--TA4f0niHM6tHt3xR
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

So, despite all the threats, I've gone ahead and kept using LFS on
my mp3 partition (because, really, who cares; I can just rip the CDs
again, being as I do own them all), and now I seem to be bit, but
not by anything I've seen described here before. I will, as of
earlier today (quite a while after I started running this kernel),
get an lfs_cleanerd stuck in physio thrashing the disk. This
doesn't seem to affect the disk's performance (I'm reading mp3s off
of it right now, and it's working just fine), but it doesn't strike
me as physically healthy for the platters.

top's output:

  PID USERNAME PRI NICE   SIZE   RES STATE      TIME   WCPU    CPU COMMAND
  792 root      -5    0  5368K 5060K physio     0:47 12.21% 12.21% lfs_clea=
nerd

And ps -lwwp 792:

UID PID PPID CPU PRI NI  VSZ  RSS WCHAN  STAT TT    TIME COMMAND
  0 792  131  11  -5  0 5368 3776 physio DL   ?? 0:52.00 /usr/libexec/lfs_c=
leanerd -b -n 4 /mp3=20

The disk (df -k):

/dev/wd0a   127195868 41428750  73047532    36%    /mp3

(Point is, there is plenty of free space for LFS to play with here.)

The kernel:

NetBSD grappa 1.5ZC NetBSD 1.5ZC (GRAPPA) #4: Fri Apr  5 17:58:15 EST 2002 =
    gr@grappa:/usr/obj/.i386/sys/arch/i386/compile/GRAPPA i386

(Sources are maybe a day older than that; I've really got little way
of double-checking at this point, as I've cvs updated my tree
several times since I built the snapshot I'm using.)

Killing this lfs_cleanerd just causes another one to be spawned from
the parent. Everything's fine for a while, then this thing starts
chewing up processor and disk again. I just killed both parent and
child and restarted lfs_cleanerd, and neither's gotten to spinning
out of control just yet.

Is this a known aberrant behavior I've just missed paying attention
to? Have I got myself a corrupt ifile? (How would I tell?)

Fwiw, I *did* just write a lot of new stuff to this disk *while*
reading from it over the past couple of days (and could easily
artificially replicate that situation), but this just showed up
during the day today (kind of hard to tell exactly when; I was at
work). And, in any case, I've done that frequently before without
problems.

Konrad, you out there? Anything I could do here to help debug?
(Upgrading to a more recent -current is quite possible; in fact, I
think I've got a 20020430 snapshot lying around I haven't started
using yet, and could easily build a newer one.)

--=20
gabriel rosenkoetter
gr@eclipsed.net

--TA4f0niHM6tHt3xR
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (NetBSD)

iD8DBQE8/Cp59ehacAz5CRoRAkT2AJ9vCQSGfWflC6iYC+AFXjHuj5tfAACfeePq
LUjnhMn+KG3xMJtZqyZPGMM=
=flRg
-----END PGP SIGNATURE-----

--TA4f0niHM6tHt3xR--