RFC: Move PSEUDO_LOCALSTATEDIR for filesystems outside the filesystem

RFC: Move PSEUDO_LOCALSTATEDIR for filesystems outside the filesystem

[Peter Seebach]

Got a bug related to this that probably only affects our build
environment, but it caused me to note:

When we're using export/dist as a user-mode NFS root (under pseudo),
the PSEUDO_LOCALSTATEDIR is export/dist/var/pseudo.

This seems a bit weird to me -- I don't think pseudo's db should be
part of the filesystem it's maintaining. And I'm wondering whether
there is a historical reason for this that I need to be aware of,
before I propose moving it to something like export/pseudo.

... And if not, I propose we move it to something like export/pseudo.

-s
-- 
Listen, get this.  Nobody with a good compiler needs to be justified.

Re: RFC: Move PSEUDO_LOCALSTATEDIR for filesystems outside the filesystem

[Mark Hatle]

On 8/14/12 8:41 PM, Peter Seebach wrote:
> Got a bug related to this that probably only affects our build
> environment, but it caused me to note:
>
> When we're using export/dist as a user-mode NFS root (under pseudo),
> the PSEUDO_LOCALSTATEDIR is export/dist/var/pseudo.
>
> This seems a bit weird to me -- I don't think pseudo's db should be
> part of the filesystem it's maintaining. And I'm wondering whether
> there is a historical reason for this that I need to be aware of,
> before I propose moving it to something like export/pseudo.
>
> ... And if not, I propose we move it to something like export/pseudo.

The NFS export scripts and such expect the pseudo DB to be available to them. 
The filesystem can be anywhere the user wants, so somehow we have to know where 
it was extracted, and the only answer we came up with at the time was to put the 
pseudo database into the filesystem directory.  It doesn't cause any harm to the 
target that is being booted, and it ensures that the NFS server setup scripts 
know where the both the root and pseudo DBs are located.

--Mark

> -s
>