Recent changes to bugs

Total order not respected

2012-07-11T13:45:41Z

I believe I've come across a problem with the total ordering in the sequenceruniform protocol. The transaction execution (and outcome) trace diverges when using Appia, but not when using JGroups. Details are attached.

Bug on the SETO protocol when uniformity period is set to 0

2010-09-24T12:41:12Z

When the uniformity period is set to 0, the protocol does not deliver messages.
This applies to the SETO and SequencerUniform total order protocols.

New sequencer based protocol with uniformity

2010-09-24T12:39:16Z

Create a new sequencer based total order protocol based on SETO, but without the estimated optimistic delivery.

GroupInit constructor does not store Endpt argument

2009-02-15T16:56:58Z

The GroupInit constructor does not store the value of the endpt parameter. This prevents the correct clonning of the event.

I need the endpt parameter to be stored as I am extending GroupInit and when using the super() method to call this constructor I need to have the properties cloned in the child class.

ViewState Serialization

2009-02-15T15:55:58Z

There is a bug in the file ViewState.java.

When deserializing a ViewState object, a null pointer exception is raised.

The patch to correct the problem in the attachement.

Wrong cast in TcpCompleteSession

2008-10-13T07:49:26Z

Class: net.sf.appia.protocols.tcpcomplete.TcpCompleteSession
Method: private void handleChannelClose(ChannelClose e)
line: 311

((TcpReader)(it.next())).setRunning(false)

Iterator it iterates on instances of SocketInfoContainer but it is casted to TcpReader.
A ChannelClose event throws a ClassCastException.

wrong packages, GossipServer and Validate

2007-11-23T15:16:45Z

On the bin package the path to the GossipServer class is wrong.

Where it is:

java org.continuent.appia.gossip.GossipServer "$@"

Should be:

java net.sf.appia.gossip.GossipServer "$@"

Am I right? I'm using version 4.0 of Appia

NPE on the total sequencer layer

2007-10-08T12:45:12Z

Exception in thread "main" java.lang.NullPointerException
at org.continuent.appia.protocols.total.sequencer.TotalSequencerSession.orderAlgorithm(TotalSequencerSession.java:382)
at org.continuent.appia.protocols.total.sequencer.TotalSequencerSession.cleanCounters(TotalSequencerSession.java:364)
at org.continuent.appia.protocols.total.sequencer.TotalSequencerSession.handleView(TotalSequencerSession.java:331)
at org.continuent.appia.protocols.total.sequencer.TotalSequencerSession.handle(TotalSequencerSession.java:91)
at org.continuent.appia.core.EventScheduler.consumeEvent(EventScheduler.java:173)
at org.continuent.appia.core.Appia.instanceRun(Appia.java:141)
at org.continuent.appia.core.Appia.run(Appia.java:253)
at org.dmd.appia.test.Main.main(Main.java:60)

SendableEvents do not free the bytes in memory manager

2007-08-24T12:59:26Z

These protocols should call the discardAll() method of the message in order to free the "memory" in the memory manager and allow that the asyncgo method do not block when the threshold is reached.

This was intentional, in order to get better performance... The protocols should always discard the bytes of the message before it is garbage collected, but this proved incompatible with OO programming and the programmers of protocols always forget this.

The deallocation of "bytes" should be ensured by the kernel, but the previous mechanism should be kept and seen as an optimization.

Excepton on message reception (jGCS)

2007-08-06T17:19:13Z

Exception:

...
CAUSAL Processing event net.sf.jgcs.appia.protocols.top.JGCSGroupEvent@18e546
Exception in thread "PullPushThread" java.lang.IndexOutOfBoundsException: Index: 3, Size: 3
at java.util.ArrayList.RangeCheck(ArrayList.java:546)
at java.util.ArrayList.get(ArrayList.java:321)
at net.sf.jgcs.appia.AppiaMembership.getMemberAddress(Unknown Source)
at net.sf.jgcs.appia.AppiaDataSession$PullPushWorker.run(Unknown Source)
at java.lang.Thread.run(Thread.java:595)
CAUSAL Processing event org.continuent.appia.protocols.total.seto.SeqOrderEvent@160088f
...