According to james <at>:
> DoN. Nichols wrote:
> <snip>
> >>> smartmontools-5.33

[ ... ]

> > Of course, only some drives recognize it. It depends on just
> > how old your drives are.
>
> ok.. now this is odd. i got it installed and, supprising enough, got
> some readings. all have a shutdown temp of 65c and the front drives
> range from 42-44c while the back range from 49-51c. none ever got above
> 51 at any time when it was shutting down on me. may be over heating
> components rather than the drive shutting itself down. i also notice
> that most of the drives showed a "powered on" time of around 200k hours
> while the top and bottom drives in the back showed 15-17k hours... i'm
> assuming that's the amount of time the box has been on rather than
> runtime of the life of the drive?

Hmm ... that is on the order of two years steady on time. Has
the box really been on that long? And the 200K hours works out to 22.82
years. That does not sound even remotely possible, even if it is the
sum of all powered on time since they were manufactured.

Is it possible that you have been getting power glitches which
may have corrupted the NVRAM in the drives?

> <snip>
> > Also -- be very careful not to put drives so equipped into the
> > bottom slots on a MultiPack housing, as it will catch on the RFI
> > flashing on the bottom of the doorway, and you will have great
> > difficulty getting the drive back out.
>
> no such attachments on them.

Good. They would be a big help with the A1000/D1000 enclosure,
but a big problem with the Multipack.

> <snip>
> > BTW -- what is the power rating on the drives you have in there?
>
> noticing the "powered on" time discrepency i'm lead to believe that
> maybe the Multipack's power supply may be the issue. not able to handle
> the load?

That -- or poor connections on the power connector to the
backplane, or to the individual drives.

Back in the days when I was using AT&T UnixPC (7300/3B1)
machines (68010 CPUs), one of the failure modes was overheating of the
connector from the power supply to the motherboard. It was a fairly
stiff ribbon cable, with an IDC single-row connector from the
motherboard which plugged onto a row of pins sticking up from the power
supply. The 7300 version had the power for the disk drive (all of 20
MB) running through this cable and connector. The 3B1, which had up to
a 67 MB drive, had a separate cable directly from the power supply to
the drive.

When the system got old, the pins on the connector would
oxidize, and start to overheat. Depending on which pin it was, it could
glitch memory, the disk controller, or the CPU itself. And at least
once, the heat on the pin melted back the solder connecting that pin to
the traces on the CPU board, leaving it arcing and seriously glitching
things.

At the time, the best cure was a contact treatment called
"Cramolin", which has since been replaced by "DeOxIt" from the same
company, as the original Cramolin apparently fell afoul of the anti-CFC
rules which make Freon so hard to get these days. You would spray both
connectors, mate and un-mate one several times to break down the oxide
film, wipe off what you could access, and then re-spray with a new
treatment of Cramolin before reassembling. (They supplied two colors,
red and blue, with the one being for the first treatment, and the other
for the final assembly. In reality, they differed only in the color
additives, but this was their way of making sure that you gave two
applications with a wipe-off between. :-)

You might check the connector from the power supply to the
backplane on your Multipack to see if the pins show any signs of
overheating. If so, some similar high-grade contact treatment might
help, or you might do well to pick up a spare Multipack and swap the
drive into it.

While you have the box apart, spraying the 80-pin connectors
into which the drives plug will probably not hurt.

Be careful when you disassemble and reassemble the Multipack
housing to not lose the clear plastic moulded piece which goes over the
activity LEDs to pipe the light out the front of the box.

> <snip>
> > Will the DEC housing hold the other drives which you will be
> > pulling from the Multipack? If so, this might be a good way to go, as
> > the DEC housing may well have better cooling airflow.
>
> the drives i'm using in the Multipack are from an Alpha that has the
> same setup as the DEC enclosure. the drives are actually interchangable
> between my other running AlphaServer 1200 and the enclosure.
> the DEC has 2 "squirrel cage" fans in the back that are about 6" in
> diameter. made for volume rather than speed... i'll have to do a
> little more tinkering to figure this out.

O.K. I suspect problems with the power supply in the Multipack,
and the easiest fix is to pick up a used one from eBay as a replacement.

> thanks again. i'll let you know if the same drives work in the DEC. if
> so then i'll just put the Multipack back on my Ultra 10. btw, know of a
> software stripe solution (free) for Solaris 10?

Start with "man metainit". That will point you to a bunch of
other commands (in the "SEE ALSO" section):

================================================== ====================
SEE ALSO
mdmonitord(1M), metaclear(1M), metadb(1M), metadetach(1M),
metahs(1M), metaoffline(1M), metaonline(1M), metaparam(1M),
metarecover(1M), metarename(1M), metareplace(1M),
metaroot(1M), metaset(1M), metassist(1M), metastat(1M),
metasync(1M), metattach(1M), md.tab(4), md.cf(4),
mddb.cf(4), md.tab(4), attributes(5), md(7D)

Solaris Volume Manager Administration Guide
================================================== ====================

It is amazing the number of things which are buried in Solaris
10 which are not obvious until someone else stumbles across them and
points them out to you. :-)

Good luck,
DoN.
--
Email: <dnichols@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---