activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel Laügt (JIRA) <j...@apache.org>
Subject [jira] [Updated] (AMQCPP-491) Connection didn't switch to the slave broker when the master broker is down
Date Thu, 13 Jun 2013 18:19:20 GMT

     [ https://issues.apache.org/jira/browse/AMQCPP-491?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Daniel Laügt updated AMQCPP-491:
--------------------------------

    Description: 
Master/Slave configuration with a shared data directory:
ActiveMQ Configuration (activemq.xml):
<persistenceAdapter>
  <kahaDB directory="\\host1\kahadb"/>
</persistenceAdapter>

URL:
failover:(tcp://host1:port1,tcp://host2:port2)?randomize=false

Test case:
1) Run ActiveMQ on host2, it becomes the master
2) Run ActiveMQ on host1, it becomes the slave
3) Run a program that opens an ActiveMQ connection with the url
4) Shutdown ActiveMQ on host2
5) The ActiveMQ connection is not able to switch to host1.

Explanation:
The test case has randomize=false to be able to reproduce the problem anytime. If randomize=true,
the problem occurs randomly.

At step 3, ActiveMQ-CPP tries to open first a connection with host1. It fails as host1 is
the slave. Then it tries to open a connection with host2. This time, it succeeds but it forgets
to put back the failure hosts (host1) to the broker list. At this point, the broker list =
(host2).

At step 5, ActiveMQ-CPP tries to open a connection with the broker list. As the broker list
= (host2), it will try to open a connection only with host2.


  was:
Master/Slave configuration with a shared data directory:
ActiveMQ Configuration (activemq.xml):
<persistenceAdapter>
  <kahaDB directory="\\host1\kahadb"/>
</persistenceAdapter>

URL:
failover:(tcp://host1:port1,tcp://host2:port2)?randomize=false

Test case:
1) Run ActiveMQ on host2, it becomes the master
2) Run ActiveMQ on host1, it becomes the slave
3) Run a program that opens an ActiveMQ connection
4) Shutdown ActiveMQ on host2
5) The ActiveMQ connection is not able to switch to host1.

Explanation:
The test case has randomize=false to be able to reproduce the problem anytime. If randomize=true,
the problem occurs randomly.

At step 3, ActiveMQ-CPP tries to open first a connection with host1. It fails as host1 is
the slave. Then it tries to open a connection with host2. This time, it succeeds but it forgets
to put back the failure hosts (host1) to broker lists. At this point, the broker list = {host2}.

At step 5, ActiveMQ-CPP tries to open a connection with the broker list. As the broker list
= {host2}, it will try to open a connection only with host2.


    
> Connection didn't switch to the slave broker when the master broker is down
> ---------------------------------------------------------------------------
>
>                 Key: AMQCPP-491
>                 URL: https://issues.apache.org/jira/browse/AMQCPP-491
>             Project: ActiveMQ C++ Client
>          Issue Type: Bug
>          Components: Transports
>    Affects Versions: 3.7.0
>         Environment: Windows 7, VS2010
>            Reporter: Daniel Laügt
>            Assignee: Timothy Bish
>            Priority: Minor
>         Attachments: FailoverTransport.cpp.patch
>
>   Original Estimate: 24h
>  Remaining Estimate: 24h
>
> Master/Slave configuration with a shared data directory:
> ActiveMQ Configuration (activemq.xml):
> <persistenceAdapter>
>   <kahaDB directory="\\host1\kahadb"/>
> </persistenceAdapter>
> URL:
> failover:(tcp://host1:port1,tcp://host2:port2)?randomize=false
> Test case:
> 1) Run ActiveMQ on host2, it becomes the master
> 2) Run ActiveMQ on host1, it becomes the slave
> 3) Run a program that opens an ActiveMQ connection with the url
> 4) Shutdown ActiveMQ on host2
> 5) The ActiveMQ connection is not able to switch to host1.
> Explanation:
> The test case has randomize=false to be able to reproduce the problem anytime. If randomize=true,
the problem occurs randomly.
> At step 3, ActiveMQ-CPP tries to open first a connection with host1. It fails as host1
is the slave. Then it tries to open a connection with host2. This time, it succeeds but it
forgets to put back the failure hosts (host1) to the broker list. At this point, the broker
list = (host2).
> At step 5, ActiveMQ-CPP tries to open a connection with the broker list. As the broker
list = (host2), it will try to open a connection only with host2.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message