Hi Gregory,
On Mon, Nov 22, 2004 at 02:06:57PM -0500, Gregory J. Sharp wrote:
> Today we successfully tested an interface between Suez and Xrootd. We
> are able to read files into Suez from an xrootd server. We can cache
> files from the HSM to the xrootd servers, so the rest should all be
> downhill ;-)
Excellent news! As Andy said, we are particularly interested in feedback
on operational issues that come up as you try to run the system day in
and day out....
There are a couple of things that do come out when one is trying to operate
the system that we really need to document someplace:
o You should set things up such that the xrootd (and olbd) servers are
automatically restarted. While they shouldn't crash very often, nothing
is perfect. Hence if they do crash in some rare situation, restarting
them automaticaly (within 15 minutes, say) will at least allow clients
to continue with crashing due the fault-tolerance built into the client.
[If you save the core file from any crash, too, that is welcome.]
o Depending on how many clients you have, you might want to look at
increasing the number of file descriptors on the servers. While
xrootd is fairly economical with this (compared to the AMS, for
example) the linux default of 1024 can be too small if you have more
than 250 clients. (Given the defaults Andy put into the xrootd.)
Obviously this is most important on the redirector machine.
Once you get things stable I'll probably ask you for a description of
your system (hardward, etc.) as well as your config files (if you don't mind)
for our "examples" documentation....
Pete
-------------------------------------------------------------------------
Peter Elmer E-mail: [log in to unmask] Phone: +41 (22) 767-4644
Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland
-------------------------------------------------------------------------
