INFO-VAX Fri, 12 Jan 2007 Volume 2007 : Issue 24 Contents: Re: %APB-F-NOIOVEC Re: DECWindows SET/DISPLAY & CREATE/TERM/DETACH problem on Alphaserver DS10L Re: ODS5 and hardlinks Re: ODS5 and hardlinks Re: ODS5 and hardlinks Re: ODS5 and hardlinks Replacement for the TK70 Re: Replacement for the TK70 Re: RUN SYS$SYSTEM:DCL on Itanium VMS 8.2-1 Re: RUN SYS$SYSTEM:DCL on Itanium VMS 8.2-1 strange LINK error Re: strange LINK error Re: Suggestion: SET FILE/SHADOW= Re: Windows on Integrity Podcast Series ---------------------------------------------------------------------- Date: 12 Jan 2007 02:54:56 -0800 From: "Camiel" Subject: Re: %APB-F-NOIOVEC Message-ID: <1168599296.786597.54820@l53g2000cwa.googlegroups.com> mb301@hotmail.com schreef: > Camiel wrote: > > I'm still trying to emulate an AlphaServer, and I've got the SRM > > console up and running. I can issue various show commands, set > > environment variables, buildfru, etc... However, when I try booting, I > > get an error message "%APB-F-NOIOVEC Failed to create IOVEC". I'm > > trying to figure out the cause of this, but it would really help if I > > know what this message means, and what could cause it on a real > > AlphaServer. Has anyone ever seen this message? I can't find it with > > HELP/MESSAGE, nor with Google. > > > > Camiel. > > What emulator are you using? I'm in the process of writing my own emulator... > have you tried getting some more information by doing a conversational > bootstrap Flags? Yes I have tried the additional message flags. That made no difference. Except that with some flags (I don't recall excactly which ones) I get a kernel stack not valid trap before I get any other messages... I've tried both OpenVMS 7.3-1 and 8.3. Camiel. ------------------------------ Date: 12 Jan 2007 07:42:59 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: DECWindows SET/DISPLAY & CREATE/TERM/DETACH problem on Alphaserver DS10L Message-ID: <+rncQTQTtf8y@eisner.encompasserve.org> In article <1168552763.671601.148190@i56g2000hsf.googlegroups.com>, "urbancamo" writes: > Please help! > > I've got a VAXStation 4000/90 and an Alphaserver DS10L. I can > SET/DISPLAY from the VAX and CREATE/TERM/DETACH a DECTerm on my Linux > laptop, but the Alphaserver refuses to open one. I am using OpenVMS > 7.3-1 on both. The DS10L does not have a graphics card installed, the > VAX does. I wasn't aware this would make any difference when starting X > applications on a remote terminal. > Are you sure the Linux X server is allowing the Alpha to connect? Usually xhost is the control mechanism, but I could conceive of a mechanism which uses back-translation of the IP address. Try an explicit xhost+ of the Alpha's IP address, and if that doesn't work, a onetime xhost+ (allowing everyone in). I'm assuming you're using the same protocol to access the Alpha from Linux as you do the VAX, not getting SSH on one and TELNET on the other. ------------------------------ Date: 12 Jan 2007 05:21:11 -0800 From: "AEF" Subject: Re: ODS5 and hardlinks Message-ID: <1168608071.271017.87730@a75g2000cwd.googlegroups.com> JF Mezei wrote: > AEF wrote: > > But did you see the manual reference? > > > Yes I did. The info in it was about teh same as what I had already read. It > doesn't discuss why hardlinks should or should not be enabled. Well, that depends on what you want. If you want regular VMS behavior, don't enable hard links. If you want Unix-like behavior, enable them. > > If VMS engineering decided that hardlinks were not the default/automatic, > then it means that there are some reasons why someone would want to have a > disk without hardlinks. They are probably not enabled by default for backward compatibility, I assume. > > So what is the caveat with using hardlinks ? Are there any ? Or is it just A file you delete that others have created directory entries for will continue to exist after you delete