INFO-VAX Tue, 21 Oct 2008 Volume 2008 : Issue 568 Contents: Re: AlphaServer 1000a screen blanks after self test Re: AlphaServer 1000a screen blanks after self test Re: AlphaServer 1000a screen blanks after self test Re: AlphaServer 1000a screen blanks after self test Re: AlphaServer 1000a screen blanks after self test Re: Enhancing DCL, was: Re: How do I add 2 letters to a long RE: OpenVMS Book Wins award Re: OpenVMS Book Wins award ---------------------------------------------------------------------- Date: Mon, 20 Oct 2008 13:29:05 -0700 (PDT) From: johnwallace4@yahoo.co.uk Subject: Re: AlphaServer 1000a screen blanks after self test Message-ID: On Oct 20, 2:10 pm, jacquesal wrote: > On 19 oct, 15:56, johnwalla...@yahoo.co.uk wrote: > > > > > My vague recollection is that the LCD is driven by the main processor, > > there's no microcontroller or similar to drive the boot and diagnostic > > process, though some of the more upmarket Alphaservers do have this. > > Thus it's probably no great surprise that when the LCD doesn't do the > > expected thing, nothing else works further on than that. When you get > > something onscreen, what do you get? (I'd usually suggest using the > > serial console, except you're not getting that far... maybe disable > > the onboard VGA, *and* remove the PBXGA, as per User's Guide [1] Table > > 9-7?) > > > You did say "The memory appears to be recognized (when I can get that > > far in the boot console)" - exactly what is leading you to think > > that? > > > The Users Guide Chapter 6, Memory section, says that memory must be > > configured in groups of 5 SIMMs: "A minimum of one memory bank (five > > modules) is required.". This confirms my recollection that this system > > is intended to be used with ECC. Whether it will *work* (by any > > sensible definition of work) without ECC is a slightly different > > question, which others may be able to confirm if your Service Guide > > doesn't shed any light (especially anyone with real life AlphaServer > > 1000A experience). > > > In your situation I'd be looking for a SIMM to match any one of the > > existing ones, and reduce it to one bank of five identical SIMMs as > > the next step, just to see what the system does when not semi- > > permanently stuck in a tight loop handling uncorrectable memory errors > > (which *might* be what it's doing at the moment). > > > Don't forget that once you do get it live, you might want to check > > your firmware is reasonably recent. You are also supposed to run the > > relevant EISA config utility for VMS. This may be important even if > > you are not using any EISA cards, unless you are 100% confident you > > don't need any changes from the current EISA config; EISA config > > values can in principle change with firmware changes and OS changes, > > so it's good to be confident. Incorrect EISA config values can cause > > Bad Things to happen at inconvenient times. > > > Good luck > > John > > Hi John, > > Will try to briefly respond. Not next to the machine right now, but > the discussion about the SIMMs has come up again further along in the > thread so I'll try to summarize. > > I received the machine in this condition with 3 quadruplets of SIMMs. > Now the service guide mentions that the banks 0, 1, 2 and 3 each > contain 4 SIMMs _plus_ an additional bank of 4 ECC SIMMs (1 for each > 'ordinary' bank). The bank 0 is listed as obligatory with the one ECC > SIMM placed in a specific slot of the ECC bank in order to match it. > (If you can see this in your head, you may note that something is not > adding up already.) > > When I pulled all the memory and started from scratch, my first > attempt was to fill what I thought to be the bank 0 and place what I > presumed to be an ECC SIMM in the correct slot. When I turned the > machine on and few times and it displayed the blue boot console, it > complained that the bank 1 was filled and that the bank 0 was > mismatched! > > I then looked more closely at the mother board and realized that there > is _NO_ ECC bank, contrary to the service guide! Just banks 0 through > 3 for 'ordinary' RAM. Below the bank 0, however, one notices that > there are solder points for an eventual ECC bank. Moreover, what I > thought to be ECC SIMMs must therefore be 'ordinary' SIMMs although > I'll compare the product