[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: radmind anomaly



Willie Northway <willn@xxxxxxxxx> writes:
> The machine had problems - so I brought it up from CD, and ran fsck on 
> each of the partitions from the CD... ahh! The version of fsck *on* the 
> hard drive is v 1.35, but that might not be the same on the CD.
> 
> If it's older than 1.35, then perhaps we should cut a new one.
> 
> Thanks for the tip Marcus.

Yup, it's older, and that needs to be done.  Rumour has it that
Albert is looking into doing this.  There's a list of other things
that could be done with the bootcd -- my personal list of things to
consider includes:
	updated fsck
	more recent radmind build
	reiserfs tools
	no radmind client certs -- script to fetch certs
		somehow instead (scp?).
	better build process using special cert
		(works experimentally...)
	2.4.28 generic kernel
		(built, not tested)
	probably not lvm stuff -- I take it this was considered
		unhealthy on the imap machines...
	lmsensor stuff -- see if this works on any of our hardware
	include some form of low-level disk exerciser
I believe albert is doing at least the first two or three,
very probably not the others.  I heard rumours he's using make,
which is perfectly reasonable.  I'll probably try to catch up with
him to talk about some of the rest of this stuff.

I'm nevertheless highly curious what version of reiserfs clunis has been
trying, what his luck has been, and if he's built any transcripts
of the tools.

Which of us remembers the radmind CA password?  I can't remember
if this was Patrick, sweda, or somebody else...

An interesting question to ask here might be where do people
want to see .iso images end up in the long run?  I've been stuffing
them into AFS here:
	/afs/umich.edu/group/itd/build/mdw/umce/bootcd
but other equally plausible areas might be:
	/afs/umich.edu/group/itd/umce
	somewhere on equilibrium
	somewhere on sixthday
	it's passed around in secret and nobody knows where
		they really originate

					-Marcus