- 28 3月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Added a new demo named "BFTList", which is very similar to "BFTMap", but uses Lists instead of Maps.
-
- 21 3月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
-
- 04 3月, 2012 2 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Started changing the state transfer protocol to not perform checkpointing. Such procedure should be delegated to the applcation instead. THIS IS AN UNSTABLE COMMIT!!!!!! USE AT OWN RISK!!!!!!!!!!!!!!!!!!
-
由 liquidsnake@sapo.pt 提交于
Started changing the state transfer protocol to not perform checkpointing. Such procedure should be delegated to the applcation instead. THIS IS AN UNSTABLE COMMIT!!!!!! USE AT OWN RISK!!!!!!!!!!!!!!!!!!
-
- 02 3月, 2012 1 次提交
-
-
由 marcelmhs@gmail.com 提交于
Removed the methodes invoke(byte[], boolean readOnly) from ServiceProxy. Fixed demo code to work with the new version of ServiceProxy.
-
- 28 2月, 2012 2 次提交
-
-
由 marcelmhs@gmail.com 提交于
Changed ServiceReplica to set the ReplyListener to null upon receive of invoke() call.
-
由 marcelmhs@gmail.com 提交于
- Changed TOMMessageType from REQUEST and READONLY_REQUEST to ORDERED_REQUEST and UNORDERED_REQUEST respectively - Added method in TOMSender to send messages to target replicas instead of send to all replicas
-
- 18 2月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Fixed a bug on the Round object, that could have an updated array of processes in relation to the ServerViewManager. This bug would trigger an arrayoutofbound exception.
-
- 16 2月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Added an optimization that decreases the leader change timeout upon a new view that removes the current leader from the group. There is also a debugging 'try-catch' on the "Round" class, that I'm using to track a bug that I couldn't resolve yet.
-
- 09 2月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Updated the state transfer protocol to also include the latest view, in order to avoid delayed replicas from getting blocked after applying the state. If the view had changed in the middle of a state transfer and the view updates where already erased from the state log, the replica would alreays discard client requests, since they where for a view ahead of the one it had. Also fixed a bug that wouldn't let the replica execute the requests contained in a state transfer log, also quase by the fact that a client might had a different view from the replica.
-
- 04 2月, 2012 1 次提交
-
-
由 reiser@cs.fau.de 提交于
-
- 01 2月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Fixed a bug that would manifest in a delayed replica: if a timeout occurred, the replica would be forever stopped, and the bug prevented it from triggering the state transfer protocol. With the bug gone, the replica will ask for the state. Regarding the aforementioned bug, I added a parameter to the config file dubbed "timeoutHighmark". It should be lower than "paxosHighmark" and greater than "revivalHighmark". This will make the replica trigger the state transfer more quickly, given the conditions just described. Fixed also a bug on the clients: when there were no "currentView" file, an exception would be thrown, and the clients would not read the host.config file. This no longer happens. Finally, added some debugger messages in the leader change protocol and also in the servicereplica class.
-
- 28 1月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
-
- 27 1月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Fixed a bug on the leader election. It didn't work well together with the reconfiguration. Modified the LC protocol plus the state transfer to fix this issue.
-
- 19 1月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
-
- 14 1月, 2012 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Small additions to the Throughput demo. Ir now has an option for DoS and shows the maximum throughput on the servers
-
- 09 1月, 2012 1 次提交
-
-
由 marcelmhs@gmail.com 提交于
- ServiceReplica.receiveMessages() is now implementing the code for batch message processing; - BatchExecutable.executeBatch() signature was changed to has an array of MessageContext as parameter.
-
- 04 1月, 2012 2 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Fixed a bug on the leader change protocol that would block all replicas if the delayed replica triggered again a leader change before the others finished the same protocol. Identified a bug on the method "send" of class NettyClientServerCommunicationSystemServerSide, but its cause remains misterious: if invoked 2 times for the same message (why it sometimes is invoked 2 times for the same message is not known), the client may not receive a reply on its end, and if no sufficent replies are received, the client blocks. If locks are used in this method, the problem seems to disappear. Removed various locks from the leader change protocol which were rendered useless in previous commits. The Proposed thread now unlocks by this order: I am the leader -> the consensus as finished -> I have message to order
-
由 marcelmhs@gmail.com 提交于
Now Executable is the top class. It has the executeUnordered method, as all executables can process readonly messages. There are two classes extending Executable. SingleExecutable has the executeOrdered method to process a sigle command. BatchExecutable has the executeBatch to process multiple commands. BatchExecutable was also changed to have a matrix of bytes as parameter, corresponding to an array of commands, having an array of bytes. Demo classes was changed to use the new Executable structure. ServiceReplica was also changed.
