[rescue] Vendor software support [was Re: RPi vs SS20 benchmarks (interesting)]

[Jonathan Patschke]

On Mon, 4 Jan 2016, Dave McGuire wrote:

>  In my experience, "in the old days" (this may have improved later on)
> when SPARCstation-10s were current, Sun employees were the people who
> knew the least about Sun hardware.

I take it you've never experienced the fun of aftermarket warranties.

At $position[-2], we had Northrup doing on-call service for some
UltraSPARC-III machines (presumably because they were cheaper than Sun and
only marginally more expensive than a well-stocked spares cabinet).
What started as a bad memory module on a Blade 2000 escalated to a full
system replacement when their service tech ignored the torque spec and
tightening instructions for the CPU card...

...a CPU that he pulled and replaced because he misunderstood the memory
test message to be a cache failure message.

My experience has been that hardware support is only useful for being able
to order replacement parts from the local depot, as opposed to ordering
through the catalogue.  Occasionally they send a minion out who is
capable of turning a screwdriver, and that has the potential to save time.

At that same position, we had a ton of Dell systems that booted from the
network (using the disks as non-bootable cache).  That flummoxed the poor
field servoids, so I wrote a boot-sector program which we mass-installed
to inform the tech that the machine was actually okay and that I would
sign-off on it, if they saw the message.


Ed Mitchell (formerly of this list) had a funny story about Sun hardware
support wherein they replaced a burnt-out V240 power supply with a memory
module.  The tech had come out, diagnosed the problem, and ordered the
wrong part for Ed to install upon receipt, as they're marked
customer-serviceable.  I seem to recall that he sent them a photo of the
memory module lying in the power supply slot with the caption, "failure
condition persists."


-- 
Jonathan Patschke
Austin, TX
USA