using 2 --direct tungsten instances in the same host, one of them fails

What steps will reproduce the problem?
1. Install 2 instances of Tungsten with the --direct option
./tools/tungsten-installer \
    --direct \
    --master-host=db1 \
    --master-port=7101 \
    --slave-host=db2 \
    --slave-port=7102 \
    --master-user=root \
    --slave-user=root \
    --master-password=msandbox \
    --slave-password=msandbox \
    --master-log-directory=$HOME/sandboxes/tr_dbs/node1/data \
    --service-name=Castor \
    --thl-port=12112 \
    --rmi-port=10101 \
    --channels=1 \
    --home-directory=$TUNGSTEN_BASE \
    --thl-directory=$TUNGSTEN_BASE/thl1 \
    --relay-directory=$TUNGSTEN_BASE/relay1 

./tools/tungsten-installer \
    --direct \
    --master-host=db1 \
    --master-port=7101 \
    --slave-host=db3 \
    --slave-port=7103 \
    --master-user=root \
    --slave-user=root \
    --master-password=msandbox \
    --slave-password=msandbox \
    --master-log-directory=$HOME/sandboxes/tr_dbs/node1/data \
    --service-name=Pollux \
    --thl-port=22112 \
    --rmi-port=10101 \
    --channels=1 \
    --home-directory=$TUNGSTEN_BASE \
    --thl-directory=$TUNGSTEN_BASE/thl2 \
    --relay-directory=$TUNGSTEN_BASE/relay2 

2. start the replicator

3. create a table in the master

What is the expected output? 
  both instances replicate the event

What do you see instead?
  one slave has got less relay logs than the other, and the replicator log complains about EOF event in the relay log

$ ls -R relay*
relay1:
Castor

relay1/Castor:
mysql-bin.000002  mysql-bin.000003  mysql-bin.index

relay2:
Pollux

relay2/Pollux:
mysql-bin.000003  mysql-bin.index

$ grep -i error tungsten/tungsten-replicator/log/trepsvc.log 
INFO   | jvm 1    | 2011/05/19 15:37:25 | 2011-05-19 15:37:25,787 ERROR 
extractor.mysql.RelayLogTask Relay log task failed due to exception: EOF packet 
received
INFO   | jvm 1    | 2011/05/19 15:38:25 | 2011-05-19 15:38:25,811 ERROR 
extractor.mysql.MySQLExtractor Failed to extract from mysql-bin.000003 (106)
INFO   | jvm 1    | 2011/05/19 15:38:25 | 2011-05-19 15:38:25,811 ERROR 
replicator.management.OpenReplicatorManager Received error notification, 
shutting down services: Event extraction failed: Relay log task has 
unexpectedly terminated; logs may not be accessible
INFO   | jvm 1    | 2011/05/19 15:38:25 | 2011-05-19 15:38:25,815 INFO  
replicator.management.OpenReplicatorManager Sent State Change Notification 
ONLINE -> OFFLINE:ERROR
Original issue reported on code.google.com by g.maxia on 19 May 2011 at 1:48
Merged into: #59
zkfan / tungsten-replicator

using 2 --direct tungsten instances in the same host, one of them fails #78