-
- 14 12月, 2011 3 次提交
-
-
由 marcelmhs@gmail.com 提交于
-
由 marcelmhs@gmail.com 提交于
- Added tests for killing leaders; - Added Logger to log tests in console; - Added ConsoleTest to execute tests in Console window instead of JUnit.
-
由 liquidsnake@sapo.pt 提交于
Fixed a bug on the state transfer protocol, that would render the leader change protocol from changing leader a second time. However, the leader change is still unstable when ciclying through replicas. I'm investigating the issue. This commit as a lot of debugging message related to the leader change protocol.
-
- 13 12月, 2011 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Fixed a bug on the state transfer protocol. It will now longer abort if the replica which should send the complete state remains mute; instead, it will re-initialize the protocol. This way, the system will no longer freeze when trying to add a replica to the group while the original leader is down (because there are not enough replicas to have a quorum during the first phase of the consensus algorithm). Also done a minor change in the implementation: upon a timeout, the computations will be passed to the CS thread, in a similar way in which is done with the leader change protocol. Some refractoring was also done.
-
- 10 12月, 2011 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Fixed a small bug that would render the state transfer protocol less efficient, although still functional. Found a small defect on the protocol, but not fixed yet (I want to discuss the issue first with Professor Bessani). Also, substituted a printstack trace by a system out, which would be displayed when the CS couldn't connect to another replica. Since this was very periodic, it would make difficult to do debugging with such an excpetion constantly appearing.
-
- 08 12月, 2011 2 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Small but important change in the leader change implementation: When a replica suspects the leader, such computations will be done by the CS thread isntead of the RequestTask Thread. This shall reduce the changes of concurrency and race conditions, thus, making the code more resilient.
-
由 liquidsnake@sapo.pt 提交于
Small but important change in the communication system: when a connection to a replica is lost, the OutQueue is erased. This is meant to avoid gaps in the sequence of messages upon reconnection, and thus, avoid the "break" of upper protocols. Also, small refractor in TomUtil.
-
- 07 12月, 2011 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
-
- 06 12月, 2011 4 次提交
-
-
由 marcelmhs@gmail.com 提交于
-
由 marcelmhs@gmail.com 提交于
- Changed ServiceReplica to implement TOMReceiver; - Created interfaces Executable and Recoverable, so the replicas can choose which methods to implement; - Changed replicas to use ServiceReplica instead of extend it.
-
由 liquidsnake@sapo.pt 提交于
Modified slightly the throughput client and server. You can now specify the size of the application state.
-
由 liquidsnake@sapo.pt 提交于
-
- 03 12月, 2011 1 次提交
-
-
由 marcelmhs@gmail.com 提交于
-
- 02 12月, 2011 1 次提交
-
-
由 marcelmhs@gmail.com 提交于
Moved executeOrdered method to the server implementation. Added ant code to compile and export Jar of the project. Added JUnit tests for the keyvalue demo. Refactored keyvalue code.
-
- 24 11月, 2011 2 次提交
-
-
由 marcelmhs@gmail.com 提交于
No commit message
-
由 marcelmhs@gmail.com 提交于
No commit message
-
- 19 11月, 2011 1 次提交
-
-
由 liquidsnake@sapo.pt 提交于
Huge update: merging corrections of Eduardo with current version of the repository. The reconfiguration protocol should work properly now!
-
- 16 9月, 2011 4 次提交
-
-
由 liquidsnake@sapo.pt 提交于
No commit message
-
由 liquidsnake@sapo.pt 提交于
No commit message
-
由 liquidsnake@sapo.pt 提交于
-
由 liquidsnake@sapo.pt 提交于
-