1 INFO-VAX	Mon, 12 May 2003	Volume 2003 : Issue 261       Contents:) Re: AlphaStation 255/500 power up problem  Re: Another AS 500/266 question  Another AS 500/266 question ' Re: creating licenses for my own demos? A Re: DECnet-Plus (DECnet PhaseV) (DECnet/OSI) migration necessary?   Re: Determining Disk Booted From Free VMS hardware < Re: How Alpha will save Itanium - must reading for Bill Todd" Re: INDEX.SYS size and performance' Re: OpenVMS Memory/Performance Question 1 Re: Pinout of power connector on DECserver 90TL ? 1 Re: Pinout of power connector on DECserver 90TL ? 1 Re: Pinout of power connector on DECserver 90TL ? < Re: Problem to localy logon on a Pathworks 6.1 domain member simh emulator and cluster  Re: simh emulator and cluster  RE: Spamfilter for OpenVMS?  RE: Spamfilter for OpenVMS?  TCPIP SET CONF SMTP /SUB=HIDDEN   F ----------------------------------------------------------------------  % Date: Sun, 11 May 2003 17:27:10 -0400 0 From: "Alan Boyles" <alan.boyles@mindspring.com>2 Subject: Re: AlphaStation 255/500 power up problem/ Message-ID: <vbtg086mph3qde@corp.supernews.com>   I If you plug a power supply in and turn it on can you test current that is I coming out with  a volt meter ?  If so, what should the output look like?    Alan  ; "Alan Boyles" <alan.boyles@mindspring.com> wrote in message ) news:vbpk74sin2cg2e@corp.supernews.com...  > Group: > G > I have just received 2 AlphaStations (a 255/400 and a 255/500) and am  havingJ > some problems.  When I try to turn either of the boxes on they will veryI > briefly run, maybe half a second to a second and then shut down.  I was I > thinking power supply but I find it strange that both boxes do the same  > thing.  Any ideas ?  >  > Thanks >  > Alan >  >    ------------------------------  + Date: Mon, 12 May 2003 00:37:50 +0000 (UTC) - From: lewis@BOXTER.mitre.org (Keith A. Lewis) ( Subject: Re: Another AS 500/266 question. Message-ID: <b9mqcu$9qs$1@newslocal.mitre.org>   Paul Repacholi <prep@prep.synonet.com> writes in article <87isshduvw.fsf@prep.synonet.com> dated Mon, 12 May 2003 05:40:51 +0800: @ >Anyone any idea on how to check the built in ethernet on these?; >Mine seems to only work for a few % of packets :( I've got 9 >a DE450 or 500 in at the moment, but I'd like to get the ( >onboard unit sorted if at all possible.  I High packet loss rate is often caused by a mismatch between the speed and L duplex settings of the card and the settings of the hub it's plugged into.     >>>SHOW EWA0_MODE    "twisted" means 10half- "BNC" means 10half through the coax connector F "auto-select" means use twisted if there's a connection, otherwise BNC   (For your hub, RTFM.)   + --Keith Lewis              klewis$mitre.org > The above may not (yet) represent the opinions of my employer.   ------------------------------  % Date: Mon, 12 May 2003 05:40:51 +0800 , From: Paul Repacholi <prep@prep.synonet.com>$ Subject: Another AS 500/266 question- Message-ID: <87isshduvw.fsf@prep.synonet.com>   ? Anyone any idea on how to check the built in ethernet on these? : Mine seems to only work for a few % of packets :( I've got8 a DE450 or 500 in at the moment, but I'd like to get the' onboard unit sorted if at all possible.    tnx.   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Sun, 11 May 2003 17:53:04 -0400 " From: Beyonder <beyonder@vrx.nbet>0 Subject: Re: creating licenses for my own demos?8 Message-ID: <udgtbvg622al6s2f3ncsmsbljti7th3s6b@4ax.com>  D On Fri, 9 May 2003 19:27:15 -0400, John Santos <JOHN@egh.com> wrote:7 >HMMMM!  A lot of erroneous assumptions in this thread.  > ) >1) Sign up for CAS^H^H^HASAP^H^H^H^HDSPP    done!    >2) Request the PAKGEN PAK  C did that. got some nonsense email asking me for my HP pavilion BIOS / version and serial number (which I said "HUH?")   * >3) Register the PAKGEN PAK on your system1 >4) Follow the instructions for generating PAK's.  >  >It now magically works.  > thanks for the other info, I'll test license register/generate just for my own amusement ;)   B.   ------------------------------  % Date: Sun, 11 May 2003 23:05:31 -0400 * From: JF Mezei <jfmezei.spamnot@istop.com>J Subject: Re: DECnet-Plus (DECnet PhaseV) (DECnet/OSI) migration necessary?) Message-ID: <3EBF0F75.67C0E720@istop.com>    > Bob Ceculski wrote: 
 > > [snip]F > > Javascript is used by our web/graphics design people ... what else > > do you suggest they use?  + HTML first. Javascript only when necessary.   M Remember that many have disabled javascript by default because of abuses made E by some web sites, and javascript generally slows things down anywasy I (especially when you one page that could fit inside 2k of text requires 5 ' separate javascript URLs to be fetched.   J the official way for redirects is through the web server, or via the <metaL tag. You can chieve your desired results in such a way that even though with1 very limited browsers can still access your site.   M Also, bear in mind that there are many blind people who still access the web  K via speed synthesis and those systems prefer the least amount of images and  javascript stuff.    ------------------------------  # Date: Mon, 12 May 2003 00:42:38 GMT ! From: rob.buxton@wcc.spam.govt.nz ) Subject: Re: Determining Disk Booted From $ Message-ID: <3ebeebd8.16619117@news>  1 On Fri, 9 May 2003 23:28:04 +0100, "John Travell"  <john@travell.uk.net> wrote:D I think this will just show you which is considered to be the Shadow? Set Master, but not necessarily the device that was booted off.   D If you've split and reformed the shadow set since the last boot then? the "Master" will simply be the device that was not dismounted.   ? Some of the f$getdvi lexicals can be used to look at the Device C Members, but again, these just relate to the shadow sets and do not  give you what you were after.      > < >"Jack Trachtman" <Jack.Trachtman@vmmc.org> wrote in message7 >news:69d784c4.0305091349.f99ded2@posting.google.com... 
