Hi Brent,<br> Thanks for the detailed obseration regarding NFS re-export. <br><br><div class="gmail_quote">On Wed, May 13, 2009 at 8:59 AM, Brent A Nelson <span dir="ltr"><<a href="mailto:brent@phys.ufl.edu">brent@phys.ufl.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;"><br>
Note that this is the first time I've tried using cluster/ha; </blockquote><div> </div><div>Let us check what is the effect of cluster/ha in this behavior. Seems like its causing issues, as thats the only untested part after this feature got introduced. <br>
<br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Note that df of the GlusterFS filesystem looks just fine, so it seems odd that anything would be confused about filesystem size</blockquote></div><br>Ya, its true, when you do 'df' the call is initiated from the top most layer (fuse in most of the case), and all inode initialization etc happens in the right way, where as in the earlier case (of calculating free disk space), the call gets initiated by distribute translator itself, hence there may be issues. Anyways, we will get back with our findings.<br>
<br>Regards,<br clear="all"><br>-- <br>Amar Tumballi<br><br>