. I have tried using NCP TELL, but its time stamp is from the local node. I would much prefer to be able to do it simply, preferably with DCL and the standard VAX utilities. Any solutions spring to mind? ================================================================================ Note 6.1 Remote VAX time access 1 of 11 OPNNET::CAMERONJ12 4 lines 25-AUG-1992 17:34 -------------------------------------------------------------------------------- write sys$output f$file_attributes("''node'::netserver.log;","cdt") ? ================================================================================ Note 6.2 Remote VAX time access 2 of 11 OPNNET::HAWLEYD13 "Dave Hawley, IT Manager, Tektroni" 8 lines 25-AUG-1992 17:52 -< I'll try it ... >- -------------------------------------------------------------------------------- I'll try it, but will PSI access allow it? I can't do: $ COPY X.COM node:: It looks like file-level access to interactive users has been denied, so the f$file_attributes may also be barred. No way but to try it, though. Many thanks. ================================================================================ Note 6.3 Remote VAX time access 3 of 11 OPNNET::CAMERONJ12 13 lines 25-AUG-1992 18:06 -------------------------------------------------------------------------------- Re: Note 6.2 by OPNNET::HAWLEYD13 > I'll try it, but will PSI access allow it? I can't do: > > $ COPY X.COM node:: Hmmm. Can you $ COPY node::X.COM [] If not, then yes, you can't use the F$FILE_ATTRIBUTES trick. James ================================================================================ Note 6.4 Remote VAX time access 4 of 11 OPNNET::HAWLEYD13 "Dave Hawley, IT Manager, Tektroni" 8 lines 25-AUG-1992 18:07 -< Worked well up to a point >- -------------------------------------------------------------------------------- OK, I tried it as you suggested. Success in most cases, except where the remote DECNET login has been disabled (removed from UAF, I suspect). This precludes the use of remote login ala your solution. HOWEVER, it is a lot better than I had before, so I will see what I can do with that. Of course, further suggestions are welcome! ================================================================================ Note 6.5 Remote VAX time access 5 of 11 OPNNET::CAMERONJ12 5 lines 25-AUG-1992 18:09 -------------------------------------------------------------------------------- Re: Note 6.4 by OPNNET::HAWLEYD13 > Of course, further suggestions are welcome! Glad to see it worked... ================================================================================ Note 6.6 Remote VAX time access 6 of 11 OPNNET::ROSSR12 "Rollo Ross, Uni SA" 5 lines 26-AUG-1992 09:41 -< A different solution? >- -------------------------------------------------------------------------------- > Of course, further suggestions are welcome! I'm not clear about what you were trying to do. If you are trying to keep your VAX clocks synchronised, do you know about the British Telecom X.29 talking clock? You SET HO/X29 to it, it produces a line of output giving the UK time (and timezone), and disconnects. ================================================================================ Note 6.7 Remote VAX time access 7 of 11 OPNNET::HAWLEYD13 "Dave Hawley, IT Manager, Tektron" 28 lines 27-AUG-1992 10:10 -< REAL WORLD ... >- -------------------------------------------------------------------------------- The X.29 talking clock sounds great, and I have another project in mind now that you have informed me of it. However, the original problem is that we have a number of offices world wide, and they all have (a) differing timezones, and (b) daylight saving regimes. When we go to phone/FAX urgent info it would be of significance to know if the destination office is actually open. Eg. Head office in Portland Oregon can be anywhere from 15 to 19 hours behind us. The zone difference means that in some cases we have to call from our office before 9am here to catch people in the office before 5pm there. At other times we can be as late as 1pm here! Solution: "world" clock that knows the time in Portland every day. Since our VAXes contact each other regularly many times per day, simply ask the other VAX to tell you its time, and make that time available for display to interested parties. By extension, our European sites can be similarly polled. Problem: They are all security mad, so I can't just make a small command procedure to SHOW TIME to SYS$NETWORK, copy it to the non-priv DECnet account then do a remote task run (TYPE xxx"TASK=0 ..."::) to capture the info. X.25 PSI Security is