- 27 1月, 2019 1 次提交
-
-
由 João Sousa 提交于
-
- 23 1月, 2019 1 次提交
-
-
由 João Sousa 提交于
-
- 22 1月, 2019 6 次提交
-
-
由 João Sousa 提交于
-
由 João Sousa 提交于
-
由 João Sousa 提交于
This reverts commit 1a2aecd5.
-
由 João Sousa 提交于
-
由 João Sousa 提交于
Server-side netty channels and now synchronous Refined control flow mechanism to also oversee the number of pending replie waiting to arrive at clients.
-
由 João Sousa 提交于
-
- 21 1月, 2019 1 次提交
-
-
由 João Sousa 提交于
-
- 18 12月, 2018 1 次提交
-
-
由 João Sousa 提交于
Fixed race condition in reconfiguration that would occur when batch execution was slower than consensus processing.
-
- 23 11月, 2018 1 次提交
-
-
由 João Sousa 提交于
Fixed vulnerability that would cause the system to block if a client issued a malformed/invalid reconfiguration request.
-
- 30 10月, 2018 1 次提交
-
-
由 João Sousa 提交于
-
- 28 9月, 2018 1 次提交
-
-
由 João Sousa 提交于
Changed the way to detected leader replay attacks. This is now done using the last request submitted to the delivery thread from each client.
-
- 07 8月, 2018 1 次提交
-
-
由 João Sousa 提交于
Also replaced all System.out.println and ex.printStackTrace methods with adequate SLF4J invocations. Included xml configuration file for logback.
-
- 19 7月, 2018 1 次提交
-
-
由 João Sousa 提交于
Created new option to force all replicas to receive the same number of requests per batch. Not to be confused with the batch used for the PROPOSE message from the ordering protocol.
-
- 04 11月, 2017 1 次提交
-
-
由 João Sousa 提交于
Last two commits rendered most of the constructors in TOMMessage useless, so they were removed from the rest of the code.
-
- 06 5月, 2017 1 次提交
-
-
由 João Sousa 提交于
Fixed a bug on the client that would make the virtual machine exaust the heap space if asynchronous clients aggressively sent requests to the servers.
-
- 22 11月, 2016 2 次提交
-
-
由 snakejerusalem 提交于
-
由 snakejerusalem 提交于
-
- 23 9月, 2016 1 次提交
-
-
由 snakejerusalem 提交于
-
- 14 9月, 2016 1 次提交
-
-
由 snakejerusalem 提交于
-
- 17 5月, 2016 2 次提交
-
-
由 Joao Sousa 提交于
-
由 Joao Sousa 提交于
Renamed ServiceReplica.shutdown() to ServiceReplica.kill(). Also made some modifications to make the act of a replica leave the group more robust.
-
- 16 5月, 2016 2 次提交
-
-
由 snakejerusalem 提交于
After adding the ServiceReplica.shutdown() and ServiceReplica.restart() methods, I used them to fix a bug related with the reconfiguration protocol. The bug happened when a replica was removed from the group and them re-added to it, and it would prevent the replica from correctly resuming execution. The main idea is to "clean" the replica after it is removed from the group using the aforementioned commands, and when it is re-added, it is a tabula rasa.
-
由 snakejerusalem 提交于
Implemented the ServiceReplica.shutdown() method, which stops the service execution at a replica. It will shutdown all threads, stop the requests' timer, drop all enqueued requests, thus letting the ServiceReplica object be garbage-collected.
-
- 24 12月, 2015 2 次提交
-
-
由 Joao Sousa 提交于
Performed refractors on a few variables that are of the type 'CertifiedDecision', so that nomenclature becomes consistent.
-
由 Joao Sousa 提交于
-
- 23 12月, 2015 3 次提交
-
-
由 Joao Sousa 提交于
-
由 Joao Sousa 提交于
-
由 Joao Sousa 提交于
-
- 22 12月, 2015 1 次提交
-
-
由 Joao Sousa 提交于
MessageContext now has more info about the each request that is delivered to the application. This info can be used to recreate the TOMMessage object to which the request was originally associated.
-
- 17 12月, 2015 2 次提交
-
-
由 snakejerusalem 提交于
Do not panic! I only performed refractores on all variables/methods whose names contained the sequence 'eid'. This sequence of letters (which meant 'execution ID') are now replaced by 'cid' (which means 'consensus ID'). I performed this refractor to make the rest of the code match the renaming of class 'Execution' into 'Consensus'. There were a lot of them across the entire code, hence why so many classes where affected.
-
由 joao 提交于
-
- 16 12月, 2015 1 次提交
-
-
由 Joao Sousa 提交于
ServiceReplica will now invoke the new method of the Recoverable interface, so that applications can use the raw decision to build their logs. This commit also includes minor changes required from the modification of the previous commits.
-
- 12 12月, 2015 2 次提交
-
-
由 Joao Sousa 提交于
Do not panic! I only made a refractor to the name of the LastEidData class. Now it is called CertifiedDecision. I also refractored some of its methods names.
-
由 Joao Sousa 提交于
MessageContext now holds a LastEidData object instead of a set of ConsensusMessage's. If I had added such type of objects instead, it would have saved me time for what I am implementing now (which will come in the next commit) . I feel dumb :(
-
- 10 12月, 2015 2 次提交
-
-
由 Joao Sousa 提交于
MessageContext now only keeps the number of nonces and respective seed. It will only generate the nonces when the application explicitly asks for them via the getter.
-
由 Joao Sousa 提交于
In addition, the nonces will now also be stored in the state transfer log.
-
- 28 11月, 2015 1 次提交
-
-
由 Joao Sousa 提交于
Do not panic! I only added a new parameter to the MessageContext constructor, to know either or not operations are NoOp (the reason why these operations may happen is explained in commits from January and code comments). Recoverables now also check this parameter to know if they should deliver the operation(s) to the application.
-
- 27 11月, 2015 1 次提交
-
-
由 Joao Sousa 提交于
Do not panic! I only performed a refractor to rename the 'Execution' class to 'Consensus', given that the content and usage of the class reflects the implementation of such protocol. Also performed a refractor in some variables and methods that either represent an instance of the class or a consensus ID.
-