On Fri, 2005-11-04 at 19:43 -0800, Doug Cutting wrote: > Rod Taylor wrote: > > I tried running one datanode per machine connecting back to the same SAN > > but it seemed pretty clunky. A crash of any datanode would take down > > the entire system (no data replication since it's a common data-store in > > the end). Reducing it to a single datanode did not have this impact. > > Why use NDFS at all? Why not just mount the SAN on all hosts? You're > not using NDFS as a distributed file system, but rather as a centralized > file system.
I was unable to make the mapred branch work by using 'local' as the filesystem and having more than one tasktracker. Tasktrackers were unable to complete any work, although it was quite a while ago when I last tried (September). -- Rod Taylor <[EMAIL PROTECTED]>