FULLY invoked. Thus, here I am asking the questions ... The one provided is OK so far, and better than I have been able to do. I remain open to any other alternatives. ================================================================================ Note 6.8 Remote VAX time access 8 of 11 OPNNET::MCGILLC8 9 lines 27-AUG-1992 10:15 -< Talk to them ! >- -------------------------------------------------------------------------------- If you guys are all in the same company, then get them to cooperate ! Tell them the problem and ask them to set up a specific DECnet object and command procedure that does exactly what you want (and nothing more). Problem solved. You don't have a technical problem - you have a people problem ! Get your company's act together ! ================================================================================ Note 6.9 Remote VAX time access 9 of 11 OPNNET::CAMERONJ12 10 lines 27-AUG-1992 10:18 -------------------------------------------------------------------------------- Re: Note 6.8 by OPNNET::MCGILLC8 > If you guys are all in the same company, then get them to cooperate ! > Tell them the problem and ask them to set up a specific DECnet object > and command procedure that does exactly what you want (and nothing > more). The way I've solved this people problem in the past is to provide them with a complete kit for them to check out and install. If they believe they don't have to do much, they are more likely to cooperate. ================================================================================ Note 6.10 Remote VAX time access 10 of 11 OPNNET::HAWLEYD13 "Dave Hawley, IT Manager, Tektron" 11 lines 27-AUG-1992 10:44 -< Little guys >- -------------------------------------------------------------------------------- I think you may be right about the people problem, but we also communicate to outside organisations on public networks like the X.25 links we use. We have had instances in the US of network hackers shooting at our VAXes, and our company is high-tech, so the shutters come DOWN. Alas, that also makes inter-company comms difficult too. The problem of time differences in Europe and the US is not half as great as between Australia and those sites. Therefore WE have been the initiators in the "world clock" push. However, it is a "nice-to-have", not absolutely essential. If only the buggers were there to take our phone calls of complaint ... ================================================================================ Note 6.11 Remote VAX time access 11 of 11 OPNNET::ROSSR12 "Rollo Ross, Uni SA" 9 lines 27-AUG-1992 11:46 -< DTE number for talking clock >- -------------------------------------------------------------------------------- > The X.29 talking clock sounds great, and I have another project in mind > now that you have informed me of it. Probably would have helped if I'd told you the number... Here's the number and an example of what happens when you call it: $ set ho/x29 023421920100605 Call connected to remote DTE THU 27/08/92 02:41:12 HRS BST Call cleared by remote DTE ================================================================================ Note 7.0 Oracle V5 on u3100/80 & VMS5.5 1 reply OPNNET::GARRETTR3 18 lines 25-AUG-1992 18:01 -------------------------------------------------------------------------------- We have just purchased a MicroVAX 3100 model 80, with 72Mb memory and VMS 5.5. We currently have Oracle V5 on our existing MicroVAX II and want to move it accross to the new 3100. We also have Oracle V6 to install and start migrating our oracle V5 applications to the oracle V6. 1. What things do I need to watch out for with Oracle V5 on the 3100? 2. Do I need any special VMS or INIT.ORA paramaters? 3. Are there any special little traps I need to look for as well? 4. Any special tips for installing Oracle V6 onto the 3100? Thanks, Robin Garrett ================================================================================ Note 7.1 Oracle V5 on u3100/80 & VMS5.5 1 of 1 OPNNET::POWELLH27 10 lines 26-AUG-1992 14:07 -< Ours went OK. >- -------------------------------------------------------------------------------- I wish you luck! We have moved to V6.0 recently and are now commencing SQL*Forms V2.3 to V3.0 upgrade. Our V6 upgrade went OK but we were almost forced to buy TPO. If you have any problems the hot-line is reasonably good value. ================================================================================ Note 8.0 Compressing VMS mail files. 