Can I put on my paranoid hat (tinfoil) and ask if they
refuse to work with something with a mismatch? Do they
refuse to connect to dcache.gridpp.rl.ac.uk if it presents
somethingelse.gridpp.rl.ac.uk?
This (gedanken) experiment assumes your CA manager allows
you to install a certificate/key - temporarily - on a host
where it does not belong. I am sure for the Greater Benefit of
Science the CA can condone this ...?
Cheers,
--jens
> -----Original Message-----
> From: GRIDPP2: Deployment and support of SRM and local storage
> management [mailto:[log in to unmask]]On Behalf Of Steve
> Traylen
> Sent: 28 February 2005 16:22
> To: [log in to unmask]
> Subject: GSI and hostname-whatever
>
>
> Hi
>
> We added
>
> dcache-tape.gridpp.rl.ac.uk -> dcache.gridpp.rl.ac.uk
>
> as a CNAME today.
>
> Now things like
>
> globus-url-copy file:/etc/group \
>
> gsiftp://dcache-tape.gridpp.rl.ac.uk/pnfs/gridpp.rl.ac.uk/data
> /dteam/swansea
>
> and
>
> srmcp file:////etc/group \
>
> srm://dcache-tape.gridpp.rl.ac.uk:8443/pnfs/gridpp.rl.ac.uk/da
> ta/dteam/newcastle
>
>
> both work despite the fact that host is presenting a CN of
> dcache.gridpp.rl.ac.uk.
>
> It is good they both work since these clients include the major
> implementations of GSI (globus C and globus COG).
>
> Whilst this is slightly disturbing it is extremely useful. Things
> like grid-support.ac.uk are particularly interesting.
>
> We will now look into setting a second fictitious SRM
> endpoint on this
> CNAME that will write to pools that are backed by tape. The different
> object endpoints will have different SARoots in the info system which
> is what really defines the particular pool and so if to tape
> or not to
> tape.
>
> dcache.gridpp.rl.ac.uk:/pnfs/gridpp.rl.ac.uk/data/dteam
>
> for disk and
>
> dcache-tape.gridpp.rl.ac.uk:/pnfs/gridpp.rl.ac.uk/data/tape/dteam
>
> for tape backed disk or something like that.
>
> There will be nothing stopping people mixing these two up
> however I think
> we will decide fairly early on if these will ever be split
> up. My instant
> reaction is that we avoid some worms from a can if we don't
> split them up
> ever.
>
> Steve
>
>
> --
> Steve Traylen
> [log in to unmask]
> http://www.gridpp.ac.uk/
>
|