Re: RFC: New required package: libblkid1
Martijn van Oosterhout <kleptog@svana.org> writes:
> On Wed, Mar 26, 2003 at 02:31:35PM -0500, Matt Zimmerman wrote:
> > On Wed, Mar 26, 2003 at 08:14:54PM +0100, Goswin Brederlow wrote:
> > > That means it needs to change on nearly daily basis here.
> >
> > "needs"? It's a cache, and if you don't mind it scanning your devices when
> > you do a uuid-based mount, then you don't need it and if you insist on
> > having a read-only root partition, everything still works for you. Where's
> > the problem?
>
> I'm just trying to understand something. If you have 6000 partitions
> mounting by UUID without a cache, does this mean that mount would scan the
> partitions over 10 million times to complete it's job? Is this the problem
> being solved? Obviously mount -a could optimise but which 600 seperate mount
> commands I imagine this would be an issue.
Given that the major/minor numbers are pretty limited even having 6000
partitions would be pretty hard.
I think thats realy just said as an exageratede example to show the
point.
MfG
Goswin
PS: How often do you think a system with 6000 partitions would get rebootet anyway? Who cares if it takes slightly longer? The UUIDs should all remain in the block caches so it shouldn't take that long to scan them all.
Reply to: