SARA-NIKHEF tier1 report:

We have upgraded to gLite3.0.0. We have encountered several bugs in yaim
that we have submitted to GGUS. We still need to test the gLite-CE and
the FTS. The lcg-CE and torque server are running on one node and the
Glite-CE on an separate node. We have splitted up de FTS over two nodes.
The thing I want to mention here is the BLAHParserPBS server that is
supposed to run on the torque server. That is a server that runs as root
and does not contain any form of authentication. It exposes the pbs logs
to the outside world. To reduce the security risk we have wrapped it in
a iptables rule so that only the gliteCE can access it. In addition, a
proper init.d script was not provided so we had to write our own that
would properly daemonise it. 

Further we have been testing our storage infrastructure and investigated
the bottlenecks that we encountered during the disk2tape tests. We found
out the the paths on our SAN that were used for reading from disk and
writing to tape caused two raid controllers in the same enclosure to
continuously copy data to each others cache leading to performance
degradation. We have corrected this. We have further improved the
performance by making sure that data copied to our disk servers in our
SAN immediately is immediately written to disk. All in all this has lead
to the fact even with a for these tests stripped infrastructure we could
write to tape with about 90 MB/s at times. For the hardware that we have
available at the moment we should be able to do about twice this rate in
the near future.