<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type"> </head> <body bgcolor="#ffffff" text="#000000"> We have ISCSI based storage and are using GFS during a demo period, but I don't think we can afford it after the demo period is over. The fencing is another issue for us. I need to learn more about that. I'm not sure I'm fully grasping all I need to there. I'm still trying to get nfs file systems mounted on a remote host. :) Thanks, Randy Lon Hohberger wrote: <blockquote cite="mid:20070822153807.GF4669@redhat.com" type="cite"> <pre wrap="">On Wed, Aug 22, 2007 at 11:37:11AM -0400, Lon Hohberger wrote: </pre> <blockquote type="cite"> <pre wrap="">On Mon, Aug 20, 2007 at 04:39:05PM -0400, Randy Brown wrote: </pre> <blockquote type="cite"> <pre wrap="">Right. That's the way I understood it to be. Using ext3 would require us to have to umount and remount the file systems to the each host after the failure, though, correct? In other words, would require administrator interaction. GFS would do this automatically without impacting the users. </pre> </blockquote> <pre wrap="">One would hope that the failover process handles mounting and unmounting. You do, however, need shared storage and suitable fencing mechanisms in place. </pre> </blockquote> <pre wrap=""> ... that is, with ext3. With GFS, you just mount it on both nodes and the recovery occurs in the background. Again, shared storage + fencing are required. </pre> </blockquote> </body> </html>