<div dir="ltr">Are you sure all nodes in your network and clients are working at Gigabit speeds ? I've had problems with NICs negotiating 100Mbps due to faulty equipment, cables etc. Do some basic tests with ethtool, iperf etc. Replica + 100Mbps would explain your problem. </div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Aug 20, 2015 at 9:14 AM, shacky <span dir="ltr"><<a href="mailto:shacky83@gmail.com" target="_blank">shacky83@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">2015-08-20 17:27 GMT+02:00 Prasun Gera <<a href="mailto:prasun.gera@gmail.com">prasun.gera@gmail.com</a>>:<br>
> Your gluster node cannot be an nfs client either. If you are using autofs or<br>
> even fstab for mounting other NFS shares, it will stop gluster nfs from<br>
> starting due to the NLM conflict.<br>
<br>
</span>I don't use autofs nor fstab for mounting NFS share, but Proxmox is<br>
mounting shares itself.<br>
Anyway, I don't necessary want to use NFS on my Gluster volume, it was<br>
only a try to solve the poor performance problem.<br>
Do you have hint about this? 5 MB/sec of throughput on the GlusterFS<br>
volume is very very bad!<br>
</blockquote></div><br></div>