Quiz 9 Flashcards
A system is k-fault tolerant if:
It can survive when k processes failed
It can survive when all of its processes except k processes failed
It can survive when k+1 processes failed
It can survive when the majority of its processes failed
It can survive when k processes failed
What major problems can we face when client process crashes while its request has already been forwarded to the server for processing?
The client cannot fulfil its job
The client’s request may create orphan that may lock up files and tie up valuable resources
The client’s request may create orphan that may waste server’s CPU time
Old reply from the server can create confusion for the client when it reboots and sends a fresh request
The client’s request may create orphan that may lock up files and tie up valuable resources
The client’s request may create orphan that may waste server’s CPU time
Old reply from the server can create confusion for the client when it reboots and sends a fresh request
A major problem with reliable multicasting with feedback messages is:
ordering of multicast messages
faulty processes
feedback implosion
feedback implosion
How many processes are needed to form a byzantine agreement in a group with k faulty processes?
3k+1
3k
2k+1
k+1
3k+1
A system is said to be fault tolerant if:
it can continue to operate in the presence of faults
a partial failure can crash the system but it can be repaired later
it can tolerate crash failure but not arbitrary failure
it can tolerate crash and omission failures but not arbitrary failure
it can continue to operate in the presence of faults
A type of failure(s) where the server’s response can violate integrity of data is:
Omission failure
Byzantine failure
Crash failure
Response failure
Timing failure
Byzantine failure
Response failure
A general technique to provide process resilience in distributed systems is to:
use fault-removal technique
organize several identical processes into a group
forecast the fault so that systems can take action when the fault occurs
organize several identical processes into a group
A client can retransmit the request when dealing with lost reply messages from server to client. Which of the statements below is/are correct:
Server may not care about duplicate requests if the request is idempotent
Server needs not care about the retransmitted request at any time
Server needs to care about duplicate requests if the request is non-idempotent
Server may not care about duplicate requests if the request is idempotent
Server needs to care about duplicate requests if the request is non-idempotent
One challenge with non-hierarchical feedback control to improve scalability in reliable multicasting is that
Sender has to send messages to coordinator
Receivers that received original message correctly have to process retransmitted messages that are useless to them
Receivers have to send acknowledgement for messages they did not receive
Receivers that received original message correctly have to process retransmitted messages that are useless to them
How many processes are needed in a leaderless protocol to obtain a correct result from a majority vote in the event of crash or arbitrary failures by k processes
k+1
3k+1
2k+1
2k+1
