<html><head><meta http-equiv="Content-Type" content="text/html; charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class="">Hi,<div class=""><br class=""></div><div class="">Using NFS is discouraged for MPI-IO (and PnetCDF as well) applications, because</div><div class="">it requires a special setting in NFS lock daemons to make it work correctly</div><div class="">for MPI-IO.</div><div class="">See ROMIO user guide at <a href="ftp://ftp.mcs.anl.gov/pub/romio/users-guide/node12.html" class="">ftp://ftp.mcs.anl.gov/pub/romio/users-guide/node12.html</a></div><div class=""><br class=""></div><div class="">If the HPC cluster you are using has a parallel file system, such as Lustre or</div><div class="">OrangeFS, then you should use it.</div><div class=""><div class=""><br class=""><div class="">
Wei-keng
</div>
<div><br class=""><blockquote type="cite" class=""><div class="">On Apr 16, 2018, at 3:07 AM, o_Orz <<a href="mailto:x70062924@qq.com" class="">x70062924@qq.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class="">Hi:<br class=""> I runned WRF program on ARM HPC cluster. In the cluster, all nodes share working directory by NFS.<br class=""> While using netcdf, I found everything fine except that the performance was poor.<br class=""> So I used PnetCDF. On single node, it functioned well. But when it is used in cluster, the netcdf file it generated is wrong and unreadable. I cannot figure oit why. Is it that PnetCDF does not support multi-node application? Is it that I didn't use distribute file system? Or some other reason?<br class=""><br class="">Best Regards<br class="">Xie Bin</div></blockquote></div><br class=""></div></div></body></html>