1 INFO-VAX	Tue, 07 Sep 2004	Volume 2004 : Issue 496       Contents:1 Re: cluster member crashes during boot (again...) 1 Re: cluster member crashes during boot (again...)  FTP Mirror Freeware  Re: Future of Decnet Re: Future of Decnet Re: Future of DecnetB Re: How to monitor RAM usage and automatically send warning email? Marketing of tech into campuses  Unscribe! Re: VMS Cluster and IP Subnetting ! Re: VMS Cluster and IP Subnetting ! Re: VMS Cluster and IP Subnetting  Re: VMS marketing opportunity  Re: VMS marketing opportunity  Re: VMS marketing opportunity   F ----------------------------------------------------------------------   Date: 6 Sep 2004 20:40:31 +0200 . From: huber@NOBODY-mppmu.mpg.de (Joseph Huber): Subject: Re: cluster member crashes during boot (again...)+ Message-ID: <ABO$cPo6gMPA@vms.mppmu.mpg.de>   W In article <m33c1vi9rs.fsf@pixie.isrnet>, Rodrigo Ventura <yoda@isr.ist.utl.pt> writes:  > E > I see. In the master node, which has 64MB, the SHOW MEMORY commands D > tells that 3441 pages are permanently allocated to OpenVMS. If theC > page size is 8K, this means 26MB. In a 32MB machine, 5MB seems an F > awfully small space for user memory... Although, this master node isB > running DECwindows, while I don't want DECwindows running on the
 > satellites.   ; First of all of course You have to be able to boot at all,  C if You have the bootable distribution CD, try to boot that. It is a  minimally configured system.  E Then of course, don't  start DECwindows: either use SYSMAN STARTUP to F see which startup file to remove, or  put in the systartup_vms.com for those nodes:1 $ DEFINE/SYSTEM/EXEC DECW$IGNORE_WORKSTATION TRUE 1 to inhibit DECW server components startup, and/or 1 $ DEFINE/SYSTEM/EXEC DECW$IGNORE_DECWINDOWS TRUE   to inhibit DECW startup at all.  > F > Is it possible to tune the system parameters such that VMS uses lessH > memory? One problem I had with DECwindows was lack of free GBLSECTION;F > so I had to modify MODPARAMS.DAT to add a few more. Since this files? > seems common to all nodes, I could reduce this (and/or other)  > parameter(s).   J MODPARAMS.dat is in SYS$SPECIFIC:[SYSEXE], i.e. each node can has it's own parameter set.B Also see the comments in modparams.dat (or the template file): it B includes either a bootnode specific or a satellite-specific common) file, so You can group system-parameters.     --  6    Joseph Huber, Muenchen  http://www.huber-joseph.de/   ------------------------------  $ Date: Mon, 6 Sep 2004 23:21:36 +0100< From: "Alex Daniels" <AlexNOSPAMTHANKSDaniels@themail.co.uk>: Subject: Re: cluster member crashes during boot (again...)6 Message-ID: <413ce2e3$0$22749$db0fefd9@news.zen.co.uk>  9 "Rodrigo Ventura" <yoda@isr.ist.utl.pt> wrote in message  # news:m3k6v7if77.fsf@pixie.isrnet...  > A > This time I added with SYSGEN two swapfiles and one pagefile in G > [SYSn.SYSEXE], and booted with -fl 0,20000, but no avail: it crashes! > > It is so verbose that I can't figure out exactly when/how it/ > crashes. The anal/crash is not very explicit:  >   L The reason I suggested you boot with -fl x,20000 was not to stop the crash, - but to get the verbose output, as you put it.   J You mentioned you have no remote console access, so when it crashes press J HOLD if you have a VMS keyboard, or otherwise CTRL/S, then transcribe and L post the two lines of the 'verbose output' preceding the start of the crash.  I Also go into SYSGEN on the two booted nodes, that are in the cluster and   post the output of...    show shadowing show shadow_sys_unit show shadow_sys_disk show shadow_max_unit  I Then boot the third one (that is crashing) to SYSBOOT and again post the  K output of the above, along with "show vaxcluster", "show niscs_load_pea0".  K Then do 'show scsnode' and check it's not the nodename of one of the other   boxes.   Alex     ------------------------------  $ Date: Mon, 6 Sep 2004 18:37:25 -0400# From: "Hal Kuff" <kuff@comcast.net>  Subject: FTP Mirror Freeware- Message-ID: <chiora$49v@library1.airnews.net>   < Hi, anyone know who the current maintainer of ftp_mirror is?   ------------------------------   Date: 6 Sep 2004 19:52:55 -0500 - From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: Future of Decnet 3 Message-ID: <igz+OdUUaqCT@eisner.encompasserve.org>   W In article <m3vferiic6.fsf@pixie.isrnet>, Rodrigo Ventura <yoda@isr.ist.utl.pt> writes: B >>>>>> "Fabio" == Fabio Cardoso <fabiopenvms@yahoo.com.br> writes: > B >     Fabio> May be its a good solution in case of lack of Decnet; >     Fabio> Routers.  Does it support DDCMP for instance ?  > E > I have no idea of what is that, but looking for that acronym in the 3 > project FAQ (July 2004) one can find this answer:  > H > ---------------------------------------------------------------------- > 1.8 Does it do DDCMP?  > H > We don't currently support DDCMP but some work is going on (slowly) toD > add it to the feature list. If you need DDCMP then don't hold your6 > breath it could be a very long time before it works.  ( Alpha VMS does not support DDCMP either.   ------------------------------   Date: 6 Sep 2004 19:52:25 -0500 - From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: Future of Decnet 3 Message-ID: <fvXEmCAChjeZ@eisner.encompasserve.org>   b In article <chgj8g$7nq$1@pcls4.std.com>, moroney@world.std.spaamtrap.com (Michael Moroney) writes:a > Roy Omond <Roy.Omond@BlueBubble.UK.Com> wrote in message news:<2primiFoo3r6U1@uni-berlin.de>... C >> Hint: think about how the echoing of characters you type on your C >> keyboard appears on the screen.  Then think how CTERM might have B >> implemented this in order to reduce network traffic (bearing in< >> mind that it is a very, very clever piece of software :-), >> The rest is left as an exercise etc. etc. > L > Not just the local echo is "smart".  Being able to handle single characterJ > reads as well as bursts of data was harder than it looks on VMS.  I onceL > had to write software that captured and processed output of another deviceJ > connected to a serial line of a Microvax II.  The easiest way to captureK > such data was 1 character $QIO reads, right?  Yes - except a MV2 couldn't J > keep up at 9600 baud (~960 QIOs per second).  Had to implement a bizarreK > hodgepodge where a single character read would be followed by a series of I > large reads with zero timeout, and such a read getting zero characters  H > would be followed by a 1 character (no timeout) read, repeat sequence.J > CTERM, on the other hand can handle large data bursts on a microvax justJ > fine with little cpu overhead.  I still don't know exactly what trick it > uses.   - CTERM is inside the driver (on the host end).    ------------------------------  % Date: Mon, 06 Sep 2004 21:51:37 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net> Subject: Re: Future of Decnet + Message-ID: <413D2239.7CE66040@comcast.net>    Rodrigo Ventura wrote: > C > >>>>> "Fabio" == Fabio Cardoso <fabiopenvms@yahoo.com.br> writes:  > B >     Fabio> May be its a good solution in case of lack of Decnet; >     Fabio> Routers.  Does it support DDCMP for instance ?  > E > I have no idea of what is that, but looking for that acronym in the 3 > project FAQ (July 2004) one can find this answer:  > H > ---------------------------------------------------------------------- > 1.8 Does it do DDCMP?  > H > We don't currently support DDCMP but some work is going on (slowly) toD > add it to the feature list. If you need DDCMP then don't hold your6 > breath it could be a very long time before it works. > E > If you want to help with the development then please contact us via @ > the mailing list and I'll tell you what thoughts I have on the
