[SunRescue] Q on "optimal" OS for Sun4c machines, now that So laris 8 won't runn

[BSD Bob the old greybeard BSD freak]

> >>I keep clones of the OS HD's
> 
> Is there anything special you're using to do this or are you backing up the
> drive and restoring onto the other?

My goal on the old hardware, is to keep a spare clone drive on the shelf
for any system that I consider must be quickly on-line (my home web
server, my home ftp archive server, my system development boxes, and
my writing boxes).  To do that, I generically keep a system drive of
each type set up as follows:

1.  the drive is a full classical single-drive built unix system,
    complete and bootable and what I call a minimal system drive.

2.  make it bootable and set up as ID0, with a root, swap, var and usr
    file systems.  It does not matter what the drive is, just that it
    is sufficiently large to contain the desired file systems, and
    come up with all needed system scripts set, and have little free
    space for occasional addins in /usr/local.

3.  write a set of fstab files such that the default bootable fstab
    is set for hd0 file systems only.  Then have additional fstabs
    for mounting hd1-hd7, as may be needed for different machines.
    Once up, cp the desired fstab over the orignal fstab and remount.

4.  set up the etc scripts to contain all needed configuration files
    for a generic system type (e.g., sun3, sun4, sun4c, decstation,
    vax, whatever) set up to a base default system configuration on
    my intranet, on an open IP address.  Once up and installed, I
    can quickly adjust half a dozen scripts and and it then can be
    any of a dozen or more machines of its architecture class.
    Once the target system is set, reboot and it is up and running.

5.  set up a set of disklabels in /root or wherever so that I can
    quickly set up any additional drives by merely ivoking the
    disklabel on the proper protofile and newfsing and mounting.

In addition to the system drives, I keep a few blank drives that have
been freshly lowlevelformatted on my VAXen as emergency spares for
other mounted drives.

That works for me.  I run about a dozen Sunboxes of 3/4/4c class,
half a dozen VAXen, half a dozen MIPS/Alphas, and some odd x86 based
Free/Net/OpenBSD toyz.  That usually means I keep 5 different class
architecure system disks on the shelf, and a couple of spare blanks
drives, and if luck is with me, I can have any of those machines up
and rebuilt and restored in about 15 minutes.  The odd things, I
keep the sytem files on-line in reserve, and can boot from a floppy
or tape, and reinstall on-line in about 30 minutes.

It is crude, and probably not the most efficient way to do it, but
it is the handiest way to do it for may random assortment of junque.
It is so easy just to reach over on the shelf and grab a fully
installed operational system drive and just chuck it in and go.

> >>The hardware, other than HD's, seems to run forever, but, I usually
> >>wind up nuking a few HD's a year from motor/platter/powerburp bellyup.
> >>The old drives are good, but, reaching their end of life.  But, at
> >>a buck or two each in surplus, I find it good karma to just keep
> >>a few clones handy, on the shelf, ready to load in a couple of minutes.
> 
> LOL!  I've just lost about 3 of my older PC hard drives this year but _none_
> of my SCSI drives in my Sun/SGI boxen.

Well, most of my drives are old surplus things, anyway, some as old as
12-15 years, so my end-of-life rate is a little higher than if I were
to buy new stuff.  Interestingly, my PC unix dieoff rate is about the
same on unix boxes, but I don't run any windoz boxes to compare that
against, and wouldn't want to, either....(:+}}.  My ESDI unix drives
seem to hold up well, but the old MFM drives give me occasional fits.
I usually try to refit scsi if possible on those boxes.

> Seriously though I think I've had maybe two SCSI drives fail on me in the
> whole 10 years I've been doing this.  I think they're just made better (for
> the most part).  Although I think in recent years they've begun just putting
> SCSI hardware on IDE drives - I think someone mentioned IBM doing this.
> That means we _can_ expect the reliability of SCSI drives in general to go
> down, maybe not fast but it will happen unless something prevents it.

Yes, very true, and I get very good service out of my scsi drives.  But,
they were castoff surplus when I got them, so I really can't expect much
better life that what I get out of them.  I usually consider that if my
VAXen cannot lowlevelformat them, then they are toast, and real dumpster
fodder.  If the VAXen can run them up, then I can reasonably expect maybe
two years continuous duty out of them, but usually get more than that,
on any of the junque unix toyz.

> Anyway the drive swap out idea is a good one.  The moving parts in the hard
> drive make it far more prone to failure than anything else in the machine
> although there is much to be said for keeping the case extremely cool to
> prevent things like the drive spindle motor from heating up.  (The 800meg in
> my SS2 generates _massive_ amounts of heat!)  If I could I'd seriously
> consider mounting my Sparcs with extra ventilation holes cut into them and
> fans.  If the FCC wouldn't b**** I'd run them in a rack without the covers.
> I don't think the noise of one would bother the others too much.

I don't like to put the high speed drives in my pizza boxes.  I tend to
get much better heat tolerance out of using the older DEC RZ26 or quantum
or seagate equivalent type drives as opposed to the Maxtors and the HP
drives that one can fry an egg on.  I usually mount the hot things
in external boxes with better cooling.  On the old VME crates, they
can handle most anything.

> Of course I could always talk to www.kryotech.com for a custom cooling
> solution. ;-)

Nah, pentiumtoyz need that kind of horsepower.  Old sparcs or 68K toyz
know how to dissipate the heat and survive.....(:+}}...  My alpha could
probably use a head-cooler, though.  It does get rather hot-headed.

Bob