Dear Wiki user, You have subscribed to a wiki page or wiki category on "Hadoop Wiki" for change notification.
The "ZooKeeper/GSoCFailureDetector" page has been changed by AbmarBarros. http://wiki.apache.org/hadoop/ZooKeeper/GSoCFailureDetector?action=diff&rev1=7&rev2=8 -------------------------------------------------- * Naohiro Hayashibara, Xavier Défago, Rami Yared, Takuya Katayama, '''The Φ Accrual Failure Detector''', srds, pp.66-78, 23rd IEEE International Symposium on Reliable Distributed Systems (SRDS'04), 2004 == Specific objectives == + + The ones with strike-through have already been finished + - 1. Write pseudo-codes for the proposed algorithms + 1. --(Write pseudo-codes for the proposed algorithms)-- - 1. Create FailureDetector interface + 1. --(Create FailureDetector interface)-- - 1. Write implementations and tests of the FailureDetector interface based on the proposed algorithms + 1. --(Write implementations and tests of the FailureDetector interface based on the proposed algorithms)-- - 1. Refactor client-side code of the client-server monitoring to use the proposed FailureDetector interface + 1. --(Refactor client-side code of the client-server monitoring to use the proposed FailureDetector interface)-- - 1. Make the failure detection and its parameters configurable on the client + 1. --(Make the failure detection and its parameters configurable on the client)-- - 1. Refactor server-side code of the client-server monitoring to use the proposed FailureDetector interface + 1. --(Refactor server-side code of the client-server monitoring to use the proposed FailureDetector interface)-- 1. Refactor the code of the server-server monitoring to use the proposed FailureDetector interface 1. Make the failure detection and its parameters configurable on the server (to server-server and client-server monitoring) 1. Evaluate the QoS metrics with experimentation @@ -42, +45 @@ * [[https://issues.apache.org/jira/browse/ZOOKEEPER-811|ZOOKEEPER-811: Refactor server to server monitoring]] * [[https://issues.apache.org/jira/browse/ZOOKEEPER-810|ZOOKEEPER-810: Write Forrest docs]] * [[https://issues.apache.org/jira/browse/ZOOKEEPER-812|ZOOKEEPER-812: Evaluate QoS metrics]] + == Progress Report == + + ==== Community bonding period ==== + * Studied ZooKeeper code regarding failure detection + * Studied the proposed failure detection algorithms + + ==== 05/Jun/10 ==== + * Discussed whether FD instance should run in the same thread of the application + * Proposed first version of the FailureDetector interface. It does not consider application messages as heartbeats. + * Adapted ClientCnxn class to use the proposed interface. + + ==== 11/Jun/10 ==== + * Written pseudo-codes for the proposed failure detection algorithms + * Started discussion on how could the application scheduling interval could interfere on adaptive FD methods. + + ==== 22/Jun/10 ==== + * Attached the classes of the initially proposed FD methods (Phi Accrual, Chen, Bertier, Fixed Heartbeat) and the corresponding unit tests. + * Included suggestions Flavio gave concerning package naming and method scope. + + ==== 28/Jun/10 ==== + * Started discussion on how to configure the FD method and its parameters at the client and server sides. + + ==== 02/Jul/10 ==== + * Enhanced pseudocodes documentation + * Created appMessageReceived() and appMessageSent() methods. These methods allow the Failure Detector to use application messages as heartbeats, which reflects the ZooKeeper case. + * Added command line options to the client side in order to configure failure detector method and its parameters. + * Unit tests expanded to comply with new methods. + * Enhanced javadocs for each Failure Detector implementation + + ==== 08/Jul/10 ==== + + * Adapted failure detectors to work on both client and server sides of the client-server monitoring. + * Refactored server-side code to use the FailureDetector interface. + * Created a new FailureDetector, which groups monitoreds by their tick time, similar to what the SessionTracking does. + * Made the server-client failure detector (and its parameters) options of the ZooKeeper configuration file. + + === Currently working on === + * Refactoring the code of the server-server monitoring to use the proposed FailureDetector interface + + ---- == Design decisions == @@ -52, +95 @@ * Concurrency issues must be handled. * Benefits * The FD will run in a more independent way, and it will notify the application of changes using listeners or callbacks. If it runs in the same thread, the application must signal the failure detector of changes, and also retrieve its status periodically. In other words, the application code will be coupled to FD code. + * Decided to use the FD on the same thread of the application
