<div dir="ltr">Hi everyone,<div><br></div><div>We have a heterogeneous cluster. Part of the nodes have two nvidia gpu cards.</div><div><br></div><div>slurm.conf looks like:</div><div><br></div><div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">NodeName=compute-0-[0-6] CPUs=32</blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">NodeName=compute-1-[0-5] CPUs=16 Gres=gpu:2 Weight=100</blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><br></blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">PartitionName=hipri DefaultTime=00-1 MaxTime=00-1 MaxNodes=1 PriorityTier=9 Nodes=compute-0-[0-6]</blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">PartitionName=higtx DefaultTime=00-1 MaxTime=00-1 MaxNodes=1 PriorityTier=9 Nodes=compute-1-[0-5]</blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">PartitionName=cpu   DefaultTime=01-0 MaxTime=15-0 MaxNodes=1 Nodes=compute-0-[0-6]</blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">PartitionName=gtx   DefaultTime=01-0 MaxTime=15-0 MaxNodes=1 Nodes=compute-1-[0-5]</blockquote><div><br></div></blockquote><div>gres.conf look like:</div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">NodeName=compute-1-[0-5] Name=gpu File=/dev/nvidia0</blockquote><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">NodeName=compute-1-[0-5] Name=gpu File=/dev/nvidia1</blockquote><div><br></div></blockquote><div><br></div><div>Preempting works for hipri partition, but not for higtx partition</div><div><br></div><div>By the way, if I suspend job in cpu partition, the pending jobs in cpu partition start to run. But if I suspend job in gtx partition, nothing happens.<br><div><div><br></div>-- <br><div dir="ltr" class="gmail_signature"><div dir="ltr"><div><span style="font-family:"times new roman",serif">Gong, Zheng （龚正）<br>Doctoral Candidate in Physical Chemistry<br>School of Chemistry and Chemical Engineering<br>Shanghai Jiao Tong University</span><span style="font-family:"times new roman",serif"><br></span></div><div><span style="font-family:"times new roman",serif"><a href="http://sun.sjtu.edu.cn/" target="_blank">http://sun.sjtu.edu.cn</a></span></div></div></div></div></div></div></div>