Hi Moinak,
Thanks for taking a look!
On Mon, 2007-11-19 at 16:26 +0530, Moinak Ghosh wrote:
> No. In fact the microroot is too big at 146MB. The LiveKit miniroot
> in comparison is about 67MB. One of the reasons is that the microroot
> contains both 32 and 64bit binaries even though the 64bit binaries
> are not used. These should ideally be in separate microroot archives.
Aha.
> > Given though that the system was pretty responsive under nv_70, I'm
> > trying to work out what's going on - I'm thinking next, I might try
> > extracting to UFS to see if ZFS is the problem, but if anyone has better
> > ideas ?
> >
>
> I feel it is a ZFS thingy combined with the small 110MB available RAM
> after the microroot grabs it's share. I think doing the install to UFS will
> be a good test.
Okay, I'll give it a go whenever I get time. (the system's useless at
the moment, so this is in my best interests :-)
> In addition though normal lofi sucks up page cache like a sponge
> Alok's build of clofi has a couple of additional bug-fixes that resolve this
> issue
> and map pages with appropriate segmap flags. So clofi does not look like a
> problem. Try touching a file in /kernel and running bootadm update-archive
> manually on an usual SNV image (not Indiana) and you will see what I mean.
Will try to take a look, thanks!
cheers,
tim
--
Tim Foster, Sun Microsystems Inc, Solaris Engineering Ops
http://blogs.sun.com/timf
