On 08/19/2011 12:55 AM, solarflow99 wrote: > i'm going through the guide too and I just wonder why you would even > need fencing if you are not using something like GFS? Imagine a network > with many app servers running on their own storage, they would probably > just link to a database to read/write any data.
The cluster communication and membership layer is separate from the resource management layer. The membership layer is concerned with having nodes in a known state. So, when a node "disappears" (network failure, hard failure, kernel panic, hung/blocked), the node falls into an unknown state. The cluster blocks at will remain blocked until the lost node is put into a known state. This is done using fencing. By allowing the cluster to reliably reach out and force-off a node, the node then is in a known state and recovery can begin. All of this happens well below the resource manager, and is not dependant on whether something like GFS2 is in use or not. -- Digimer E-Mail: [email protected] Freenode handle: digimer Papers and Projects: http://alteeve.com Node Assassin: http://nodeassassin.org "At what point did we forget that the Space Shuttle was, essentially, a program that strapped human beings to an explosion and tried to stab through the sky with fire and math?" _______________________________________________ rhelv6-list mailing list [email protected] https://www.redhat.com/mailman/listinfo/rhelv6-list
