[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: diskless package issues...



Junichi Uekawa (Junichi Uekawa mikilab) wrote:

On Mon, 01 Jul 2002 19:32:47 -0400
Adam C Powell IV <hazelsct@mit.edu> wrote:

Hi,
I have three questions. First, when I chroot to /var/lib/diskless/default/root and install new packages, /etc and /var updates don't propagate to the host /etc and /var directories. Are there plans to make this work -- for example, to have a "diskless-update" script to update all of those dirs?
I think that would be very difficult to implement, considering that the differences on each individual clients would be pretty difficult to track, to say the least.

Hmm...  How about ending diskless-newimage with something like:

touch $diskless-rootdir/lastupdate

(I don't know what the real diskless-rootdir variable should be, nor diskless-hostdirs below.) Then diskless-update could have something like:

for newconffile in \
 `find $diskless-root/etc -newer $diskless-rootdir/lastupdate -print`; \
do \
 for hostdir in $diskless-hostdirs; do \
   cp -i $newconffile `echo $newconffile | sed s/$diskless-root/$hostdir/`;
 done; \
done

Unfortunately, this wouldn't work for /var because it would copy things like /var/run and /var/spool, which is bad.

Or one could look in $diskless-rootdir/var/lib/dpkg/info for new .conffile files and use those to update the directories... Is there some kind of hook mechanism for installing conffiles in dpkg?

Second, woody seems to have a new way of installing base using packages in basedebs.tar instead of base.tgz. Are there plans to update diskless accordingly, e.g. using dpkg --install --root?
diskless-createbasetgz exists for creating  base.tgz.
It should be possible to run debootstrap on each run to diskless-newhost.

I see, that's cool. (Do you mean diskless-newimage instead of diskless-newhost?)

However, requiring diskless-createbasetgz to be called once,
and using the resultant base.tgz is faster than requiring debootstrap to be ran on each time.

That makes sense.

Third, is there a way to make the console work so users can log in there? I always get something like "id 1 respawning too fast, disabling for 5 minutes".
Init is in /sbin/init, AFAIK, and it is a shell script installed by diskless-image-{simple,secure} package.
You may want to read it to see what might be causing it...
and yes, it should probably be respawning too fast by the look of it.

I see. I guess it wasn't meant for console use. :-) Maybe the console entries in inittab should be commented?

One could probably switch to runlevel 5 and run a *dm just fine for a workstation cluster...

Well, this isn't my intended application, so I'll let somebody else experiment with that.

I haven't had much time to work with this package recently, and help would be appreciated.

Okay, we'll see what we can come up with.

Thanks again for a really nice package!
--

-Adam P.

GPG fingerprint: D54D 1AEE B11C CE9B A02B  C5DD 526F 01E8 564E E4B6

Welcome to the best software in the world today cafe! <http://lyre.mit.edu/%7Epowell/The_Best_Stuff_In_The_World_Today_Cafe.ogg>




--
To UNSUBSCRIBE, email to debian-beowulf-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org



Reply to: