Seth Mos wrote:
>
> At 14:01 20-8-2001 -0500, yocum@xxxxxxxx wrote:
> >I'd test on a non-RAID system if I had one (how often do you hear that?
> >;-). I suppose I could try it on my desktop. I'll let you know how that
> >goes, too. We're moving *lots* of data around (a few hundred GB/week at
> >least). I'm trying to get the data analysts to use rcp instead of NFS, but
> >that'll take some time.
>
> Slightly OT.
>
> how about scp and then using publickey authentication. I just managed to
> set something like that up for the kt.zork.net mirror.
>
> You won't even need to enter passwords then although it is highly risky
> when you use this on laptops.
scp is OK for small amounts of data, but not so good for hundreds of GB,
since it will encrypt the data before transferring it, which takes a bit
more CPU time.
Yes, I know you can recompile ssh to encrypt passwords and keys and send
data in the clear, but I just haven't gotten around to it. Besides, a) rcp
is in all the scripts the developers use to move stuff around, and b) the
entire Lab will be kerberized by Dec 31, so we'll be using kerberized rcp
(no .rhosts or host.equiv files laying around) and no passwords, either.
Back on topic: at least one data analyst moved about 40GB of data around
last night via NFS from OSF1 to Linux using NFVv3 and there were no inode
corruptions. The kernel being used *is* 2.4.8-xfs which was compiled with
gcc-2.96-85 (because I didn't get around to recompiling with kgcc yet,
mkp).
Cheers,
Dan
--
Dan Yocum
Sloan Digital Sky Survey, Fermilab 630.840.6509
yocum@xxxxxxxx, http://www.sdss.org
SDSS. Mapping the Universe.
|