El Wednesday 03 September 2008 18:23:09 Petter Reinholdtsen escribió: > [Noel David Torres Taño] > > That is what I did at first :) but I tried to create an (elegant) > > solution capable of being used by any people and not only for me > > while the background problem is addressed by "the experienced > > hackers". It's the way I understand I can give to "the community" > > something for all what I've received ;) > > And know that it is much appreciated. But your proposed solution will > only work for some machines, and slow down a lot of others. I wish > there was a way to figure out if some disk are going to show up soon, > but as far as I know, there is none. Perhaps a better way is to check > if the root device is available, and if not, wait a bit. This way, > the boot will be more robust while still being quick in the common > case. > > Something like > > while [ ! -f $rootdev ] ; do > echo "missing root dev $rootdev" > sleep1 > done > > in front of the fsck call. There should be a timeout or loop > counter to make sure it do not wait forever. > > Or, perhaps the boot should be event based and fsck only called when > the event from the kernel for the given device arrive. Then we get > another issue, which is to wait for the root file system to be mounted > before continuing. > > Happy hacking,
But that is for checkroot.sh , isn't it? I'm talking about non-root devices for checkfs.sh About the "your proposed solution will only work for some machines, and slow down a lot of others" part, I think that the last patch I send doesn't slow any machines at all excepting machines expressely configured to do so. About event based boot, I hope we will have it in the (near) future, but I'm trying to give a solution for today's systems. Thanks for your time :) Noel Torres er Envite
signature.asc
Description: This is a digitally signed message part.
