<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Mar 1, 2016 at 3:47 PM, Fedele Stabile <span dir="ltr"><<a href="mailto:fedele.stabile@fis.unical.it" target="_blank">fedele.stabile@fis.unical.it</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hello to all,<br>
my question of the day is: can I use GlusterFS on a 32-node cluster<br>
HPC?<br>
Let me explain:<br>
- each node in the cluster has one extra disk that is a brick of a<br>
glusterFS cluster-wide mounted<br>
- interconnection network is 40GB/s Infiniband<br>
- users run MPI jobs<br>
<br>
Everything works perfectly except that a user complains that his MPI<br>
job after gluster implementation is slower than 50%<br>
<br>
I made a linpack test that gives optimal results if runs on 16 nodes,<br>
but if runs on 32 nodes<br>
it happens that one of the gluster nodes disconnects and the linpack<br>
don't finish.<br>
<br>
Can anyone help me?<br></blockquote><div><br></div><div>Hi Fedele,</div><div><br></div><div>We are interested to solve this, we have received similar description from other users too.</div><div>Note that I am talking only about number of clients and not about MPI. MPI workload should run fine on Gluster.</div><div><br></div><div>Some questions:</div><div>1. What version of Gluster?</div><div>2. What OS?</div><div>3. Is the volume set to RDMA mode or are you using IPoIB?</div><div>4. Do all the clients work on the same file from the volume?</div><div><br></div><div>Thanks,</div><div>Raghavendra Talur</div><div><br></div><div><br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Fedele<br>
<br>
_______________________________________________<br>
Gluster-users mailing list<br>
<a href="mailto:Gluster-users@gluster.org">Gluster-users@gluster.org</a><br>
<a href="http://www.gluster.org/mailman/listinfo/gluster-users" rel="noreferrer" target="_blank">http://www.gluster.org/mailman/listinfo/gluster-users</a><br>
</blockquote></div><br></div></div>