codes this evening just to be 100% sure. > > (Just to assure you that I'm not crazy, this is supposed to be a 1000A > and not a 1000.) > > So for troubleshooting I've placed these four non-ECC SIMMs in what is > really the bank 0 and I turn on the machine. When it successfully > starts and brings me to the boot console, it indicates the 256 MB of > RAM is present and that the other banks are empty. The chevron prompt > DOES appear. This however does NOT stop the screen from going black > after one or two minutes and the machine becoming unresponsive. > > If the memory enigma is solved, I suppose the next step is to > investigate the console vs serial option. This revision of the > motherboard does not have a jumper for onboard VGA -- in fact there is > no onboard VGA. Going back in time a bit -- when I received the > machine, it lacked a video card and I had initially connected a null > modem cable. It was already displaying the strange symptom of a blank > Operator Control Pannel (the "OCP" LCD) most of the time, so it was > not suprising that nothing showed up on the terminal (minicom on a > Linux box connected via the null modem cable). > > (As part of my diagnostic process, I had picked up a cheap PBXGA-AA/ > AN, placed it in a PCI slot and connected it to a monitor. By chance, > after a few power cycles, the OCP once displayed the self tests and > there was output to the monitor (the blue SRM console) which > eventually turned off course. This is when I began to investigate a > possible problem with the memory.) > > FYI, an additional test that I performed was to pull everything -- > memory, SCSI, floppy, keyboard, expansion cards -- just to see if the > OCP would activate in a more consistent fashion. This changed > nothing. I also tried powering it on without the CPU daughter card > and this did not help either (of course if the OCP is dependant on the > CPU, this was to be expected). > > At this point, I'm beginning to think that there is some sort of > serious hardware fault (e.g. defective power supply or crack in the > board as opposed to a bad or mismatched SIMM). My hope is that (a) > someone with a similar machine may have solved such a problem or (b) > there may be a way to diagnose the faulty hardware before I send the > machine to be recycled! > > - Alex Forgot this earlier - "there is no onboard VGA" is another difference between the EV4/21064 AlphaServer 1000A and the EV5/21164 AlphaServer 1000A, according to the Systems+Options ordering info for the later AlphaServer 1000As. So, we very likely have an EV5/21164 AlphaServer 1000A, but the only docs we have found so far relate only to the EV4 AlphaServer 1000A, unless anyone has seen docs for the EV5 flavour? I'm going to make an off the wall suggestion - take a look at the AlphaStation 600 docs instead; try http://h18002.www1.hp.com/alphaserver/workstations/retired/a600.html for starters. You will see that there were two flavours of AlphaStation 600 too. Do either of them look like your box, eg do either of the slot configurations match? The more likely match of the two, since you've already mentioned a "CPU daughtercard", is the 600A. Table 4-3 in the AlphaStation 600A Users Guide describes the beep codes. One beep on its own is said to correspond to "successful system startup", which matches what you see. There is no description for "no beeps". Both the AlphaStation 600s had a PCI bridge. These allowed more PCI slots to be used but it could also make choice of slot important. Primary slots (on the "near" side of the PCI bridge) tended to be preferred, as not everything worked right on the far side of a PCI bridge, especially with early firmware versions. You want your PBXGA to be in a primary slot, probably your SCSI adapter and eventually maybe the network card too. The reason I'm suggesting this is that I have a(nother) vague recollection that there was a similar (same?) motherboard between the EV5 AlphaServer 1000s and one of the AlphaStation 600s. Even if it's not identical, the AlphaServer 600 internals are likely going to be more relevant to you than the AlphaServer 1000 EV4 internals. I still can't work out quite why you'd get as far as the three chevrons occasionally and then after a while the monitor goes blank and the machine goes unresponsive. Are all the fans spinning properly every time you power up (some of these systems had fan failure detection which may not have caused an instant shutdown, in fact I can't remember how they would respond to fan failure, but...)