Description of problem: Need to add a short section on persistent clusters along the lines of this: In the unlikely event that every member of persistent cluster crashes and none of them have a clean store (e.g. if a power failure shut down all members simultaneously) you can recover as follows: 1. Choose one member and force its store to a clean state as follows: qpid-cluster-store --mark-clean <path to broker's data directory> If the brokers all fail so close together that there is no clean store, then there's no easy way to determine which store is the "best". 2. Start the member with the clean store first, wait for it to log a "ready" message. 3. Then start the remaining members, they will download state from the first member.
*** This bug has been marked as a duplicate of bug 598653 ***