 >> [VMS V7.3]  >>8 >> All our VMS systems have their system disks shadowed. >>9 >> How can I find out which disk of the pair was actually  >> used to boot from?  >>
 >$ ana/sys >SDA> show dev DSAx  > = >One of the devices in the member list will mention 'master'. J >I do not have a shadow set to hand to refresh my memory of exactly where.L >Undoubtedly there are other ways to identify this, but I do not know of any >way to do it from DCL.  >  >-- 
 >John Travell ! >VMS crashdump expertise for hire  >john@travell.uk.net >http://www.travell.uk.net/  >  >  >  >---' >Outgoing mail is certified Virus Free. ; >Checked by AVG anti-virus system (http://www.grisoft.com). B >Version: 6.0.478 / Virus Database: 275 - Release Date: 06/05/2003 >  >    ------------------------------  % Date: Sun, 11 May 2003 19:39:20 -0400 % From: "John Vottero" <John@mvpsi.com>  Subject: Free VMS hardware/ Message-ID: <vbtnpakou2224a@news.supernews.com>    Free to a good home:  F 1 TK50Z-GA TK50 tape drive - It worked the last time it was plugged in (which was a long time ago).E 1 RRD40-DA CDROM drive - Also worked the last time it was plugged in.    Located in Columbus, Ohio - USA   A I don't plan to put a lot of effort into shipping these anywhere.    John Vottero   ------------------------------    Date: 11 May 2003 22:49:25 -0500+ From: young_r@encompasserve.org (Rob Young) E Subject: Re: How Alpha will save Itanium - must reading for Bill Todd 3 Message-ID: <sPGC9yy5yUqe@eisner.encompasserve.org>   _ In article <I4WdnRo-U46VYyCjXTWcqA@metrocast.net>, "Bill Todd" <billtodd@metrocast.net> writes:  > : > "Rob Young" <young_r@encompasserve.org> wrote in message/ > news:Jj+dR7teY0eX@eisner.encompasserve.org... A >> In article <ovqcndxS1b2GNCGjXTWcog@metrocast.net>, "Bill Todd" " > <billtodd@metrocast.net> writes: >> >>= >> > "Rob Young" <young_r@encompasserve.org> wrote in message  >> >> > >> >> < >> >> Additionally, HP has the highest 4-CPU result on tpmC: >> >G >> > Quite likely only because IBM doesn't bother submitting scores for 	 > systems M >> > that small.  If their 32-processor system scaled perfectly linearly then  > a I >> > 4-processor version would yield a score of over 85,000 tpmC, so take  > your >> > choice: >> >>F >> http://www.tpc.org/tpcc/results/tpcc_result_detail.asp?id=103042401 >>( >> 85000 tpmC is a lot less than 121000. > K > But it's more than twice the 41,142 that you'd get if you scaled down the M > SuperDome configuration linearly to 4 processors, which, of course, was the = > point (not that I'm surprised you failed to understand it).  >   ; 	You are scaling in the wrong direction.  The 4-CPU Madison ? 	config is with a chipset that doesn't go higher in CPU counts. 9 	SuperDome as we know it is old news.  What Hein hints at @ 	is next-gen boxes.  HP has had Compaq for a year now, certainly: 	the HP+Compaq folks have been burning the midnight oil on> 	a next-gen box and Hein hints at a next-gen box.  My point of< 	course is to show the CPU itself is certainly more powerful> 	than anything currently shipping.  That is if tpmC and SAP SD 	is to be believed.      >   So even if IBM; >> were to submit a 4 CPU result (not sure - the IBM config B >> describes 8-way MCM.  Not sure if the p690 goes down to 4-way). > F > Nope - 8-way is the smallest configuration IBM bothers selling:  the > comparison was hypothetical. >   	 	Roger.      >> >>A >> >  either close to linear scaling to large processor counts is L >> > possible in this benchmark, which means that SuperDome's scalability is > justL >> > as lousy as a cursory glance makes it appear to be, or the new POWER4+s > are B >> > likely faster in a 4-processor configuration than Madison is. >> > >> >> Not at all. > N > Yes, Rob - clueless though you obviously are, the above are the only logicalK > conclusions one can draw from the existing data:  either the new POWER4+s K > would be faster in a 4-processor configuration than the new Madisons (but B > scale somewhat less than linearly to the score they reach in theF > 32-processor configuration, or, if they were slower in a 4-processorN > configuration than the new Madisons, they would be scaling close to linearlyJ > up to 32 processors while SuperDome's scaling was so shitty that despiteL > HP's performance lead at 4 processors it fell behind by more than a factorM > of 2 per processor at 64 processors.  Take a few days if you need to figure 	 > it out.  >   6 	If you take a hypothetical 85000 for a Power4+ 4-way,@ 	that is 21000 per CPU.  21250 * 32 = 680000.  That is the exactA 	number.  Which is impossible.  No architecture scales linerally. = 	So your scaling *down* to 85000 for a 4-way is off the mark. E 	I'll give you a few days to come up with a better number than 85000  = 	for a hypothetical IBM 4-way.  It would do better than that.   < 	Conversely, HP is showing a 121000 number for 4 Madisons.  > 	That is 30250 per CPU.  Looking at various architectures fromE 	small CPU counts to larger (obviously across chipsets/architectures) < 	I am seeing 75%-80% scaling 4 to 16 CPUs.  Maybe they can't5 	hit 75%.  If they hit 72%, they do better than 680K.   @ 	Current SuperDomes are obviously constrained.  As Andrew points< 	out, their bandwidth cross-section isn't exactly burning up1 	the charts - but SuperDome is long in the tooth.    > ...  > I >> >> http://www.tpc.org/tpcc/results/tpcc_result_detail.asp?id=103042401  >> >> ' >> >> 121000 tpmC.   At $4.97 per tpmC.  >> >> C >> >> So all you have left is "$ per metric" and of course cheering = >> >> on IBM and Opteron - never acknowledging your circle is  >> >> getting tighter. >> >> @ >> >> Oh... at $4.97 per tpmC for that 4 processor Madison, your< >> >> "$ per metric" argument is looking a bit weak at that. >> >A >> > Not really, Rob:  if you're looking at small systems and are  > cost-conscious, ? >> > then the 4-processor Opteron at $2.76 per tpmC is for you.  >> > >>> >> Well that's a nice whitebox number with decent performance. >>@ >> So you do still have "$ per metric" argument - Opteron versus >> Madison.  > L > Price/performance is certainly *one* legitimate knock on Itanic, Rob:  theJ > problem is that you keep bleating that it's the *only* legitimate knock, > which is simply false. >  >   Reviewing: >>. >> Madison is considerably more powerful than: >># >> Opteron in TPC-C, SAP SD, SpecFp  > I > Probably true for SPECfp.  Haven't seen any Opteron numbers for SAP SD: L > have you (if so, please provide a reference), or are you just spewing more > vapor? >   > 	No Opteron results.  But sifting back and forth at sap.com weB 	see a recent HP 4-way 2.0 GHz Xeon does 145000 dialog steps/hour @ 	supporting 480 users with 1.93 seconds response, 2420 "SAPS" - H 	whatever all that means.  A fully decked out 4-way 2.0 GHz Xeon box by H 	Dell shows it can crank 78000 tpmC - (this same CPU config by IBM does D 	53000 tpmC).  That Rackserver box does 80000+ tpmC with an Opteron H 	4-way.  Let's give it a benefit of a doubt and suppose it can "somehow": 	do 20% better than a Xeon comparing SAP SD benchmarks (itC 	probably wouldn't do that much better - but I need to build a case E 	that "Madison blows away Opteron at SAP SD").  That would place the   	Opteron at:     			 580  users 			 good response  			2900 SAPS 			175000 dialog steps/hour     $ 	4-way Madison SAP SD numbers by HP:  
 			860  users  			good response 			4320 SAPS 			259000 dialog steps/hour     N > TPC-C, however, depends heavily on system configuration and tweaking, as IBMC > has just demonstrated by improving its 32-processor configuration G > performance by 59% with less than a 31% rise in clock rate.  So while F > Madison's current TPC-C score certainly is substantially higher thanI > Opteron's, to determine whether Madison is *in fact* 'considerably more J > powerful' than Opteron in a 4-processor configuration we'd have to applyM > similar amounts of expertise to the configuration and equivalent amounts of I > hardware (for example, the Opteron configuration used only 294 disks to > > Madison's 468, and only 32 GB of memory to Madison's 64 GB). >   ? 	The *presumption* being it was IO bound or memory bound.  More < 	than likely CPU bound as they would toss disks at it to getB 	a higher number.  Memory is cheap - they would toss memory at it,> 	perhaps it is maxed out however.  The very reason the Madison< 	config has so much more.  You are touching on a feature but9 	are neglecting to flesh it out.  They didn't stick those @ 	drives in the Madison config because they had room.  They stuck@ 	them in there because the Madison could drive them.  After all,> 	extra disks drive up system cost driving down your $/tpmC and= 	doing *nothing* but spinning.  By focusing on the disk count 9 	you are actually seeing how much more powerful a Madison < 	processor is compared to Opteron - for this benchmark.  But6 	I expect to see similar symptons in other benchmarks.  N > Opteron is just getting started, Rob.  Itanic2 has been out in the field for
 > a year,   C 	Madison isn't even shipping yet.  Power4+ isn't even shipping yet. A 	In fact, that IBM number hit the streets the day it could.  i.e. D 	you can order that Power4+ config 6 months after tpmC publish date.  D > and its TPC-C figures (on the same hardware) have risen noticeablyG > over that period.  I expect the same of Opteron's as its vendors gain  > expertise in benchmarketing.? >> Power4 same but add SpecInt - for low CPU count comparisons.  > N > POWER4+ (not POWER4) is the competition now, Rob.  And it blew the doors offJ > Madison in TPC-C (more than twice the performance per processor - and atM > only about half the power consumption per processor as well) by so large an N > amount that it is very probable that it would exceed Madison's per-processorK > performance at the smallest POWER4+ system size IBM ships (8 processors).  >   > 	Probably not.  But we will never know.  IBM 8-way results forB 	a p690 will be rarer than hen's teeth.  After all, that 8-way MCM@ 	has a sticker price of $350000.  Even if higher than a mythical@ 	8-way Madison (doubt it) the $/tpmC would give Madison way too @ 	much marketing muscle (that 4-way shows each Madison listing atC 	$9900 an 8-way Madison config shouldn't be that much of a jump in   	per-CPU prices).    >>< >> Who is to say a Madison box isn't lurking in the wings to! >> make 32-64 CPU results higher?  > M > Well, 'higher' won't quite cut it, at least for SuperDomes (NEC boxes don't K > seem to scale quite as poorly):  the SuperDome results will have to *more B > than double* their per-processor performance to match POWER4+'s. > ! >   Hein basically hints at that:  >> >>N > http://groups.google.com/groups?selm=3E63CCD2.9A6E9124%40eps.zko.dec.com&oe= > UTF-8&output=gplain  >>L >> "What the Itanium and it's follow ups needs is a good large SMP SYSTEM toJ >> be put in and flourish and a good OS and DB to get everything out of itJ >> that is potentially in there. Obviously HP is working on such solutions. >> but I can not say when/where/how-many-CPUs. > L > Pity that HP already owns a world-beating large MP system architecture but > has declared it superfluous. >   9 	Yeah, beat a dead drum.  But as others have pointed out, > 	NEC, SGI, Fujitsu, IBM, Dell and others are shipping Itanium  	servers also.   >>B >> The NEC benchmark is just a stake in the ground. A first step." > G > By that reasoning, the new SuperDome result (with per-processor TPC-C < > performance only 64% of NEC's) is a major step *backward*. >   @ 	Hein wrote that after the SuperDome numbers were published, but: 	you knew that.  An *obviously* if the NEC folks can scaleA 	Madison, anyone else can and as Andrew points out, the SuperDome @ 	doesn't have world-class bandwidth.  The next-gen box certainly 	should.   >> >>: >> The inference of course is there is higher numbers than* >> what NEC demonstrated coming out of HP. > J > While it's not surprising that that is *your* inference, Rob, the actualH > *implication* of Hein's words is that HP is indeed aware of how poorlyG > SuperDome scales (it would, after all, be difficult not to be) and is N > *trying* to do something about it.  However, since they've had SGI's systemsM > to copy for the better part of the past decade and failed to do so with any L > apparent success, why anyone would expect them to have any better luck nowM > is not quite clear:  my guess is that they're waiting for the Alpha team to M > bail them out (again) 3 or 4 years from now with a server-on-a-chip product N > that eliminates the problem, but feel free to divulge any interesting serverK > development in the pipe (that you can substantiate:  your own glowing but N > vague expectations aren't particularly interesting to anyone but you) that I > might not be aware of. >   4 	Bill, all server vendors ship high-end servers thatB 	outperform the prior generation.  Until Sun came up with the 15K,C 	they were stuck with spins of the UE10000 which had its own issues > 	(other than Zinc Whiskers).  Obviously, they have a much more 	powerful box.  B 	SuperDome came out Sept 2000, a lifetime in this industry.  There? 	has to be a follow-on tuned specifically with Madison in mind, < 	or one would think.  Either way, their next-gen box will be 	an improvement on SuperDome.    >   Wonder how soon . >> the IBM folks will have to eat these words: > K > My guess would be never.  Don't forget that next year the already-leading M > POWER boxes get POWER5 processors with SMT and on-chip offload hardware for K > another major performance boost, while Itanic2s get - more on-chip cache.  >   : 	But that will be a while.  After all the Power4+ hardware= 	ships Nov 2003.  Certainly a new HP high-end server shows up = 	before then (with similar or later ship dates - but no doubt  	well before Power5).    >>+ >> http://www.theinquirer.net/?article=9416  >>L >> It took a further dig at HP and Intel. Adalio Sanchez, general manager of > the I >> Eserver division, said: "We don't just assemble boxes with third party  >> components".  >>+ >> I'm sure that fired up the troops at HP.  > L > I don't think that using the word 'fired' in conjunction with HP employees > is very tactful, Rob.  >  >   Surely they < >> are pushing hard to hit 681000 tpmC with 32 Madisons ;-).= >> Looking at some examples, I'm seeing 75-80 percent scaling  >> factor across architectures.  > M > Why not provide the numbers that lead you to this conclusion, Rob?  Then we M > can figure out whether you understood the data you're drawing it from.  For N > example, some NUMA architectures (such as EV7, POWER4/4+, SPARC in its largeG > boxes, SGI's boxes, and, within its glueless MP range, Opteron) scale H > considerably better than that, while other server architectures (e.g.,2 > bus-structured ones) tend to have more problems. > G > The only examples we have of Madison's scaling are the 4-processor HP K > result, the 32-processor NEC result that yields only a 4.25x higher score N > using 8x as many processors, and the pitiful 64-processor HP result that yieM > lds only a 5.44x higher score using 16x as many processors.  That works out N > to a performance improvement of 62% for each doubling of processor count forJ > the NEC case and only a 53% performance improvement for each doubling of) > processor count for the SuperDome case.  >   B 	Exactly.  And one would suppose a next-gen box would scale better> 	than NEC's current shipping hardware.  You are presuming that@ 	you need on-chip switches to scale.  Not so.  IBM doesn't have  	them.  ) >   Ideally, an HP 32 next-gen box should 
 >> be able to A >> get 720K to 760K tpmC *if* their 4-way result is any indicator  > E > But of course it has already proven *not* to be any indicator, Rob: K > expecting anything that close to linear scaling out of HP servers is pure M > wishful thinking until there's at least *some* kind of evidence to point to < > that they've found a way to get their server act together. >   ? 	That isn't linear.  32 * 30250 would be linear or 900000 tpmC.  	720K is assuming 75% scaling.  ; >> (the CPU is certainly capable of it - let's see how they < >> did on their next gen box.  Bet it has higher bandwidth). > N > Any time you have anything more substantial than your own hopes to offer up,$ > Rob, I'll be happy to evaluate it. >   D 	Name a high-end server vendor that shipped a next-gen high-end box C 	that didn't outperform the previous generation.  It will certainly  	have higher bandwidth.    >>@ >> Madison is dominating the 4-CPU space for performance.  Sure,E >> Opteron is in a nice niche there.  Value 4-CPU space.  But Opteron @ >> will get its competition from 3.0 GHz Xeons coming as Madison@ >> is the high performer in that space (running real benchmarks,E >> i.e. SAP, TPC - non-clustered, SpecFp - neck and neck with Opteron  >> on SpecInt).  >>> >> Interesting to see IBM being squeezed terribly in the 4-CPU- >> space by Xeon and now Opteron and Madison.  > K > Er, *you* were the one who pointed out above that IBM doesn't even bother J > selling a POWER system that small, so how they could be getting squeezedL > there is an interesting question.  But Itanic will indeed be squeezed:  byN > POWER on the high side and Opteron (and Xeon, where 32 bits are adequate) onN > the low side.  And IBM appears happy to give Opteron the support it needs toL > stake out a major presence there - offering similar performance to Itanics# > at a small fraction of the price.  >   < 	We will see what IBM does with the Opteron.  Up until p690,# 	IBM shipped 4-CPU pseries servers.    				Rob    ------------------------------    Date: 11 May 2003 20:04:45 -0700. From: spamsink2001@yahoo.com (Alan E. Feldman)+ Subject: Re: INDEX.SYS size and performance = Message-ID: <b096a4ee.0305111904.1f9783aa@posting.google.com>   ` "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message news:<3EBC527E.3083807B@fsi.net>... > "Alan E. Feldman" wrote: > > e > > brandon@dalsemi.com (John Brandon) wrote in message news:<03050908582736@dscis6-0.dalsemi.com>...  [...] M > > > > Now, while it may still be a good idea to estimate the largest number L > > > > of file extents you ever expect to have on the disk and set /HEADERSG > > > > to that number, there is much less need to worry about actually M > > > > running out of space in the file header for INDEXF.SYS because of the 0 > > > > new extension algorithm mentioned above. > > > P > > > True.  However taking a little time to determine the needs for the disk isR > > > always a good idea.  And the initial creation of the disk is a lot less time > > G > > Agreed, as I already said above. I guess I should have said that at F > > least it's not like the VMS .LE. v5 days in which this problem wasF > > much more likely to happen. I found that with the old algorithm ofI > > extending INDEXF.SYS in chunks of 1000 blocks that you'd be likely to J > > get the HEADERFULL problem when you reached about 50000 blocks on yourB                                                       ^^^^^^^^^^^^  7 Uh, make that approx. 50000 *files*, not blocks. Sorry.   	 > > disk.  > F > ...depending upon the values of /MAXFILES and /HEADERS at INITIALIZEG > time, and depending on how fragmented INDEXF.SYS has become (i.e., is * > the extent map in the header full yet?).  E I should have mentioned that for the 50000-file header-fillup rule of B thumb I was talking about disks that were INIT-ed with the defaultF value for /HEADERS, which IIRC is, or at least back then was, 16 (withC the default value for /MAXIMUM_FILES). I would think that since the D new algorithm extends INDEXF.SYS in bigger, and hence fewer, chunks,< the file header for INDEXF.SYS would not fill up as quickly.   Disclaimer: JMHO Alan E. Feldman    ------------------------------  # Date: Mon, 12 May 2003 01:00:43 GMT ! From: rob.buxton@wcc.spam.govt.nz 0 Subject: Re: OpenVMS Memory/Performance Question$ Message-ID: <3ebeef5c.17519812@news>  B On Fri, 9 May 2003 17:11:51 GMT, Rick Dyson <rick-dyson@uiowa.edu> wrote:  C Things have changed a bit with the various IO Caches so I'm not too + sure what are "recommended" limits anymore.   E Rather than the free memory, look towards the Page Read IO stats from  the Monitor IO Page.D Alas, it's a long while ago now and I've forgotten the general rulesE of thumb, but this value indicates Reads from the Pagefiles and is an  indication of Hard Faulting.F If you're excessively hard-faulting then you need to look further intoE Memory Tuning, and that would include either reviewing UAF Records to , the Sysgen Parameters to buying more memory.  ? As others have said, moving Pagefiles off the System disk is an  option. F You've also got two very differently sized Pagefiles, in the past this@ could cause problems if the smaller one filled up to beyond 90%.# I'd suggest evenly sized Pagefiles.     N >This is probably a "dumb guy" question, but I need to find the answer quickly$ >and haven't found it on my own yet. > L >On my system, a MONITOR CLUSTER shows that my "%Memory In Use" is running a
