following is needed by a membership protocol? 1. A failure detector 2. A mechanism to disseminate information about joins, leaves, and failures of processes 3. Both a failure detector and a dissemination mechanism ANSWERS: 3
Membership: Failure Detectors at 04:56
Which of the following is preferable in a failure detector? 1. 100% completeness 2. 100% accuracy 3. Long time to detect failures 4. Large number of messages to detect failures ANSWERS: 1
Membership: Failure Detectors at 08:20
Which of the following heartbeating techniques ensures completeness no matter how many failures occur simultaneously, and no matter how large the group is? 1. Centralized heartbeating 2. Ring heartbeating 3. Ring heartbeating where every node heartbeats to 3 successors and 3 predecessors 4. All-to-all hearbeating ANSWERS: 4
Membership: Gossip-Style Membership at 05:34
In gossip-style failure detection, why should an entry not be deleted right after its detected as failed? 1. Because other processes may not have deleted that entry and it may be added back 2. Because that node may not have actually failed 3. Because one cannot delete membership entries ANSWERS: 1
Membership: Dissemination and Suspicion at 03:51
In a system with N processes using the SWIM failure detector, given enough network bandwidth, how long does it take for a failure, after it is detected at one process, to disseminate through the group? 1. O(log(N)) time 2. O(N) time 3. O(1) time 4. None of these options