it and it will be charged against your disk quota. This is discussed in the manual reference I gave. > a question of improving the handling of the deletion of files that have > many pointers to them ? That depends on whether you consider hard links an improvement. My recommendation is that if you're not sure what to do, don't enable hard links. If you have a good reason to use them, then enable them. AEF ------------------------------ Date: 12 Jan 2007 05:12:49 -0800 From: "AEF" Subject: Re: ODS5 and hardlinks Message-ID: <1168607568.930268.66680@38g2000cwa.googlegroups.com> David J Dachtera wrote: > JF Mezei wrote: > > > > That newfangled ODS5 beast is all foreign to me... I don't yet have 20 > > years experience with VMS (this comes in may this year :-), so pardon my > > newbieness in this matter... > > > > I've tried to find documentation that *really* described what hardlinks are > > all about. Best I could come up with is in comp.os.vms. > > > > Can someone confirm that SET VOLUME/VOLUME_CHARACTERISTICS=hardlinks (and > > the equivalent INIT /VOLUME_CHAR=HARDLINKS changes nothing in usage for the > > disk (AKA SET FILE/ENTER SET FILE/REMOVE ) except that it implements some > > form of usage count for each file so that a DELETE won't actually release > > the blocks unless there are no aliases left for the file ? > > > > Is this the only difference ? > > Well, here's what I understand of hard- and soft-links. Perhaps someone more > UN*X-savvy can chime in and correct me... > > As implemented on VMS, what we call a "hardlink" is really more closely akin to > a "softlink" or "symbolic link". It's just like an alias, except that a link > counter in the file header gets incremented or decremented (as appropriate). > Unlike UN*X, however, there is no "chain of links" that can be followed to find > all of the links to a file(-chain). A hard link on VMS is like an alias created with SET FILE/ENTER, except that if you delete the primary entry, the file lives on. You have to delete all directory entries for a file to delete the file when hard links are enabled. > > As I understand it, were VMS to implement something more closely akin to a UN*X > hardlink, we would actually allocate a second file header ("inode" in UNX* > parlance) to point to the same (chain of) extent(s) ("file-chain" in UN*X > parlance). In that case, DELETE (or the filesystem ACP) would have to implelment > a check of all of the file headers to ensure that all the headers referencing > the same (chain of) extent(s) get processed on an appropriately qualified DELETE > command. VERIFY (ANALYZE/DISK) would have to be "smartened" to track this as > well. There is only one file header with hard links. The are, however, multiple directory entries for the same file. So you might have BLAH.BLAH (1023,23,0) BLOW.BLOW (1023,23,0) [some-other-directory]BLEE.BLEE (1023,23,0) If these were regular VMS aliases, deleting the primary entry would delete the file, leaving the aliases hanging. If hard links were enabled on the volume, the file would live until all directory entries were deleted. I know not what these file-chains you refer to are. > > > If one does a SET VOLUME on an existing disk, must one do a ANA/DISK/REPAIR > > to build those number of links to each file ? Yes. > Good question... Though, I'd expect that the default behaviour is probably to > set a link count of one(1) even when a VOLUME is not SET for HARDLINKS. Look for hard links in the index of http://h71000.www7.hp.com/doc/82FINAL/aa-pv5mj-tk/aa-pv5mj-tk.HTMl (System manager's manual vol. 1: essentials for more info.) > -- > David J Dachtera > dba DJE Systems > http://www.djesys.com/ > [...] AEF ------------------------------ Date: 12 Jan 2007 07:56:54 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: ODS5 and hardlinks Message-ID: In article <45A6F886.3B5AA388@spam.comcast.net>, David J Dachtera writes: > > As implemented on VMS, what we call a "hardlink" is really more closely akin to > a "softlink" or "symbolic link". It's just like an alias, except that a link > counter in the file header gets incremented or decremented (as appropriate). > Unlike UN*X, however, there is no "chain of links" that can be followed to find > all of the links to a file(-chain). > What is this -chain you refer to? A command option, or just a way of saying "file or file-chain"? My studies of Ultrix32 internals leads me to beleive Bill Todd's response is more accurate: both VMS and UNIX keep a single data structure to track the file (VMS file header, UNIX inode). Hard links are multipe, indistinguishable directory entires for the same copy of that structure. This data structure and the file contents do go away unless the link count (or whatever VMS calls it) in the strucure goes to 0. Unix soft links are a different kind of inode that says the file is actually found via some other directory entry. VMS aliases are distinguishable directory entries that don't increment the link count. The data structure (original UNIX inode or VMS header) and file content go away when the original file is removed/deleted since the link count then goes to 0, leaving soft links or aliases that point to nothing. The biggest functional difference between UNIX soft links and VMS aliases is that the latter cannot span volumes. The biggest implementation difference is that the latter does not create a different kind of header. I suspect future VMS file system work will need to include a full soft link capability to improve the portability of UNIX applications. Adding a special kind of file header is a straightforward, but non-trivial approach. ------------------------------ Date: 12 Jan 2007 07:59:36 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: ODS5 and hardlinks Message-ID: In article , "John E. Malmberg" writes: > > A softlink on UNIX most closely corresponds to a logical name on OpenVMS. Whoah, Nelly. Don't go there. File system entries on UNIX haven't the capabilities of logical names. Did you ever see a group-wide soft link? A per-process soft link? Every process on UNIX that looks at the same soft link gets the same data. Not necessarily so with logical names (what's your value of SYS$OUTPUT ?). ------------------------------ Date: 12 Jan 2007 07:55:36 -0800 From: tomarsin2015@comcast.net Subject: Replacement for the TK70 Message-ID: <1168617334.461821.56340@51g2000cwl.googlegroups.com> Hello I have a number of broken tk70 tape drives. According to some older post, if the lights on the tk70 start blinking fast then the drive is no good. Well I am looking for either a way to "fix" these drives or found a cheap replacement. Its not really work the money to find a scsi tape controller that may support the tk70 drive (or is it). The prices that some dealers are asking for a replacement is high also (1 dealer was asking almost a 500.00 dollars just for the drive). So does anybody have a low price solution. I was thinking of trying to do a backup across the net, the systems are on the network, but not part of the cluster. thanks phillip ------------------------------ Date: Fri, 12 Jan 2007 11:40:19 -0600 (CST) From: sms@antinode.org (Steven M. Schweda) Subject: Re: Replacement for the TK70 Message-ID: <07011211401966_2020028F@antinode.org> From: tomarsin2015@comcast.net > I have a number of broken tk70 tape drives. According to some older > post, if the lights on the tk70 start blinking fast then the drive is > no good. Well I am looking for either a way to "fix" these drives or > found a cheap replacement. Its not really work the money to find a scsi > tape controller that may support the tk70 drive (or is it). The prices > that some dealers are asking for a replacement is high also (1 dealer > was asking almost a 500.00 dollars just for the drive). So does