Subject: Re: experiments after recent LFS changes
To: Konrad Schroder <firstname.lastname@example.org>
From: Daniel Carosone <email@example.com>
Date: 05/02/2005 14:10:49
Content-Type: text/plain; charset=us-ascii
On Sat, Apr 23, 2005 at 01:53:11PM -0700, Konrad Schroder wrote:
> On Mon, 18 Apr 2005, Daniel Carosone wrote:
> >My usual test (rsync's of htdocs,pkgsrc,src,xsrc into a vnd-hosted
> >LFS) doesn't get very far now; the writing rsync gets stuck on WCHAN
> >"buffers" and the process hangs (until i ^C it).
> Are you still seeing this problem? My tests don't match yours exactly (i=
> particular I've been testing with a disk partition, not a vnd), but I hop=
> that the page-accounting change I committed on the 19th will have fixed=
Great news! I can no longer reproduce this problem; moreover I can't
yet reproduce *any* other problem with LFS.
I currently have running:
- 2 rsyncs copying and then deleting pkgsrc in a loop
- 2 pax's making hard-link trees of the above pkgsrc copies
- 2 cleaners (the mount one, and a manual -c -d)
- 2 postmarks running 50million transactions each
on a nearly-full filesystem (that has previously been filled several
Nothing appears to be breaking! I am.. thrilled. (It's getting pretty
slow for some of the writing processes at points, especially with both
cleaners going the disk is doing a *lot* of reading, but it is Still
This is clearly only preliminary testing, and hasn't been running all
that long - but already much longer than previous such tests.=20
This lfs is in a vnd over a cgd ffs, fwiw.
It doesn't yet mean LFS is ready for real use; but for the first time
it seems worth trying to really find out. It indicates to me that more
rigorous and prolonged testing is now clearly worthwhile; I shall be
setting up a system to leave thrashing full-time like this, and
hopefully others can do likewise. =20
My fulltime test machine will be lfs over raidframe R5 on ide (no vnd)
on a multiprocessor.
> Take care,
> Konrad Schroder
Thank you so much for all your LFS work, especially recently. It
seems to have made a very big difference.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (NetBSD)
-----END PGP SIGNATURE-----