Chris G Demetriou <Chris_G_Demetriou@ux2.sp.cs.cmu.edu> writes:

> Uh, actually, it's pretty easy:
> 
> 	change the per-fs '-p' option so that it does the single (specified)
> 	file system.
> 
> 	dispatch forks/execs of per-fs checkers as necessary
> 
> 	if they return specific error codes, then (possibliy) have the
> 	master not issue further fscks, and exit with an appropriate
> 	error code.
> 
> Yeah, sure, it involves modifying a _lot_ of the existing fsck control
> logic, but the resulting solution would end up looking pretty "good"
> if done right.

I like the AFS fsck...  At least one part of it.  :)

It lists which filesystems failed at the end of the run, in case you missed
it in scroll.  fscking say, 24 disks can do that...

It also had an option for specifying the "maximum parallelism" since once
again, beating on 24 drives isn't a nice thing to do all at once usually.

--Michael