Thanks, but my 8GB nodes are listed first. The server_priv/nodes file format is:<br><br>node25 np=4<br>node26 np=4<br>...<br>node32 np=4<br>node33 np=8<br>...<br>node40 np=8<br><br>With the node33 and higher having 24GB.<br>
<br>-Shawn<br><br><div class="gmail_quote">On Mon, Jan 5, 2009 at 3:54 PM, Steve Young <span dir="ltr"><<a href="mailto:chemadm@hamilton.edu">chemadm@hamilton.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div style="">Hi,<div><span style="white-space: pre;"> </span>I'm not sure about NODEALLOCATIONPOLICY since I have mine set to PRIORITY. But I did wonder how you had your nodes listed in the torque server_priv/nodes file. I'd make sure that the 8Gb nodes were listed first before the 24Gb nodes. Just a thought. Hope it helps. </div>
<div><br></div><div>-Steve</div><div><br><div><div><div></div><div class="Wj3C7c"><div>On Jan 5, 2009, at 3:42 PM, Shawn Poindexter wrote:</div><br></div></div><blockquote type="cite"><div><div></div><div class="Wj3C7c">Our cluster has 8 nodes with 8GB of RAM and 8 nodes with 24GB of RAM. I want to set it up so that the jobs will default to the 8GB nodes if the resources are adequate. However, it is not working that way. Even if the queue is completely empty the jobs fill up the 24GB nodes before the 8GB nodes. As I understand it, NODEALLOCATIONPOLICY MINRESOUCE should do what I want and fill the 8GB nodes first with the small memory jobs.<br>
<br>Here is my maui.cfg<br><br># maui.cfg 3.2.6p13<br><br>SERVERHOST <a href="http://arjuna.mps.ohio-state.edu" target="_blank">arjuna.mps.ohio-state.edu</a><br># primary admin must be first in list<br>ADMIN1 root<br>
<br># Resource Manager Definition<br><br>RMCFG[**host name removed***] TYPE=PBS<br><br># Allocation Manager Definition<br><br>AMCFG[bank] TYPE=NONE<br><br># full parameter docs at <a href="http://clusterresources.com/mauidocs/a.fparameters.html" target="_blank">http://clusterresources.com/mauidocs/a.fparameters.html</a><br>
# use the 'schedctl -l' command to display current configuration<br><br>RMPOLLINTERVAL 00:00:30<br><br>SERVERPORT 42559<br>SERVERMODE NORMAL<br><br># Admin: <a href="http://clusterresources.com/mauidocs/a.esecurity.html" target="_blank">http://clusterresources.com/mauidocs/a.esecurity.html</a><br>
<br>LOGFILE maui.log<br>LOGFILEMAXSIZE 10000000<br>LOGLEVEL 3<br><br># Job Priority: <a href="http://clusterresources.com/mauidocs/5.1jobprioritization.html" target="_blank">http://clusterresources.com/mauidocs/5.1jobprioritization.html</a><br>
<br>QUEUETIMEWEIGHT 1 <br><br># Backfill: <a href="http://clusterresources.com/mauidocs/8.2backfill.html" target="_blank">http://clusterresources.com/mauidocs/8.2backfill.html</a><br><br>BACKFILLPOLICY FIRSTFIT<br>
RESERVATIONPOLICY CURRENTHIGHEST<br> <br># Node Allocation: <a href="http://clusterresources.com/mauidocs/5.2nodeallocation.html" target="_blank">http://clusterresources.com/mauidocs/5.2nodeallocation.html</a><br><br>
NODEALLOCATIONPOLICY MINRESOURCE<br><br><br></div></div> _______________________________________________<br>mauiusers mailing list<br><a href="mailto:mauiusers@supercluster.org" target="_blank">mauiusers@supercluster.org</a><br>
<a href="http://www.supercluster.org/mailman/listinfo/mauiusers" target="_blank">http://www.supercluster.org/mailman/listinfo/mauiusers</a><br></blockquote></div><br></div></div></blockquote></div><br>