? On the occasions when the system does come up OK are all the power on self test diagnostics successful? Other thoughts: You may want to check the resolution/refresh settings on the PBXGA switches definitely match something your monitor can do (some monitors are strange). You may want to make sure your monitor isn't going to be upset by "sync on green" - there is a console variable to configure this once you are up and running. Becomes irrelevant (temporarily?) if you're using the serial console. You may also want to make sure the Halt switch is out (if you have one). And 36bit SIMMs are important. Good luck. ------------------------------ Date: Mon, 20 Oct 2008 13:34:45 -0700 (PDT) From: johnwallace4@yahoo.co.uk Subject: Re: AlphaServer 1000a screen blanks after self test Message-ID: On Oct 20, 9:29 pm, johnwalla...@yahoo.co.uk wrote: > On Oct 20, 2:10 pm, jacquesal > wrote: > > > > > On 19 oct, 15:56, johnwalla...@yahoo.co.uk wrote: > > > > My vague recollection is that the LCD is driven by the main processor, > > > there's no microcontroller or similar to drive the boot and diagnostic > > > process, though some of the more upmarket Alphaservers do have this. > > > Thus it's probably no great surprise that when the LCD doesn't do the > > > expected thing, nothing else works further on than that. When you get > > > something onscreen, what do you get? (I'd usually suggest using the > > > serial console, except you're not getting that far... maybe disable > > > the onboard VGA, *and* remove the PBXGA, as per User's Guide [1] Table > > > 9-7?) > > > > You did say "The memory appears to be recognized (when I can get that > > > far in the boot console)" - exactly what is leading you to think > > > that? > > > > The Users Guide Chapter 6, Memory section, says that memory must be > > > configured in groups of 5 SIMMs: "A minimum of one memory bank (five > > > modules) is required.". This confirms my recollection that this system > > > is intended to be used with ECC. Whether it will *work* (by any > > > sensible definition of work) without ECC is a slightly different > > > question, which others may be able to confirm if your Service Guide > > > doesn't shed any light (especially anyone with real life AlphaServer > > > 1000A experience). > > > > In your situation I'd be looking for a SIMM to match any one of the > > > existing ones, and reduce it to one bank of five identical SIMMs as > > > the next step, just to see what the system does when not semi- > > > permanently stuck in a tight loop handling uncorrectable memory errors > > > (which *might* be what it's doing at the moment). > > > > Don't forget that once you do get it live, you might want to check > > > your firmware is reasonably recent. You are also supposed to run the > > > relevant EISA config utility for VMS. This may be important even if > > > you are not using any EISA cards, unless you are 100% confident you > > > don't need any changes from the current EISA config; EISA config > > > values can in principle change with firmware changes and OS changes, > > > so it's good to be confident. Incorrect EISA config values can cause > > > Bad Things to happen at inconvenient times. > > > > Good luck > > > John > > > Hi John, > > > Will try to briefly respond. Not next to the machine right now, but > > the discussion about the SIMMs has come up again further along in the > > thread so I'll try to summarize. > > > I received the machine in this condition with 3 quadruplets of SIMMs. > > Now the service guide mentions that the banks 0, 1, 2 and 3 each > > contain 4 SIMMs _plus_ an additional bank of 4 ECC SIMMs (1 for each > > 'ordinary' bank). The bank 0 is listed as obligatory with the one ECC > > SIMM placed in a specific slot of the ECC bank in order to match it. > > (If you can see this in your head, you may note that something is not > > adding up already.) > > > When I pulled all the memory and started from scratch, my first > > attempt was to fill what I thought to be the bank 0 and place what I > > presumed to be an ECC SIMM in the correct slot. When I turned the > > machine on and few times and it displayed the blue boot console, it > > complained that the bank 1 was filled and that the bank 0 was > > mismatched! > > > I then looked more closely at the mother board and realized that there > > is _NO_ ECC bank, contrary to the service guide! Just banks 0 through > > 3 for 'ordinary' RAM. Below the bank 0, however, one notices that > > there are solder points for an eventual ECC bank. Moreover, what I > > thought to be ECC SIMMs must therefore be 'ordinary' SIMMs although > > I'll compare the product codes this evening just to be 100% sure. > > > (Just to assure you that I'm not crazy, this is supposed to be a 1000A > > and not a 1000.) > > > So for troubleshooting I've placed these four non-ECC SIMMs in what is > > really the bank 0 and I turn on the machine. When it successfully > > starts and brings me to the boot console, it indicates the 256 MB of > > RAM is present and that the other banks are empty. The chevron prompt > > DOES appear. This however does NOT stop the screen from going black > > after one or two minutes and the machine becoming unresponsive. > > > If the memory enigma is solved, I suppose the next step is to > > investigate the console vs serial option. This revision of the > > motherboard does not have a jumper for onboard VGA -- in fact there is > > no onboard VGA. Going back in time a bit -- when I received the > > machine, it lacked a video card and I had initially connected a null > > modem cable. It was already displaying the strange symptom of a blank > > Operator Control Pannel (the "OCP" LCD) most of the time, so it was > > not suprising that nothing showed up on the terminal (minicom on a > > Linux box connected via the null modem cable). > > > (As part of my diagnostic process, I had picked up a cheap PBXGA-AA/ > > AN, placed it in a PCI slot and connected it to a monitor. By chance, > > after a few power cycles, the OCP once displayed the self tests and > > there was output to the monitor (the blue SRM console) which > > eventually turned off course. This is when I began to investigate a > > possible problem with the memory.) > > > FYI, an additional test that I performed was to pull everything -- > > memory, SCSI, floppy, keyboard, expansion cards -- just to see if the > > OCP would activate in a more consistent fashion. This changed > > nothing. I also tried powering it on without the CPU daughter card > > and this did not help either (of course if the OCP is dependant on the > > CPU, this was to be expected). > > > At this point, I'm beginning to think that there is some sort of > > serious hardware fault (e.g. defective power supply or crack in the > > board as opposed to a bad or mismatched SIMM). My hope is that (a) > > someone with a similar machine may have solved such a problem or (b) > > there may be a way to diagnose the faulty hardware before I send the > > machine to be recycled! > > > - Alex > > Forgot this earlier - "there is no onboard VGA" is another difference > between the EV4/21064 AlphaServer 1000A and the EV5/21164 AlphaServer > 1000A, according to the Systems+Options ordering info for the later > AlphaServer 1000As. > > So, we very likely have an EV5/21164 AlphaServer 1000A, but the only > docs we have found so far relate only to the EV4 AlphaServer 1000A, > unless anyone has seen docs for the EV5 flavour? > > I'm going to make an off the wall suggestion - take a look at the > AlphaStation 600 docs instead; tryhttp://h18002.www1.hp.com/alphaserver/workstations/retired/a600.html > for starters. You will see that there were two flavours of > AlphaStation 600 too. Do either of them look like your box, eg do > either of the slot configurations match? The more likely match of the > two, since you've already mentioned a "CPU daughtercard", is the 600A. > > Table 4-3 in the AlphaStation 600A Users Guide describes the beep > codes. One beep on its own is said to correspond to "successful system > startup", which matches what you see. There is no description for "no > beeps". > > Both the AlphaStation 600s had a PCI bridge. These allowed more PCI > slots to be used but it could also make choice of slot important. > Primary slots (on the "near" side of the PCI bridge) tended to be > preferred, as not everything worked right on the far side of a PCI > bridge, especially with early firmware