 >steady 80%+.  > I >My question is whether this is a report of hard RAM usage or the entire  H >virtual memory?  I do have a known problem with my primary pagefile.  II >can't extend it due to a "header is full" problem.  I am preparing for a I >maintanence cycle to defragment the system disk to address that.  If the H >high memory use is a side effect of this, then maybe it will get better >after I increase the pagefile.  > J >I have a GS60 with 4 GB of RAM and 4 5/625 CPUs.  OpenVMS v7.3.  Here are >some details: >  >$ Show MemoryA >              System Memory Resources on  9-MAY-2003 12:09:42.15  > M >Physical Memory Usage (pages):     Total        Free      In Use    Modified M >  Main Memory (4.00Gb)            524288       84150      392441       47697  >  >... > M >Swap File Usage (8KB pages):                   Index        Free        Size Q >  DISK$MISC_73:[SYS0.SYSEXE]SWAPFILE.SYS                                         M >                                                   1        4312        4312  > M >Paging File Usage (8KB pages):                 Index        Free        Size Q >  DISK$MISC_PAGE:[SYSEXE]PAGEFILE2.SYS;1                                         M >                                                 253      499990      499992 Q >  DISK$MISC_73:[SYS0.SYSEXE]PAGEFILE.SYS                                         M >                                                 254       23582       23584  > M >  Total size of all paging files:                                     523576 M >  Total committed paging file usage:                                  275964  > P >Of the physical pages in use, 15648 pages are permanently allocated to OpenVMS.   ------------------------------  % Date: Sun, 11 May 2003 14:33:08 -0400 3 From: "Homer J. Simpson" <hsimpson@burnsenergy.com> : Subject: Re: Pinout of power connector on DECserver 90TL ?9 Message-ID: <Kywva.62043$Wq.2037@fe03.atl2.webusenet.com>   3 Do you mean it will work with a H7625-AA like this? = http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3024001060     9 "Paul Repacholi" <prep@prep.synonet.com> wrote in message ' news:87el35dcxn.fsf@prep.synonet.com... 2 > Dale Dellutri <ddelQQQlutr@panQQQix.com> writes: > H > > On Sat, 03 May 2003 16:39:39 +0200, Jan-Erik S?derholm <aaa@aaa.com> wrote: > G > >> I'v got a DECserver 90TL (DSRVE-M), but no power supply.  The unit G > >> needs +5v and +12v, but I don't have the pinout of the round power 2 > >> connector on the back .  Anyone having this ? > E > > I have a (no longer used) DEcserver 90TL, and the power supply is ? > > marked H7082-AA.  It also says the output is DC 5.1V, 1.8A.  > H > Mine is a H7625-AA, 5V/2.5A and 12V/1.4A. I'd expect you would have to5 > have 12V for the line drivers or a Terminal Server.  >  > --  > > Paul Repacholi                               1 Crescent Rd.,9 > +61 (08) 9257-1001                           Kalamunda. B >                                              West Australia 6076, > comp.os.vms,- The Older, Grumpier Slashdot0 > Raw, Cooked or Well-done, it's all half baked.H > EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Mon, 12 May 2003 05:25:21 +0800 , From: Paul Repacholi <prep@prep.synonet.com>: Subject: Re: Pinout of power connector on DECserver 90TL ?- Message-ID: <87n0htdvlq.fsf@prep.synonet.com>   5 "Homer J. Simpson" <hsimpson@burnsenergy.com> writes:   5 > Do you mean it will work with a H7625-AA like this? ? > http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3024001060  >  > ; > "Paul Repacholi" <prep@prep.synonet.com> wrote in message ) > news:87el35dcxn.fsf@prep.synonet.com... 3 >> Dale Dellutri <ddelQQQlutr@panQQQix.com> writes:  >>I >> > On Sat, 03 May 2003 16:39:39 +0200, Jan-Erik S?derholm <aaa@aaa.com>  > wrote: >>H >> >> I'v got a DECserver 90TL (DSRVE-M), but no power supply.  The unitH >> >> needs +5v and +12v, but I don't have the pinout of the round power3 >> >> connector on the back .  Anyone having this ?  >>F >> > I have a (no longer used) DEcserver 90TL, and the power supply is@ >> > marked H7082-AA.  It also says the output is DC 5.1V, 1.8A.  F >> Mine is a H7625-AA, 5V/2.5A and 12V/1.4A. I'd expect you would have9 >> to have 12V for the line drivers or a Terminal Server.   D That is wrong, I've just pulled some units and checked, they are allF marked as 5.1V only. Other proble is the volts from my PS seem to haveH run off and hid :( Have to get a DIN7 socket and check the output. I use5 all mine in hubs, so I've not used the standalone PS.    --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Sun, 11 May 2003 22:00:24 -0400 ( From: David Froble <davef@tsoft-inc.com>: Subject: Re: Pinout of power connector on DECserver 90TL ?, Message-ID: <3EBF0038.6010903@tsoft-inc.com>  O Yes, but I've seen them sell for as low as $5.  Since shipping will be a major i7 expense, picking up one locally would be the best idea.t  H Note that a DEChub 90 will also provide a nice home for the 90TL, and 7  additional 90 series modules.V   Dave     Homer J. Simpson wrote:   5 > Do you mean it will work with a H7625-AA like this? ? > http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3024001060E >  > ; > "Paul Repacholi" <prep@prep.synonet.com> wrote in message ) > news:87el35dcxn.fsf@prep.synonet.com...9 > 2 >>Dale Dellutri <ddelQQQlutr@panQQQix.com> writes: >> >>G >>>On Sat, 03 May 2003 16:39:39 +0200, Jan-Erik S?derholm <aaa@aaa.com>o >>>a > wrote: > F >>>>I'v got a DECserver 90TL (DSRVE-M), but no power supply.  