<p>31/07/2019<br />
Dear Users,</p>
<p>in relation to the ongoing reconfiguration/upgrade of Galileo<br /><a href="http://www.hpc.cineca.it/center_news/galileo-reconfiguration-and-upgrade-update-gpu-production">http://www.hpc.cineca.it/center_news/galileo-reconfiguration-and-upgrade...</a><br />
cluster, we scheduled a *two-day maintenance* next August, 6th-7th. During<br />
the stop, the login nodes and the filesystems will not be available in order<br />
to perform the final data synchronization to the new storage system. </p>
<p>At the end of the maintenance users will connect to the new, upgraded<br />
environment. The number of nodes will gradually increase, in the following<br />
two weeks, to arrive to over one thousand of Broadwell nodes interconnected<br />
with an Intel Omni-Path network. The sheduler, accounting, and module<br />
environments replicate the current ones, hence no changes in submission<br />
scripts are required to continue your production. </p>
<p>We remind you that only your $WORK and $HOME data will be copied to the new<br />
storage. The current $CINECA_SCRATCH filesystem will be mounted (in read-only<br />
mode) on the login nodes for three months, to allow the transfer of the data<br />
you need to maintain to the new, initially empty scratch areas.</p>
<p>During the first week on the new system the RCM service will not be<br />
available. Please also note that, due to the tuning of the storage<br />
configuration, in the first weeks of the new production some sporadic, short<br />
unavailabilities of the $WORK filesystem might occur,</p>
<p>Best regards</p>
<p>HPC User Support @ CINECA</p>