[monolith-users] News digest, March 2004

[Lennart Karlsson]

Dear Monolith Users,

This is a digest of news items related to the usage of Monolith.
It is sent to all users on Monolith via the e-mail list
<monolith-users at nsc.liu.se>. Instructions on how to subscribe and
unsubscribe is listed at the end of this e-mail.


System Stability
=====================

The 29th of February Monolith was back in full service in our
new computer room. The move started Monday morning, six days
earlier. Most of the time was spent on removing about half of
the 600 SCALI cables, to be able to move the racks one by one,
and later to reconnect the cables and making sure that the
reconnection job was done good enough.

We have had some program crashes since then, and we have been
fine adjusting a lot of the cabling. But everything inlcuded,
we have experienced much fewer problems than we hade anticipated
and are happy with our move.

At the moment nearly everyting looks nice and nearly as good as
before the move. We hope to get the quality fully back in the
old shape within a couple of weeks. Please report any problems you
experience to support at nsc.liu.se, as usual.


Planned System Stops
=====================

On Tuesday, March 23 at 11 a.m. we will start an upgrade of
our SCALI software, including the ScaMPI libraries, to get
better performance, especially in the initialization phase of
ScaMPI jobs running on many nodes. We plan to have Monolith back
in full operations at latest at 11 a.m. on Thursday, Match 25, but
more probably already early Wednesday afternoon.


Walltime limit of jobs
=====================

We have now a wallclock time limit of 72 hours, i.e. 3 days, on
Monolith batch jobs.

This low limit has made it possible for us to plan system maintenance
stops at short notice without deleting user jobs prematurely.

We have been asked to increase this wallclock time limit to 144 hours,
i.e. six days. Because the system stability now is quite good,
we plan to implement this change at the end of next week, in case the
SCALI software upgrade (see above) will not give us unanticipated, new
problems.


As always, please mail your questions and comments to
<support at nsc.liu.se>.

Best regards,
-- Lennart Karlsson <support at nsc.liu.se>
   National Supercomputer Centre, Linkoping University
   http://www.nsc.liu.se