6 replies OPNNET::MCMORRAND11 2 lines 25-AUG-1992 19:10 -------------------------------------------------------------------------------- Can someone tell me how a system manager can compress all users' VMS mail files? ================================================================================ Note 8.1 Compressing VMS mail files. 1 of 6 OPNNET::GARSOND11 0 lines 25-AUG-1992 19:18 -< put it in SYLOGIN? (-: >- -------------------------------------------------------------------------------- ================================================================================ Note 8.2 Compressing VMS mail files. 2 of 6 OPNNET::CAMERONJ12 7 lines 26-AUG-1992 08:57 -------------------------------------------------------------------------------- Write something that will read through a list of usernames generated from an AUTHORIZE listing, enter mail, set the mail file to the user's file, perform a COMPRESS, and exit. I haven't tried it, am I missing something? James ================================================================================ Note 8.3 Compressing VMS mail files. 3 of 6 OPNNET::GARSOND11 5 lines 26-AUG-1992 09:06 -------------------------------------------------------------------------------- re .-1 You really need to read the mail profile so you know where the user's mail file is (may not be SYS$LOGIN). [Determined disk wasters may have multiple mail files too.] ================================================================================ Note 8.4 Compressing VMS mail files. 4 of 6 OPNNET::BEISSELC15 4 lines 26-AUG-1992 10:02 -< Another way to do it. >- -------------------------------------------------------------------------------- Another kludge is to submit a batch job under each user's name to enter mail, compress and exit. I di dthis before I discovered mail's SET FILE command. ================================================================================ Note 8.5 Compressing VMS mail files. 5 of 6 OPNNET::MCGILLC8 26 lines 27-AUG-1992 10:21 -< More info >- -------------------------------------------------------------------------------- > You really need to read the mail profile so you know where the user's > mail file is (may not be SYS$LOGIN). [Determined disk wasters may have > multiple mail files too.] I concur - open and read SYS$SYSTEM:VMSMAIL_PROFILE.DATA. You will need to look at this file to work out it's layout. It's not too complicated - it has lots of info and each one (if present) consists of a code (a word I think) that says what the next bit of data is, another word that says how long it is, and then the data. For each record (the primary key of which is the username), you need to look for the "maiul subdirectory" field which will be represented by a certain number (dump the file to work it out, or look at the VMS source listings). You will need to apply this info found (if any) to the default directory and devie in the UAF to get the exact location of the file. Then for each file found, do a mail SET FILE then COMPRESS. Watch out though - COMPRESS *MAY* be one of those MAIL commands that set context back to your own file (although I doubt it). For example, you can SET FILE anywhere, but if you type READ/NEW - it's back to yours again ! Craig McGill Pagefile Computing Services (Canberra) ================================================================================ Note 8.6 Compressing VMS mail files. 6 of 6 OPNNET::GARSOND11 8 lines 27-AUG-1992 12:26 -------------------------------------------------------------------------------- re .-1 > I concur - open and read SYS$SYSTEM:VMSMAIL_PROFILE.DATA. There is a callable interface to MAIL so you may want to check whether it can get the information that you need - and avoid hard-coding details of the record layout which are probably not documented and hence subject to change. ================================================================================ Note 9.0 Once upon a time ... 2 replies OPNNET::COMMONSA29 18 lines 25-AUG-1992 19:34 -------------------------------------------------------------------------------- Once upon a time I could sit down with my little uVAX and write some serious (??) software that had a chance of running on some of the bigger boxes. The only difference between my little box and the bigger boxes was really just the money that had changed hands - VMS was VMS. With the advent of POSIX and ALPHA we seem to be getting into the same situation that developers working with UNIX have had to suffer, to produce a product you need: (1) A VAX/VMS version (2) An OPEN-VMS version (VMS + POSIX) (3) An ALPHA/VMS version (4) An ALPHA/OPEN-VMS version All the hardware and variations on software are time and money ... Is this the end of development by small software houses ?? ================================================================================ Note 9.1 Once upon a time ... 1 of 2 OPNNET::CAMERONJ12 5 lines 26-AUG-1992 09:01 -------------------------------------------------------------------------------- Re: Note 9.0 by OPNNET::COMMONSA29 > Is this the end of development by small software houses ?? "Think of it as evolution in action." (Larry Niven, I think) ================================================================================ Note 9.2 Once upon a time ... 2 of 2 OPNNET::KEOGH 25 lines 26-AUG-1992 10:00 -------------------------------------------------------------------------------- > The only difference between my little box and the bigger > boxes was really just the money that had changed hands - VMS was VMS. > > With the advent of POSIX and ALPHA we seem to be getting into the same > situation that developers working with UNIX have had to suffer, to > produce a product you need: > (1) A VAX/VMS version > (2) An OPEN-VMS version (VMS + POSIX) > (3) An ALPHA/VMS version > (4) An ALPHA/OPEN-VMS version There are two distinct cases here. You can continue to send just one executable to your customers, developed in standard OpenVMS on any VAX, just like you do today. However, if you execute this application on an Alpha processor, you won't get the best possible performance. However, it may be acceptable, depending on the application. You would develop in POSIX if you want an open, standards-compliant, portable application. Then you can run (after compliation on the target platform) on any conforming platform, like some UNIX boxes. However, building and deploying Open Systems takes effort, and if your customer demands an "open" environment, then I expect that they will be prepared for the extra effort to support POSIX. ================================================================================ Note 10.0 POSIX and unix tools 1 reply OPNNET::GUNJKOA29 8 lines 26-AUG-1992 12:25 -------------------------------------------------------------------------------- Just a question about POSIX/VMS. Does anyone out there have any hints as to how to port true unix utility software (eg PERL, GNU) to VMS/POSIX? Is it easy, hard, or what? I figure that most unix programmers write NOT to the POSIX spec, but has anyone actually *tried* to convert unix gear? It would be nice to be able to access all that anonymous ftp stuff on unix nodes around the globe ... ================================================================================ Note 10.1 POSIX and unix tools 1 of 1 OPNNET::TONKINP2 10 lines 27-AUG-1992 18:34 -< POSIX where are we going ? >- -------------------------------------------------------------------------------- A question for all of you OPENVMS POSIX people. Is the POSIX user interface the same as UNIX ? If so does this mean the industry has taken two steps forward and three steps backward in user interfaces ??? Are we all going to move the "C" as a programming language ? This again seems like a step backwards ================================================================================ Note 11.0 How do I move a DNS Nameserver from one node to another 1 reply OPNNET::BEGGJ17 "Jeremy Begg, VSM Software Services" 39 lines 27-AUG-1992 09:56 -------------------------------------------------------------------------------- I have recently been given the task of integrating a VAX 4500 into an existing VAXcluster. The nodes presently in the cluster include an 8350, an 8530 and a 6610, all booting off a common HSC system disk. The 6610 has a DSSI adapater which it uses to talk to the 4500 (and all are connected by Ethernet, of course). The 4500 has been booted into the cluster (via its own DSSI system disk) and appears to run OK. I am now in the process of loading layered products onto it. The 4500 will be used mainly for non-interative, background type stuff: DECnet router, PATHWORKS server, e-mail switch, AARNet connection, etc etc. Two of the products to be loaded include DNS and DFS, both presently running on the 6610. I would like to move the DNS from the 6610 to the 4500. I began by installing DNS onto the 4500 and telling it that it was not a new DNS domain (just a new nameserver). I then tried moving the clearinghouse from the 6610 to the 4500, by following the manual: - on the 6610, STOP the clearinghouse and REMOVE it - on the 4500, ADD the clearinghouse file and START it. The ADD command generated one error message (something about no access to perform the attempted operation) but appeared otherwise to work; it served the clearinghouse even when the 6610 DNS was turned