odd results from f$getsyi("power_vector")

[Adrian Graham_1]

Hi folks,

I know the fan_vector and power_vector calls from f$getsyi() are unsupported but the machines I need to use them on (ES40s and 4100s) can't have WEBES installed because it needs an ODS-5 volume for JAVA support and customer is as yet unwilling to convert - "if it ain't broke."

Anyway, does anyone know why fan_vector only returns status for 2 fans on a 4100 (3 present and running) and why power_vector returns status for 5 power supplies in an ES40 with 3 physical PSUs?

Even the sample C routine on the HP website only returns status for 2 fans on a 4100...

Thanks in advance!

[labadie_1]

You do not need to convert a disk to Ods5.

If you have some free blocks on a disk, use LD
$ ld create disk:ods5.dsk/size=400000
$ ld connect disk:ods5.dsk
-> create a LDAx device
$ init/struct=5/nohigh/... LDAx: webes
$ mount/sys LDAx: webes

and now you install WEBES in the LDAx device

Ld is hidden, but present since Alpha Vms 7.3-1, just start it with
$ @sys$startup:ld$startup

And you can always get LD from
http://h71000.www7.hp.com/freeware/freeware70/ld/

[Richard Whalen]

I don't have a guess for why the fan_vector would return information about fewer fans than there are, but the power vector may be returning information about the maximum possible power supplies on any system.

[Hoff]

The vectors return what status is found out on the I2C bus. The AlphaServer 4100 and AlphaServer ES40 are among the boxes that do have the I2C and the vectors.

Take a look at the hardware service manuals for the boxes, and find out of the fans are all connected up and wired correctly, and if the correct types of fans are installed, etc.

IIRC, in some cases in some configurations, you won't get "readings" from the vectors, you'll just get an indication of "presence" or "absence".

Scott Belviso posted a ENV_CHECK.COM tool over at dcl.OpenVMS.org a while back which can decode the vectors, and there's a full C source module at posted at the old Ask The Wizard area in topic (2945) -- this is the C code I assume you are referring to...
http://h71000.www7.hp.com/wizard/wiz_2945.html

I won't comment on WBEM/WEBES.

ODS-5, on the other hand, is something I've been running as a data disk and as a system disk forever. It works. It's really just ODS-2 with a couple of added structures, and you're already using the "new" parsing code and selector mechanisms...

As for LD, do get the driver from Jur's site (newest) or from any available ECO for LD for the version of OpenVMS in use here (newer). The versions of LD up past V7.3-2 had some built-in bugs.

LD 9.0 http://www.digiater.nl/lddriver.html is current.

Stephen Hoffman
HoffmanLabs LLC

[Dean McGorrill]

Hi Adrian,
I use the temperature_vector to look
at my remote sites, and its not always accurate either. eh its unsupported, however
I caught an ac failure with it.

[Ian Miller.]

IIRC one of the reasons that those calls are not supported is that the results very dependent on the hardware model and configuration so interpreting what you get is not obvious at times.

Use the supported product - by avoiding it you are increasing risk not reducing it.

[Adrian Graham_1]

Hi Ian and folks,

It's not a case of avoiding WEBES, it's a case of convincing this rather large customer that these tools can be a benefit; at present they're happy for us to monitor etc so while we talk to them about the extra support tools I can use the outputs I get from the vectors as a baseline for a working machine so if anything changes it can be spotted. Of course, with the 4100 if a fan dies so does the machine....

I'd seen Stephen's monitoring DCL script after I'd written my own - the two are remarkably similar but there isn't really a different way of looking at the vectors and outputting results :)

Cheers!

I'll set the LD stuff up on a DS10 here and see what's what.

[Adrian Graham_1]

I meant, of course, Scott's monitoring script, not Stephens! It's friday :)