RFC - bootscript error reporting

IvanK. ivan at chepati.org
Thu Jan 29 10:28:38 PST 2004


On Wednesday 28 January 2004 02:08 pm, Nathan Coulson wrote:
<snip>

> I like the idea, but it sounds like a big project...  [I'll look at it
> again later], and a bit out of our boundries...  [and yet...]
>
> I do admit, I never did play with initrd's at all, but what I looked at,
> was almost a image that was mounted in place of root?

Exactly.  An alternative would be to have an extra partition on the first 
hardrive that's only, say 20MB with enough stuff to provide a comfortable 
environment for fixing, either manually or automated-like, the real system.. 
The problem with that is that many people would begrudge the space "wasted".  
Also this will not be an easy solution to implement on a already existing lfs 
system.  An initrd, on the other hand, is just a drop-in solution.

Yes, it is a big project and it best be done in stages.  Firstly, we can 
address Jeremy's concern about pausing the bootscripts.  Then we can worry 
about fancy stuff.

I have guests until Saturday, but once I see them off to the airport, I'll sit 
down and experiment with some of the ideas already in circulation.  Exciting 
stuff.

Also, should we starting using extX labeling so that we can say /, /var, and /
usr are critical, if they fail, trigger the auto-healing procedure, but if /
usr/local, /home, etc fail, continue.  This means we have to modify mountfs 
and not use -a with mount, but rather mount the critical systems first, then 
the rest.  I was never big on parallel mounting anyway.  Actually extX labels 
are not necessary for that, but it's nice to have them.

Any reliable method of corrupting an ext3 fs for testing purposes? :-)

IvanK.




More information about the lfs-dev mailing list