Diagnostic tool for alphastation (openvms)

[Ferdinand Macatangay]

Hello,

Anybody knows a diagnostic software in openvms that I can run to test my memory and other hardwares?

Thanks and Regards,
Ferdie

[Andy Bustamante]

You can try memexer from the console prompt to test memory. For other hardware, see the System Manager's Manual vol 2 at http://h71000.www7.hp.com/doc/index.html. UETP is included with the operating system. If you're having outages, reviewing the crash dump and error logs may identify issues.

What makes you think you have a memory or hardware error?

[Hoff]

Which AlphaStation?
What error(s) or symptom(s)?
Checked the console diagnostics?
Checked the service manual for the AlphaStation?

[Ferdinand Macatangay]

How do I go to the console diagnostic? I do not have the Alpha SDK and Firmware Update cd to boot. My srm console dont have the diagnostic commands.

[Steven Schweda]

> How do I go to the console diagnostic?

First, answer the basic questions:

> Which AlphaStation?
> What error(s) or symptom(s)?


> My srm console dont have the diagnostic
> commands.

What leads you to say that?

>>>help

[Bob Blunt]

Your Alpha goes through the console diagnostics, albeit somewhat limited, when you init the processor. There are some built-in tools within SRM that don't require a separate CD but each system's setup and commands can be different. I've never encountered any true *diagnostic* software for any Alpha system like we had for the VAXen. No stack of floppies, no special bootable CDs or DVDs.

There were two *exerciser* tools, one called DECvet and Qvet that were much more intense than UETP but those were usually only used to test new installations by services for customers under contract. These tools were not generally available. Some of the magic incantations that are possibly processor-specific may be available in the service guide for your system, so you might try to purchase one of those if you're not covered by a maintenance contract.

[Ferdinand Macatangay]

i want to run some diagnostic to run on the motherboard because the error is intermittent. the error message on the analyze/sys shows it is a cache problem(as volker helps me to find) but I can not turn on the alpha and wait for it to fault out again because I am trying to troubleshoot the board. there is no diag commands in my srm console. already tried the help command.

Specs:
Openvms 6.1
alphastation 255/300
SRM firmware ver 7

[Steven Schweda]

> SRM firmware ver 7

What does it _really_ say?

> there is no diag commands in my srm
> console. already tried the help command.

And what did you see when you tried it?

[Volker Halle]

Ferdie,

don't spend too much time trying to isolate an intermittent Memory or B-Cache error using diagnostics. This seldom works for intermittent problems...

$ TYPE CLUE$HISTORY will show you the crash history and you can directly see the frequencies of this problem.

If this machine is important, consider to swap both memory and motherboard. Or consider replacement with an emulator (like CHARON-AXP).

Volker.

PS: This is related to the following entry in ITRC:

http://forums13.itrc.hp.com/service/forums/questionanswer.do?threadId=1408470

[Hoff]

Re-seat memory and PCI boards.

Then (if the error persists) replace the memory.

Then replace the main board.

Then move forward to newer Alpha. Or port to Itanium. Or move to an emulator. Or port off of VMS.

Researching and servicing a fifteen year old AlphaStation 255 hardware running V6.2-1H3 that's also tossing machine check errors is an effort worth less than what it might cost to replace the box or to port the software.

The (repair) value of a fifteen year old server is limited.

A (used) Itanium box regularly show up for US$200 to US$500 from various sources, and vastly newer low-end Alpha boxes tend to be in the same price range and less.

It might well be cheapest to find a junker AlphaStation 255 and start swapping parts wholesale.

Expect to replace any disks that are even a fraction as old as this server.

[Bob Blunt]

Ferdinand, you're asking for help and you're not helping us help you. First and foremost you're asking in a user-supported forum at no cost to you. Noone, not even HP, Compaq or DEC with your paid hardware support contract could tell you more with the information you've provided. We could guess for hours and give you reams of wrong information. We're not asking for fun. The SRM consoles are different depending on the type Alpha system or family. Yup, they're similar but they're not identical. The console on a GS80/160/320 is not the same as one on an AlphaServer400.

Second, the Alpha systems don't have diagnostics like the VAX did. There aren't bootable floppies, tapes or CDs. None. The diagnostics are built into the console and we have to know what type system so the right command can be issued to access the diagnostics. And, to be honest, that sort of information isn't usually shared with customers even if they have a maintenance contract.

[Ferdinand Macatangay]

Firstly I want to give thanks for all the answers you have feedback.as what i have said in the previous post my SRM is V7.0 ,Running on 6.1 openvms alphastation 255/233.There are no diagnostic commands inside the console(no memexer, no test,no memtest,no nettest,no sys_exer) when i tried to type in help there are no diagnostic commands listed there.

[Steven Schweda]

> There are no diagnostic commands [...]

Ok, you win. I woke up one of my AlpSta 200
4/233 systems, and there's not much in there.
Newer systems (PWS, XPxxx) have more
diagnostics in the SRM console.

However, unless you have
quick_start ON
I'd guess that a lot of the time at power-up
is spent doing memory testing.

> [...] it is a cache problem [...]

If so, then a normal memory test might not
tell you much, anyway. These things use ECC
memory (yours is ECC, right?), so most simple
memory problems are not fatal. If the thing
is crashing, then I'd guess that the problem
is not a simple memory problem.

[Hoff]

"DRAM Errors in the Wild: A Large-Scale Field Study" <>

Turns out hard errors can be more common than was classically assumed, and that soft errors are rather common, and that ECC/EDC isn't a panacea, and options with Chipkill can be a better choice.

Also shows that the older the memory gear gets, the worse it gets.