versions. You want your PBXGA > to be in a primary slot, probably your SCSI adapter and eventually > maybe the network card too. > > The reason I'm suggesting this is that I have a(nother) vague > recollection that there was a similar (same?) motherboard between the > EV5 AlphaServer 1000s and one of the AlphaStation 600s. Even if it's > not identical, the AlphaServer 600 internals are likely going to be > more relevant to you than the AlphaServer 1000 EV4 internals. > > I still can't work out quite why you'd get as far as the three > chevrons occasionally and then after a while the monitor goes blank > and the machine goes unresponsive. Are all the fans spinning properly > every time you power up (some of these systems had fan failure > detection which may not have caused an instant shutdown, in fact I > can't remember how they would respond to fan failure, but...)? > > On the occasions when the system does come up OK are all the power on > self test diagnostics successful? > > Other thoughts: You may want to check the resolution/refresh settings > on the PBXGA switches definitely match something your monitor can do > (some monitors are strange). You may want to make sure your monitor > isn't going to be upset by "sync on green" - there is a console > variable to configure this once you are up and running. Becomes > irrelevant (temporarily?) if you're using the serial console. > > You may also want to make sure the Halt switch is out (if you have > one). > > And 36bit SIMMs are important. > > Good luck. And something else I forgot: the 600A manual, table 7-10, describes jumper settings including those used to enable and disable system shutdown on overtemperature and fan failure. For now, I'd want them set to disable the usual fan failure and overtemperature shutdown. ------------------------------ Date: Mon, 20 Oct 2008 17:27:14 -0400 From: JF Mezei Subject: Re: AlphaServer 1000a screen blanks after self test Message-ID: <48fcf987$0$10458$c3e8da3@news.astraweb.com> jacquesal wrote: > RAM is present and that the other banks are empty. The chevron prompt > DOES appear. This however does NOT stop the screen from going black > after one or two minutes and the machine becoming unresponsive. You might wish to time precicely the time between chevron appearing (or power on) and the time the screen goes blank. If it is some constant precise value like 1:30 seconds every time, it could mean some internal timer deciding something is wrong. If it is a variable amount of time and the machine lasts longer if it has been left powered off and cool for a while then it might be something about heat. For instance, perhaps the CPU is overheating and there is an automatic shut off ? (is the CPU fan OK, is it full of lint inside ? ------------------------------ Date: Mon, 20 Oct 2008 18:09:39 -0700 (PDT) From: jacquesal Subject: Re: AlphaServer 1000a screen blanks after self test Message-ID: <0d887b0a-2017-4d7e-9de4-e1e994db4d5a@u46g2000hsc.googlegroups.com> Alright, I'm going to try to respond to all of the questions that I've seen in the thread. Thanks for you help folks! 1. It is a pedestal system. (Steve) 2. The memory that I found installed: (johnwalla... sorry I don't see your proper name!) 4x 32 MB Viking Components unknown model 4x 8 MB Digital 54-21246-BA (36 bit according to Google) 4x 256 MB Dataram 61990 AO (36 bit as well, same source) 3. The history of the box (johnwalla...) It was a server running *BSD at a certain engineering in the United States. When they dumped it, a former student took it and left it in his closet until he gave it to me a few years later. I don't think that he even turned it on. 4. System unit part numbers (johnwalla...) Exterior: model no PB76C-FA Rev A01 Backplane: 54-23499-02 A01 CPU daughtercard: 54-24799-02 C02 and the "glue logic" chips show 21172-BA CPU: 21-43918-01 revision D (I believe that we're looking at a 21164 EV5 so it may have been field upgraded.) 5. set console graphics \ init (David Turner) I will plug this command in straight away just to be sure -- once the unit powers up again! Will keep in mind the S3Trio64 if and when I determine that the system is salvageable. Moreover, the PBXGA has jumpers that may be poorly set since I can't get my hands on a manual. However, since the system doesn't like to turn on correctly (with or without the card), I sense that that the problem lies elsewhere. 