off. So I installed DFS on the 4500 and the DNS promptly died. :-( Later on I determined that the DNS clearinghouse object had RWDTC access only for the 6610, hence the error message I saw. I have now re-ADDED the clearinghouse to the 6610 and added RWDTC access for the 4500, but in the meantime I couldn't persuade the 4500 DNS to come to life so I restored the system disk from a backup and will re-install DNS on Monday. Can anyone give me some more suggestions for moving a DNS from one node to another? Thanks, Jeremy Begg jeremy@vsm.com.au VSM Software Services Ph: (08) 414 5074 FAX: (08) 373 4911 ================================================================================ Note 11.1 How do I move a DNS Nameserver from one node to another 1 of 1 OPNNET::BEISSELC15 19 lines 27-AUG-1992 14:26 -< A sorry tale! >- -------------------------------------------------------------------------------- All I can say is Good Luck! I moved a DNS server after a reconfiguration in which a DNS server was removed from a cluster (powered down, but not dismantled). TSC were helpful, but it still took a while. We had to add the removed node back into the cluster to move the clearinghouse. It took a few weeks to do this as one server in another state that had a read-only copy of the root (.) was not running. It turned out that 6 months before this cluster had a VMS v5.5 upgrade applied by restoring ANOTHER site's system disk and fiddling with SCSNODENAME etc. Unfortunately no-one thought about DNS! As I said it took a while. After this experience I wish you luck, and would like to warn the world of the potential problems of running DNS in a network when you do not manage all the DNS servers. ================================================================================ Note 12.0 Pipes in VMS 5 replies OPNNET::DELAINEM14 15 lines 27-AUG-1992 13:03 -------------------------------------------------------------------------------- When will we be able to use pipes (or similar) to cut out the usual file writing stuff that we need... $ mcr authorize show * /full /output=file1.tmp ... $ search file1.tmp "Username","Identifier" /output=file2.temp ... etc. $ mcr authorize show * /full | search "Username","Identifier" | etc.... Mike ================================================================================ Note 12.1 Pipes in VMS 1 of 5 OPNNET::CAMERONJ12 8 lines 27-AUG-1992 13:18 -------------------------------------------------------------------------------- Just did it then; $ POSIX psx> ls -al | grep -e root Admittedly not what you asked for; but it's free isn't it? James ================================================================================ Note 12.2 Pipes in VMS 2 of 5 OPNNET::KEOGH 12 lines 27-AUG-1992 13:49 -< VMS POSIX has it now! >- -------------------------------------------------------------------------------- $ mcr authorize show * /full | search "Username","Identifier" | etc.... becomes $@sys$update:vmsinstal posix010 wherever etc. $posix % dcl mcr authorize show */full | grep "Username|Identifier" Good enuf? ================================================================================ Note 12.3 Pipes in VMS 3 of 5 OPNNET::DELAINEM14 22 lines 27-AUG-1992 14:06 -< Pipes, not POSIX >- -------------------------------------------------------------------------------- Yes, I have already played with POSIX... $ posix/run/input=command_shell It still means that I must put the commands I want to execute into a command shell... and then put the output of the command shell into another file... $ posix/run/input=command_shell/output=temp.file It is a start, but as yet I haven't found out how from Posix I can refer to: o files in my local directory and sys$login o other VMS directories and files It would be much easier if there were pipes in VMS.... Mike. ================================================================================ Note 12.4 Pipes in VMS 4 of 5 OPNNET::DAVISS25 9 lines 27-AUG-1992 14:15 -------------------------------------------------------------------------------- I am not very up on Posix, but... FIFOs (I think that's the correct name) are named, cross process pipes that are implemented for VMS Posix thru a brand new version of the mail box driver... I'd recommend you look there... Scott ================================================================================ Note 12.5 Pipes in VMS 5 of 5 OPNNET::KEOGH 2 lines 27-AUG-1992 14:19 -< exit >- -------------------------------------------------------------------------------- I think you are missing a couple of points. Come around to the OpenVMS stand and I'll show you!