Subject: Re: lfs panic on NetBSD-4.0_BETA2
To: None <tech-kern@netbsd.org>
From: Blair Sadewitz <blair.sadewitz@gmail.com>
List: tech-kern
Date: 12/19/2006 11:58:10
Oh?  Simply by using it, or is there a specific operation?  This is
one of the biggest problems I have with LFS. I'm not sure, but I think
this may also happen with null mounts if one layer is on an ffs
filesystem. It's hard to tell, because I very well could've been doing
something else to trigger it while I had a null/union mount; it's hard
to tease out causality in this case.

Apart from this, I have very specific problems that happen:

1) A deadlock involving biowait
2) A deadlock involving vnlock, which I'm fairly sure is related to layers.
3) Certain metadata update operations such as chown/chmod seem to take
inordinate amounts of CPU  and real time.  It takes me 10+ minutes at
95%+ CPU to chmod -R g+w pkgsrc.

However, I still use LFS for some things because its write performance
can be incredible.  The day when LFS has stability approaching FFS
will be a bright day for NetBSD indeed.  From there, improving the
smooth sync so it doesn't write out huge chunks of data every, for
instance, vfs.sync.*delay seconds.

-- 
Support WFMU-FM: free-form radio for the masses!

<http://www.wfmu.org/>
91.1 FM Jersey City, NJ
90.1 FM Mt. Hope, NY

"The Reggae Schoolroom":
<http://www.wfmu.org/playlists/RS/>