Gordan On Wed, 30 Apr 2008, Mickey Mazarick wrote:
We have a work around for this so that clients don't see a "blocked" situation. we have a 6 node cluster and whenever a server node comes up it lists all the files on it's "slice" of the gluster mount and then does a find on those files on the main mounted volume. In other words each server is also a client, and does the heal process right after starting up.
