[net.periphs] sc750/Eagle installation

boylan@dicomed.UUCP (Chris Boylan) (06/02/84)

My thanks to all of you who responded for my request for info
on installing our sc750/eagle and those who assured me the drive
would survive being shipped on it's ear (it did).

Here are a couple of interesting things I discovered in the process
of installing it.  By the way, we installed it ourselves because
the local DEC office was reluctant to do it because of some past
problems and everyone else I talked to wanted >1K to do it and would
not be able to for another week ... sigh.  It was easy though.

the Fujitsu manual gives incorrect sectoring switch settings for
using 48-sector tracks with the Emulex. The Emulex manual gives
the correction.  (It doesn't tell you that it is a correct, it just
tells you to do it.)

Several people suggested that I specify severe burnin when formatting
to avoid future problems.  I agree with this because it IS a pain
to have flaw out bad sectors periodically.

Something that isn't mentioned in either manual is that the Eagle
is shipped with tag lines 4 and 5 enabled.  These additional status
lines are written up as something different in the Emulex manual
so I disabled them, which made them match the Emulex document.  A
engineer friend of mine tells me that this could cause random flakeiness
in the controller (if enabled) although he hasn't seen the schematics
so he couldn't rule out the possibility that they handled them.

The Fuji manual tells you to ground the drive to the controller
ground.  Based upon conversations with the local DEC people I grounded
it to the frame.  Frankly, I don't know whether this is right but
it shouldn't be a big deal.  If anyone knows for sure that this
should be done differently, please post something.

And last, a bug report:

As long as the controller was on the first MBA everything went fine.
There is a bug in the standalone hp driver that won't let it boot off
of any other MBA. You  probably won't need this now but it may
save you trouble later. here is the fix for the above problem.


RCS file: RCS/hp.c,v
retrieving revision 1.2
diff  -r1.2 hp.c
98c98
< 		hp_type[unit] = hpmaptype(hpaddr, i, UNITTODRIVE);
---
> 		hp_type[unit] = hpmaptype(hpaddr, i, UNITTODRIVE(unit));

-- 

	Chris Boylan
	{mgnetp | ihnp4 | uwvax}!dicomed!boylan