> Today I've had two sparcstations (1 & 1+) crash after
> their user defined a new swap area using
> mkfile -n 16m /fewiofj
> and modifying /etc/fstab to refer to the new swap and putting swapon -a
> in /etc/rc.
> If the machine only crashed, I wouldn't mention it, but in
> one case hitting a key on the keyboard caused a watchdog reset as
> soon as the machine came up multi-user. And this was repeatable.
> On the Sparcstation 1+, the sync to the screen was gone
> and so was the OS: ping showed the machine down and L1-A didn't
> work. After turning the machine Off/On, we rebooted and the
> problem (with sync) reappeared. Another Off/On and we lost
> one internal disk (sd1), at this point I learned about the
> new swap area and commented it out from fstab. A bit too
> late since that swap area was on the 2nd disk, the one we lost!
I guess I hadn't made it clear enough, but I knew that
the problem was the mkfile -n. Indeed lots of respondent
told me just that.
From: pjs@euclid.JPL.NASA.GOV (Peter Scott)
According to the 4.1 Release Notes, Appendix D, p. 134, you don't
use the -n flag on Sparcstations.
(Of course I should have tought to tell my users to look there!)
If I remember correctly, there is some information missing in the
Using mkfile -n doesn't alllocate all the blocks in the the file, it is
supposed to grow as needed up to the defined limit. This is all
What is not mentioned is that it only works when you nfs-mount the file
from a server. I believe it does not work when you are using it on the
systems own disk.
> I tried to rewrite the label on sd1, but format gives me
> something like:
> sd1: Label not found
> sd1: Disk Quantum P105SSS, size unknown
> sd1: off-line
> and then only gave me the choice to select sd0.
> Sun will come tomorrow with a new drive, but I would
> really like to restore the label on the old one in the hope
> that the data on the drive is still good.
Sun came and did like a lot of people told me, gave it a light
tap. It didn't solve the problem, so a new drive was installed
and tested. Knowing that the problem was with the drive, a more
forcefull tap/shake/twist was given to it and now it works just fine.
From: tessi!joey@nosun.West.Sun.COM (Joe Pruett)
sounds like your disk needs the smack on the side fix. this is a known
problem with the quantam drives, that sometimes the arm will get stuck and
a gentle whack (someone posted a diagram showing the best place to hit it)
will unstick it.
From: firstname.lastname@example.org (Geert Jan de Groot)
Usually, the drive isn't bad. Take it out (using ESD precautions), and
tap, shake it a bit. Sometimes, the heads become stuck and the motor
cannot get the spindle running. Twisting it a bit may fix this.
It has been said in sun-managers before.
Also thanks to
Jeff Nieusma <nieusma@eclipse.Colorado.EDU>
From: email@example.com (Vikas Aggarwal)
From: firstname.lastname@example.org (Claude Scarpelli)
From: email@example.com (Thomas Narten)
From: Charles <firstname.lastname@example.org>
From: email@example.com (Mike Raffety)
From: firstname.lastname@example.org (Stewart Castaldi)
From: dagobah!mis@uunet.UU.NET (Mark Seiden)
From: Tony.Santos@Canada.Sun.COM (Tony Santos - Sys/Net Administrator Sun Toronto)
Alain Brossard, Ecole Polytechnique Federale de Lausanne,
SIC/SII, EL-Ecublens, CH-1015 Lausanne, Suisse
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:05:59 CDT