anybody > have a low price solution. I was thinking of trying to do a backup > across the net, the systems are on the network, but not part of the > cluster. I see TK70 drives on Ebay with some frequency. A KZQSA card just sold for around $25 (which may be pretty unusual), and newer DLT drives are often quite cheap (shipping exceeds winning bid). I've recently converted from yet another bad Exabyte 8mm drive to a DLT4000 for my backups, and then I found a bargain on a DLT7000, too. DLT III tapes seem to be pretty cheap (less than $2 each), too. (And MKDRIVER knows how to enable compression on one, too, unlike on the Exabyte.) For $500, you can buy a whole lot of junk, not all of which is likely to be all bad. ------------------------------------------------------------------------ Steven M. Schweda sms@antinode-org 382 South Warwick Street (+1) 651-699-9818 Saint Paul MN 55105-2547 ------------------------------ Date: Fri, 12 Jan 2007 15:31:42 +0000 From: Roy Omond Subject: Re: RUN SYS$SYSTEM:DCL on Itanium VMS 8.2-1 Message-ID: Trefor wrote: > On Alpha VMS 7-3.2 > > RUN SYS$SYSTEM:DCL > > is fine. We use it (with lots of parameters) to start another process > to run a command file. > > But ... on Itanium VMS 8.2-1 > > RUN SYS$SYSTEM:DCL > > %DCL-W-ACTIMAGE, error activating image SYS$SYSTEM:DCL > -CLI-E-IMGNAME, image file DSA0:[SYS0.SYSCOMMON.][SYSEXE]DCL.EXE > -SYSTEM-F-ACCVIO, access violation, reason mask=25, virtual > address=000000007FFD1160, PC=000000000000000E, PS=7FF97EA4 > > Any idea why? Is it an Itanium or VMS 8.2-1 problem? Y'know ... in my 25 years in VMS, I have never seen anyone do this ($ run sys$system:dcl). Where is that documented ? What does the documentation (if there is any) say it should do ? For your information, I tried it on a VMS 8.3 system on Alpha: $ sh sys/noproc OpenVMS V8.3 on node WIZARD 12-JAN-2007 15:27:15.16 Uptime 360 03:46:50 $ mc dcl %DCL-W-ACTIMAGE, error activating image DCL -CLI-E-IMGNAME, image file DSA0:[SYS0.SYSCOMMON.][SYSEXE]DCL.EXE -SYSTEM-F-ACCVIO, access violation, reason mask=25, virtual address=000000007FFD1160, PC=0000000000000003, PS=7FF9DEA3 Why are you not using RUN/DETACH ? Roy Omond Blue Bubble Ltd. ------------------------------ Date: Fri, 12 Jan 2007 15:54:48 -0000 From: "Richard Brodie" Subject: Re: RUN SYS$SYSTEM:DCL on Itanium VMS 8.2-1 Message-ID: "Roy Omond" wrote in message news:eo89l0$7sr$1$8300dec7@news.demon.co.uk... > Y'know ... in my 25 years in VMS, I have never seen anyone do > this ($ run sys$system:dcl). I tried it once on a VAX. As I recall it deleted the current process. ------------------------------ Date: 12 Jan 2007 04:47:15 -0800 From: "Pierre" Subject: strange LINK error Message-ID: <1168606034.969186.118270@i15g2000cwa.googlegroups.com> while linking a C program, I get this error %LINK-W-MULDEF, symbol decc$txsnprintf multiply defined in module DECC$SHR_EV56 file SYS$COMMON:[SYSLIB]DECC$SHR_EV56.EXE;1 here are some info on the context, as CC see them: These macros are in effect at the start of the compilation. ----- ------ --- -- ------ -- --- ----- -- --- ------------ __G_FLOAT=0 __DECC=1 vms=1 VMS=1 __32BITS=1 __PRAGMA_ENVIRONMENT=1 __CRTL_VER=70320000 __vms_version="V7.3-2 " CC$gfloat=0 __X_FLOAT=1 vms_version="V7.3-2 " __DATE__="Jan 12 2007" __STDC_VERSION__=199901L __DECC_MODE_RELAXED=1 __DECC_VER=60590001 __VMS=1 __ALPHA=1 VMS_VERSION="V7.3-2 " __IEEE_FLOAT=1 __VMS_VERSION="V7.3-2 " __STDC_HOSTED__=1 __TIME__="12:45:45" __Alpha_AXP=1 __VMS_VER=70320022 __BIASED_FLT_ROUNDS=2 __INITIAL_POINTER_SIZE=0 __STDC__=1 _IEEE_FP=1 __LANGUAGE_C__=1 __vms=1 __alpha=1 __D_FLOAT=0 any idea ? TIa, Pierre. ------------------------------ Date: Fri, 12 Jan 2007 11:29:42 -0600 (CST) From: sms@antinode.org (Steven M. Schweda) Subject: Re: strange LINK error Message-ID: <07011211294197_2020028F@antinode.org> From: "Pierre" > while linking a C program, I get this