 > subject.H > ----------------------------------------------------------------------  G DDCMP is simply DECnet over a serial line, synchronous or asynchronous. ? Think point-to-point links, much the way that Multinet supports  DECnet-IV over TCP/IP.   D.J.D.   ------------------------------  % Date: Mon, 06 Sep 2004 21:55:09 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>K Subject: Re: How to monitor RAM usage and automatically send warning email? + Message-ID: <413D230D.399A5A95@comcast.net>    Andreas Gruhl wrote: >  > Dejan Terzic wrote:  >  > > Hi!  > > @ > > I need to create command procedure which will monitor systemG > > parameters (RAM memory usage) for OpenVMS cluster consisting of two H > > nodes and if it drops below 70% send a warning e-mail to a specifiedI > > address. It can be a batch job which will start every 10 minutes. But H > > I don't know how to extract RAM usage to a variable and how to start< > > mail utility from command procedure. Can anyone help me? > 2 > Here is a small example (working on VMS 7.3-2) : >  > $ OPEN/WRITE F MEMAIL.TMP 9 > $ WRITE  F "Memory Size    = ", F$GETSYI("TOTAL_PAGES") 8 > $ WRITE  F "Free Pages     = ", F$GETSYI("FREE_PAGES")< > $ WRITE  F "Modified Pages = ", F$GETSYI("MODIFIED_PAGES") > $ CLOSE  F2 > $ MAIL /SUBJECT="Memory Size"  MEMAIL.TMP  GRUHL  B This variant should work better on V7.3-2 than on anything earlier0 (command line length - can probably be reduced):   $ pipe -D  ( WRITE sys$output "Memory Size    = ", F$GETSYI("TOTAL_PAGES") ; -C    WRITE sys$output "Free Pages     = ", F$GETSYI("FREE_PAGES") ; - H    WRITE sys$output "Modified Pages = ", F$GETSYI("MODIFIED_PAGES")) | -/   MAIL /SUBJECT="Memory Size"  sys$pipe:  GRUHL    D.J.D.   ------------------------------  % Date: Tue, 07 Sep 2004 01:03:14 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> ( Subject: Marketing of tech into campuses, Message-ID: <413D40F1.E29AB61B@teksavvy.com>  M http://news.com.com/Big+tech+on+campus/2100-1025_3-5347538.html?tag=nefd.lede   J Very good article about how many technology companies are fighting to haveM their technologies installed and used by students on campuses. (Remember when ( VMS was prevalent in universities :-) ?&   ------------------------------  $ Date: Mon, 6 Sep 2004 12:56:18 -0700% From: VMS Mangler <bill@lacounte.com>  Subject: Unscribe 4 Message-ID: <f06110400bd6271482323@[165.121.28.214]>   unsubscribe info-vax   ------------------------------  * Date: Mon, 6 Sep 2004 19:03:39 +0000 (UTC)6 From: peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER)* Subject: Re: VMS Cluster and IP Subnetting1 Message-ID: <newscache$6bvm3i$nek1$1@news.sil.at>   S In article <chi6qq$f6r$1@news3.tilbu1.nb.home.nl>, Dirk Munk <munk@home.nl> writes: O >Unless this problem has been resolved in the latest eco's of TCP/IP services,  L >you may encounter a problem with the routing configuration. For some silly P >reason TCPIP$ROUTE.DAT is placed in SYS$SYSDEVICE:[VMS$COMMON.SYSEXE], and not $ >in a SYS$SPECIFIC directory branch.  D You can redirect them whereto you like them to reside with logicals.  R >                                    Creating a default route for cluster members P >in different subnets is impossible this way. Apparently the IP engineers never # >thought of multi-site clusters....   D No problem. The files are indexed with the SCSNODE name and therefor= nodespecific even though per default placed in SYS$COMMON ...    --   Peter "EPLAN" LANGSTOEGER % Network and OpenVMS system specialist  E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------  % Date: Tue, 07 Sep 2004 00:06:42 +0200  From: Dirk Munk <munk@home.nl>* Subject: Re: VMS Cluster and IP Subnetting2 Message-ID: <chin1m$7u0$1@news3.tilbu1.nb.home.nl>    Peter 'EPLAN' LANGSTOEGER wrote:U > In article <chi6qq$f6r$1@news3.tilbu1.nb.home.nl>, Dirk Munk <munk@home.nl> writes:  > P >>Unless this problem has been resolved in the latest eco's of TCP/IP services, M >>you may encounter a problem with the routing configuration. For some silly  Q >>reason TCPIP$ROUTE.DAT is placed in SYS$SYSDEVICE:[VMS$COMMON.SYSEXE], and not  % >>in a SYS$SPECIFIC directory branch.  >  > F > You can redirect them whereto you like them to reside with logicals. >  > R >>                                   Creating a default route for cluster members Q >>in different subnets is impossible this way. Apparently the IP engineers never  $ >>thought of multi-site clusters.... >  > F > No problem. The files are indexed with the SCSNODE name and therefor? > nodespecific even though per default placed in SYS$COMMON ...  >  Are you sure ?? > It is a indexed file, but I can't find the SCSNODE name in it:   (dump /rec)   P Dump of file SYS$SYSDEVICE:[VMS$COMMON.SYSEXE]TCPIP$R on  7-SEP-2004 00:05:15.40               OUTE.DAT;1: File ID (7092,1,0)   End of file block 126 / Allocated 128  A Record number 1 (00000001), 546 (0222) bytes, RFA(0004,0000,0008)   =   20202020 20202020 20302E30 2E302E30 0.0.0.0          000000 =   20202020 20202020 20202020 20202020                  000010 =   20202020 20202020 20202020 20202020                  000020 =   20202020 20202020 20202020 20202020                  000030 =   20202020 20202020 20202020 20202020                  000040 =   20202020 20202020 20202020 20202020                  000050 =   20202020 20202020 20202020 20202020                  000060 =   20202020 20202020 20202020 20202020                  000070 =   20202020 20202020 20202020 20202020                  000080 =   20202020 20202020 20202020 20202020                  000090 =   20202020 20202020 20202020 20202020                  0000A0 =   20202020 20202020 20202020 20202020                  0000B0 =   20202020 20202020 20202020 20202020                  0000C0 =   20202020 20202020 20202020 20202020                  0000D0 =   20202020 20202020 20202020 20202020                  0000E0 =   23202020 20202020 20202020 20202020                # 0000F0 =   00000000 00000000 00000000 00000000 ................ 000100 =   00000000 00000000 00000000 00000000 ................ 000110 =   312E3033 312E3032 312E3731 32000000 ...217.120.130.1 000120 =   20202020 20202020 20202020 20202020                  000130 =   20202020 20202020 20202020 20202020                  000140 =   20202020 20202020 20202020 20202020                  000150 =   20202020 20202020 20202020 20202020                  000160 =   20202020 20202020 20202020 20202020                  000170 =   20202020 20202020 20202020 20202020                  000180 =   20202020 20202020 20202020 20202020                  000190 =   20202020 20202020 20202020 20202020                  0001A0 =   20202020 20202020 20202020 20202020                  0001B0 =   20202020 20202020 20202020 20202020                  0001C0 =   20202020 20202020 20202020 20202020                  0001D0 =   20202020 20202020 20202020 20202020                  0001E0 =   20202020 20202020 20202020 20202020                  0001F0 =   20202020 20202020 20202020 20202020                  000200 =   20202020 20202020 20202020 20202020                  000210 =                                  2020   .............. 000220    ------------------------------  % Date: Mon, 06 Sep 2004 22:11:05 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>* Subject: Re: VMS Cluster and IP Subnetting+ Message-ID: <413D26C9.70718839@comcast.net>    Andrew Rycroft wrote:  >  > Hi,  > 9 > We have 2 sites that each run on different IP sub nets.  > G > We are now going to implement an OpenVMS cluster between the 2 sites. F > The first site will be in one subnet, and the second site in anotherF > subnet. Will this cause problems for the VMS cluster, and additionalB > routing overhead ? We do have an IP alias. Are there any generalD > rules/ best business practises for configuring TCP/IP in a cluster > environment ?    O.k.    C I guess the first myth we need to lay to rest is that subnetting is H somehow a physical concept. It is not. It is a logical concept. MultipleE subnets can exist on the same ethernet segment ("between" routers, or 5 between termini if no routers exist in the scenario).   = The second myth we need to lay to rest is that TCP/IP has any > relationship OpenVMS clustering. It doesn't. Clustering uses aH proprietary, non-routable (suite of) protocol(s) known (collectively) asH SCS: System Communication Services. SCS must be bridged transparently atG the highest available priority between ethernet segments, regardless of C what other ethernet protocols co-exist on the same segments or data  path.   C Problems will arise if latency or packet-loss are beyond tolerance.   @ OpenVMS clusters exist quite happily where no TCP/IP (or network= protocol other than SCS) is present - no relationship exists.    D.J.D.   ------------------------------   Date: 6 Sep 2004 19:51:51 -0500 - From: Kilgallen@SpamCop.net (Larry Kilgallen) & Subject: Re: VMS marketing opportunity3 Message-ID: <Ivp7JrFogzOf@eisner.encompasserve.org>   ` In article <413BC679.CE2FCBFF@comcast.net>, David J Dachtera <djesys.nospam@comcast.net> writes: > JF Mezei wrote:  >>  O >> RIGHT NOW would be the best time to buys VMS specific ads in any/all Florida M >> newspapers extolling VMS's disaster tolerance virtues and VMS's ability to 9 >> have a single cluster spanning hundreds of kilometres.  > H > Well, actually, people are going to be focussed on recovery right now.  C The major advantage VMS Clustering (or MVS data vaulting) brings is C based on the notion that the same disaster is not going to hit more C than one site.  Considering the limited distances supported for VMS D clusters, if I were in Florida I would invest in one very strong and$ well protected (UPS, etc.) building.   ------------------------------  % Date: Mon, 06 Sep 2004 21:58:54 -0500 2 From: David J Dachtera <djesys.nospam@comcast.net>& Subject: Re: VMS marketing opportunity+ Message-ID: <413D23EE.FBCC2638@comcast.net>    Larry Kilgallen wrote: > b > In article <413BC679.CE2FCBFF@comcast.net>, David J Dachtera <djesys.nospam@comcast.net> writes: > > JF Mezei wrote:  > >>Q > >> RIGHT NOW would be the best time to buys VMS specific ads in any/all Florida O > >> newspapers extolling VMS's disaster tolerance virtues and VMS's ability to ; > >> have a single cluster spanning hundreds of kilometres.  > > J > > Well, actually, people are going to be focussed on recovery right now. > E > The major advantage VMS Clustering (or MVS data vaulting) brings is E > based on the notion that the same disaster is not going to hit more E > than one site.  Considering the limited distances supported for VMS F > clusters, if I were in Florida I would invest in one very strong and& > well protected (UPS, etc.) building.  H True. Then again, even given such a massive storm system, it is possibleA that a remote cluster site outside of Frances's reach may survive  relatively uneffected.   D.J.D.   ------------------------------   Date: 6 Sep 2004 22:29:12 -0500 - From: Kilgallen@SpamCop.net (Larry Kilgallen) & Subject: Re: VMS marketing opportunity3 Message-ID: <GGJaXxGzyYaw@eisner.encompasserve.org>   ` In article <413D23EE.FBCC2638@comcast.net>, David J Dachtera <djesys.nospam@comcast.net> writes: > Larry Kilgallen wrote:  F >> The major advantage VMS Clustering (or MVS data vaulting) brings isF >> based on the notion that the same disaster is not going to hit moreF >> than one site.  Considering the limited distances supported for VMSG >> clusters, if I were in Florida I would invest in one very strong and ' >> well protected (UPS, etc.) building.  > J > True. Then again, even given such a massive storm system, it is possibleC > that a remote cluster site outside of Frances's reach may survive  > relatively uneffected.  / What is the current limit on Cluster distance ?   B Certainly there were multiple places in Florida hit, and one would not want to have all sites hit.   E I have heard of a (non-VMS) company in the midwest that aligned their B data centers on an axis such that no single tornado would take outA both of them because tornados in the area always move at the same 1 angle.  I don't believe hurricanes work that way.    ------------------------------   End of INFO-VAX 2004.496 ************************