6. The video card is installed in slot 11 (primary). 7. Monitor -- it's a recent, multisync LCD that supports sync-on-green (although I'm not sure that this is necessary given the card). When it do get a console display, there is no flickering and the image is well, just perfect. Until it goes black. 8. Time the delay and check CPU temp (JF Mezei) Good idea, JF. The next time it powers up properly (over the past 20 power cycles, I haven't gotten even that far), I will time the delay and try the command for the temperature. 7. Use the 600A owner's guide (johnwalla...) I've looked through it. The memory configuration looks identical to mine and the troubleshooting guide is better written than that of the 1000a... But it still remains that the 1 beep (_only_ audible when I've obtained a "successful" power up) does not correspond to list error codes. I'm going to find out about disabling the fan tests. All fans appear to function well. Will report back! :-) - Alex ------------------------------ Date: Mon, 20 Oct 2008 18:48:58 -0700 (PDT) From: jacquesal Subject: Re: AlphaServer 1000a screen blanks after self test Message-ID: <618163fd-80df-4b9b-9868-92f60adcc92c@p49g2000hsd.googlegroups.com> On Oct 20, 9:09=A0pm, jacquesal wrote: > I'm going to find out about disabling the fan tests. All fans appear > to function well. Will report back! :-) I disabled J52 (temperature shutdown) and put the daughtercard boot jumper in fail-safe mode. The OCP lit up straight away and I heard 1-3-3 (bad RAM), before the self tests stalled! I looked at the RAM and noticed that one SIMM was poorly inserted (I had pulled it to see the model number) so I turned the machine off and reseated the SIMM. I have had no luck getting the system to power on correctly again (and I've tried six or seven times). I've disabled the fan checks and have played around with the boot mode jumpers on the CPU daughtercard. The fans turn on, but no OCP and no console. Frustrating! - Alex ------------------------------ Date: Mon, 20 Oct 2008 17:31:22 -0400 From: JF Mezei Subject: Re: Enhancing DCL, was: Re: How do I add 2 letters to a long Message-ID: <48fcfa80$0$10458$c3e8da3@news.astraweb.com> George Cornelius wrote: > This works: > > $ pipe recall/out=x.lis ; search x.lis sea Or, with a sufficiently recent version of VMS: recall/search sea ------------------------------ Date: Mon, 20 Oct 2008 18:21:47 +0000 From: "Main, Kerry" Subject: RE: OpenVMS Book Wins award Message-ID: <9D02E14BC0A2AE43A5D16A4CD8EC5A593ED7607449@GVW1158EXB.americas.hpqcorp.net> > -----Original Message----- > From: yyyc186 [mailto:yyyc186@hughes.net] > Sent: Monday, October 20, 2008 1:37 PM > To: Info-VAX@Mvb.Saic.Com > Subject: OpenVMS Book Wins award > > The Minimum You Need to Know About Service Orieted Architecture by > Roland Hughes > > Award-Winner in the Business: Technology/Computers/Internet category > of the National Best Books 2008 Awards, sponsored by USA Book News > > You can find this book in Island Computer's Web store. Hey, very coooll .. :-) Additional info: http://www.usabooknews.com/bestbooksawards2008.html scroll down to: Business: Technology/Computers/Internet Winner: The Minimum You Need to Know About Service Oriented Architecture by Roland = Hughes Logikal Solutions ISBN 978-0-9770866-6-5 Some earlier feedback on the book: http://www.openvms.org/stories.php?story=3D08/09/27/8565098 Regards Kerry Main Senior Consultant HP Services Canada Voice: 613-254-8911 Fax: 613-591-4477 kerryDOTmainAThpDOTcom (remove the DOT's and AT) OpenVMS - the secure, multi-site OS that just works. ------------------------------ Date: Mon, 20 Oct 2008 14:37:42 -0700 (PDT) From: Rich Jordan Subject: Re: OpenVMS Book Wins award Message-ID: <7587ac72-0e8a-4294-8209-a0c5697fc783@e38g2000prn.googlegroups.com> On Oct 20, 12:36=A0pm, yyyc186 wrote: > The Minimum You Need to Know About Service Orieted Architecture by > Roland Hughes > > Award-Winner in the Business: Technology/Computers/Internet category > of the National Best Books 2008 Awards, sponsored by USA Book News > > You can find this book in Island Computer's Web store. Congratulations, Roland. That is wonderful. Rich ------------------------------ End of INFO-VAX 2008.568 ************************