The unitF >>>>needs +5v and +12v, but I don't have the pinout of the round power1 >>>>connector on the back .  Anyone having this ?- >>>>D >>>I have a (no longer used) DEcserver 90TL, and the power supply is> >>>marked H7082-AA.  It also says the output is DC 5.1V, 1.8A. >>>cH >>Mine is a H7625-AA, 5V/2.5A and 12V/1.4A. I'd expect you would have to5 >>have 12V for the line drivers or a Terminal Server.n >> >>-- n> >>Paul Repacholi                               1 Crescent Rd.,9 >>+61 (08) 9257-1001                           Kalamunda. B >>                                             West Australia 6076, >>comp.os.vms,- The Older, Grumpier Slashdot0 >>Raw, Cooked or Well-done, it's all half baked.H >>EPIC, The Architecture of the future, always has been, always will be. >> >  >  >      -- l4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com6 T-Soft, Inc.  170 Grimplin Road  Vanderbilt, PA  15486   ------------------------------  # Date: Mon, 12 May 2003 02:28:13 GMTr! From: rob.buxton@wcc.spam.govt.nz.E Subject: Re: Problem to localy logon on a Pathworks 6.1 domain member$$ Message-ID: <3ebefb4a.20573102@news>  D On Fri, 9 May 2003 07:30:46 +0200, "Seghers Bruno" <tips@euronet.be> wrote:  D Okay, I don't know if this will work as I always logon to the domain$ to do any kind of setting up shares.   From the admin prompt do:a  A set admin /domain=\\server    where server is the name of the VMS  System you've got Pathworks on.i* You should then end up with a prompt like:   \\Server\Server> u  , The first Server may be the Pathworks Alias.  D From there yo should be able to do a SHOW USER and get a list of the Local Accounts.D  F I've no idea what our Local Administrator account is, so I cannot testA this, but a Logon Administrator should now try and login locally.i  D I'm not sure if you'll be able to create the shares with Permissions at this point. v   Hope this helps,   Rob.  L >The NT administrator add my pathworks in the domain, thus I have no idea of >the domain Admin password.  >lJ >I just want to be local Administrator (Admin account located in the local" >SAM file) of the pathworks server >cJ >The local administrator password is a question asked at the first time of9 >the config when you change the domain name and the role.  >5 >Thanks for your helpn >b >Seghers Bruno >Belgium/ ><rob.buxton@wcc.spam.govt.nz> wrote in message   >news:3ebb1fa0.199397238@news...G >> On Thu, 8 May 2003 20:59:21 +0200, "Seghers Bruno" <tips@euronet.be> 	 >> wrote:n >>E >> When you install Pathworks it needs an Account and Password on thepH >> domain, sufficient to be able to add itself in. I believe the account@ >> used needs Domain Admin privs. It's not a local administrator >> password. >>: >> So, when you then logon you're logging onto the Domain. >>* >> Note also passwords are case sensitive. >>H >> Finally, Pathworks can lose the plot with the NT Domain and Pathworks >> needs to be restarted.a >>< >> But, your account and password needs to be in the domain. >> >> Rob.  >> >Hi,A >> >= >> >I have installed pathworks 6.1 on a openVMS 7.2-2 system.AL >> >I configure it as a member of a resources Windowt NT domain A trusted by >a >> >Windows NT domain BgB >> >During installation, Pathworks ask me a password for the local >Administrator >> >account. >> >K >> >The server pathworks is up and running, I see from domain B the default 7 >> >shares created by pathworks (USERS PWUTIL, etc ...)l >> >I >> >My problem is that when I try to manage my Patworks server (add users-I >> >localy, add share with security permissions, etc ...) with ADMINISTERz8 >> >commands, the server says : "your are not Logged on" >> >I try to do :n >> > >> >$ADMINISTERs* >> >domainA\\Myserver> LOGON Administrator8 >> >password : 'the one I introduce during installation' >> >6 >> >You are logged on but not authantified by a server= >> >You will not have permissions..... or something like thatc >> >7 >> >Of course, I'm sure of the validity of the passwordu >> >@ >> >Does anybody can explain me how can I locally logon with the >administratorJ >> >account on the pathworks server. I am not admin of the domain, but I'mJ >> >responsible for giving access to dedicated user on dedicated shares on >the >> >pathworks server.  >> > >> >Thanks a lot for your help >> > >> >Regardsd >> > >> >Seghers Brunoa >> >Belgium  >> > >> > >> >b >s   ------------------------------  # Date: Sun, 11 May 2003 19:56:43 GMTy% From: "bayden cline" <bayden@isys.ca> " Subject: simh emulator and clusterJ Message-ID: <%Vxva.179219$kYH.136688@news01.bloor.is.net.cable.rogers.com>  L I am wondering if i can add a vax running in a simh emulator on my pc to theL vax cluster that we have working in the office and have the vax cluster abelJ to use it the emulators disk space.  all machines involved are running VMS$ 6.2.  Thanks in advance for any help   bayden   ------------------------------  % Date: Sun, 11 May 2003 22:12:15 -0500e1 From: "David J. Dachtera" <djesys.nospam@fsi.net>v& Subject: Re: simh emulator and cluster' Message-ID: <3EBF110F.D1813D84@fsi.net>i   bayden cline wrote:  > N > I am wondering if i can add a vax running in a simh emulator on my pc to theN > vax cluster that we have working in the office and have the vax cluster abelL > to use it the emulators disk space.  