error _Which_ C program? > %LINK-W-MULDEF, symbol decc$txsnprintf multiply defined > in module DECC$SHR_EV56 file > SYS$COMMON:[SYSLIB]DECC$SHR_EV56.EXE;1 That's not strange. Knowing nothing about the program, I'd guess that it was developed on an older system which lacked snprintf(), so it includes its own, and that one is bumping into the one in your newer C run-time library. Note the conditionality in SYS$COMMON:[DECC$LIB.REFERENCE.DECC$RTLDEF]STDIO.H: #if __CRTL_VER >= 70312000 int snprintf(char *__s, __size_t __n, const char *__format, ...); int vsnprintf(char *__s, __size_t __n, const char *__format, __va_list __ap); #endif while you admit that "__CRTL_VER=70320000". > any idea ? Re-compile the source. If it doesn't already have "#if __CRTL_VER >= 70312000"-"#endif" around its own snprintf(), you should add it. If you don't have the source, you might need to link it against an older C RTL. ------------------------------------------------------------------------ Steven M. Schweda sms@antinode-org 382 South Warwick Street (+1) 651-699-9818 Saint Paul MN 55105-2547 ------------------------------ Date: 12 Jan 2007 08:07:00 -0600 From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) Subject: Re: Suggestion: SET FILE/SHADOW= Message-ID: <3k31K9C7W7IK@eisner.encompasserve.org> In article <45A6F999.808761E5@spam.comcast.net>, David J Dachtera writes: > > My approach would be to have my system disk SYSTARTUP_VMS do only this: > > 1. Make sure the cluster-common files disk is MOUNTed. > 2. Invoke a "SYSTARTUP.COM" from the cluster-common files disk. > > That way, I'd only have ONE cluster-wide place to maintain such things. > I never had a cluster so homogeneous it was worth that approach, there would have been too many IF's for special processing per node. But I did have all my SYSTARTUP_VMS.COM in SYS$COMMON: and put in just a couple of IF for just a few of the satellites. The satellites tended to boot from servers having related configurations. And I got real good at writing EDT scripts and invoking them from SYSMAN so I could make simultaneous changes across all the boot servers. SYSMAN will look at logical name lists so you can quickly set environment to just the servers. ------------------------------ Date: Fri, 12 Jan 2007 09:16:36 -0500 From: "Kenneth Farmer" Subject: Re: Windows on Integrity Podcast Series Message-ID: <45a7983f$0$5199$4c368faf@roadrunner.com> So you don't have to fill in their form... http://www.infoworld.com/podcasts/windowsonintegrity.html Ken _______________________________________ Kenneth R. Farmer < 336-736-7376 www.OpenVMS.org | dba.OpenVMS.org | dcl.OpenVMS.org | vpn.OpenVMS.org "David J Dachtera" wrote in message news:45A6F439.B03DD377@spam.comcast.net... > From my daily Infoworld newsletter... > > WINDOWS ON INTEGRITY PODCAST SERIES > > Enhance business performance. > Find out how your data center can achieve more secured availability, > simplified management, and flexible capacity by tuning into a series of > on-demand Webcasts and podcasts that will show how Windows on Integrity > Servers can help your organization meet its critical business needs. > Click here to register for this series, sponsored by HP: > > http://reg.itworld.com/servlet/Frs.frs?Context=LOGENTRY&Source=IFWnesletterpodcast&Source_BC=13&Script=/LP/10009192/reg > > (Sorry if the long URL wraps.) > > I wonder about an OpenVMS ON INTEGRITY/SUPERDOME PODCAST SERIES... ? > > -- > David J Dachtera > dba DJE Systems > http://www.djesys.com/ > > Unofficial OpenVMS Marketing Home Page > http://www.djesys.com/vms/market/ > > Unofficial Affordable OpenVMS Home Page: > http://www.djesys.com/vms/soho/ > > Unofficial OpenVMS-IA32 Home Page: > http://www.djesys.com/vms/ia32/ > > Unofficial OpenVMS Hobbyist Support Page: > http://www.djesys.com/vms/support/ ------------------------------ End of INFO-VAX 2007.024 ************************