all machines involved are running VMS > 6.2.  H I should think so, but understand that you're doing disk I/O at EthernetG speeds. With a 10Mbit interconnect, that could have some fairly serious>" impacts on the production cluster.  H Remember also that only the disk devices visible to VMS can served using MSCP over the NI interconnect.  F If possible, I'd run this using a test cluster first so you can assess@ the impact. If not possible, I'd try to run this test off hours,B preferably using a scheduled downtime window and after making very? certain that all the backups have run to successful completion.r  H Please approach this with great caution until you have made very certain3 that there is no danger to your production systems.C   -- b David J. Dachteras dba DJE Systemsr http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/-   ------------------------------  % Date: Sun, 11 May 2003 11:06:14 -0700 # From: "Tom Linden" <tom@kednos.com>F$ Subject: RE: Spamfilter for OpenVMS?9 Message-ID: <CIEJLCMNHNNDLLOOGNJIIELOHCAA.tom@kednos.com>K  H Speaking of which, are there any Teergruben available for VMS, TCPIP5.1?   >-----Original Message-----A5 >From: Larry Kilgallen [mailto:Kilgallen@SpamCop.net] # >Sent: Sunday, May 11, 2003 9:40 AMS >To: Info-VAX@Mvb.Saic.Com% >Subject: Re: Spamfilter for OpenVMS?e >  > C >In article <CRmva.54822$0W.25481@fe09.atl2.webusenet.com>, Hunter 2) >Goatley <goathunter@goatley.com> writes:r >> Bob Ceculski wrote: >>> 1 >>> I thought pmdf already had spam filtering ...T >>  @ >> PMDF has support for Sieve filters and mapping rules that canE >> be used to filter out some spam, but nothing like what the productt >> we're developing will do. > H >Will what you are developing allow a Reject in cases where the usernameK >does not exist ?  I understand the RFCs allow for Bounces to be generated,BE >but those are counterproductive in this age of generally forged spam"	 >origins.n >p@ >Will what you are developing provide for a Teergrubing defense,A >or provide callouts for customers to mount their own Teergrubingl
 >defense ? >I >---' >Incoming mail is certified Virus Free.t; >Checked by AVG anti-virus system (http://www.grisoft.com).eA >Version: 6.0.476 / Virus Database: 273 - Release Date: 4/24/2003a >o --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.476 / Virus Database: 273 - Release Date: 4/24/2003   ------------------------------   Date: 11 May 2003 18:51:21 GMT3 From: gartmann@immunbio.mpg.de (Christoph Gartmann)e$ Subject: RE: Spamfilter for OpenVMS?0 Message-ID: <b9m639$m46$1@n.ruf.uni-freiburg.de>  _ In article <CIEJLCMNHNNDLLOOGNJIIELOHCAA.tom@kednos.com>, "Tom Linden" <tom@kednos.com> writes:iI >Speaking of which, are there any Teergruben available for VMS, TCPIP5.1?e  A Our mailserver is a MicroVAX 3100 from 1991. It is almost one ;-)a   Regards,    Christoph Gartmannn  H -- --------------------------------------------------------------------+H | Max-Planck-Institut fuer      Phone   : +49-761-5108-464   Fax: -452 |H | Immunbiologie                                                        |H | Postfach 1169                 Internet: gartmann@immunbio.mpg.de     |H | D-79011  Freiburg, Germany                                           |H +------------- http://www.immunbio.mpg.de/home/menue.html -------------+   ------------------------------  + Date: Sun, 11 May 2003 20:36:19 +0000 (UTC)tP From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig (remove CLOTHES to reply))( Subject: TCPIP SET CONF SMTP /SUB=HIDDEN$ Message-ID: <b9mc83$cer$1@online.de>  # What is /SUB=HIDDEN supposed to do?o  H /SUB=NAME=fully_qualified_domain is there to make sending and receiving E mail take place as if the name of the machine were different than it >C really is.  There is a logical to turn off this functionality when aE receiving or sending.  In addition, when receiving it is possible to  A specify which domains the machine will accept mail for in a list.p   But what about /SUB=[NO]HIDDEN?    The documentation says:e  '          o  /SUBSTITUTE_DOMAIN=NOHIDDENe  F             In the From: and Return-Path fields, displays the sender's,             name and fully qualified domain.  %          o  /SUBSTITUTE_DOMAIN=HIDDENy  E             In the From and Return-Path fields, substitutes the valueiH             given in name for the host name. The following example shows2             how this qualifiers [sic] can be used:  F I seem to remember, years ago with TCPIP 4.x, that this worked pretty C much as described, but not exactly.  NOHIDDEN worked as described, 3D HIDDEN worked just like NOHIDDEN, but added the HIDDEN address in a  Reply-To header.  F Now I see (with 5.0A---yes I will upgrade soon but want to understand G how /SUB=[NO]HIDDEN is supposed to behave first!) that the results are  E always the same with HIDDEN or NOHIDDEN---the setting doesn't matter!e  C Also, on VAX, TCPIP$SMTP_FROM doesn't seem to work while it does onp ALPHA. L   ------------------------------   End of INFO-VAX 2003.261 ************************