You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@activemq.apache.org by "Vadim Katz (JIRA)" <ji...@apache.org> on 2008/09/11 19:24:52 UTC
[jira] Created: (AMQ-1934) Message delivery blocks after broker
failover
Message delivery blocks after broker failover
---------------------------------------------
Key: AMQ-1934
URL: https://issues.apache.org/activemq/browse/AMQ-1934
Project: ActiveMQ
Issue Type: Bug
Affects Versions: 5.1.0
Environment: Windows XP, ActiveMQ 5.1.0, Spring 2.5
Reporter: Vadim Katz
Attachments: activemq-failover-test.zip
Start two brokers with shared FS. One is locked out because journal is locked by the master as it should be. Then run attached junit test case. The test does the following:
1. Sends and receives a message while both master and slave brokers are up.
2. Requests manual shutdown of master broker and expects slave to become master within 30 sec
3. Sends another message which is never received
You can observe via admin console that the message sent after broker failover is delivered to the queue; however it is not delivered to a waiting listener. Also, you can observe the logged invocations of TranportListener. Notice, how onCommand() calls are no longer continuously made after failover (except for the first two).
Eclipse/Maven based test project is attached.
Please see "sends/receives block during failover" discussion for more information. thanks,
vadim
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (AMQ-1934) Message delivery blocks after broker
failover
Posted by "Vadim Katz (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/activemq/browse/AMQ-1934?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=47570#action_47570 ]
Vadim Katz commented on AMQ-1934:
---------------------------------
Tried with 5.2.0 and still experience the problem
> Message delivery blocks after broker failover
> ---------------------------------------------
>
> Key: AMQ-1934
> URL: https://issues.apache.org/activemq/browse/AMQ-1934
> Project: ActiveMQ
> Issue Type: Bug
> Affects Versions: 5.1.0
> Environment: Windows XP, ActiveMQ 5.1.0, Spring 2.5
> Reporter: Vadim Katz
> Attachments: activemq-failover-test.zip
>
>
> Start two brokers with shared FS. One is locked out because journal is locked by the master as it should be. Then run attached junit test case. The test does the following:
> 1. Sends and receives a message while both master and slave brokers are up.
> 2. Requests manual shutdown of master broker and expects slave to become master within 30 sec
> 3. Sends another message which is never received
> You can observe via admin console that the message sent after broker failover is delivered to the queue; however it is not delivered to a waiting listener. Also, you can observe the logged invocations of TranportListener. Notice, how onCommand() calls are no longer continuously made after failover (except for the first two).
> Eclipse/Maven based test project is attached.
> Please see "sends/receives block during failover" discussion for more information. thanks,
> vadim
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (AMQ-1934) Message delivery blocks after broker
failover
Posted by "Rob Davies (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/activemq/browse/AMQ-1934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Rob Davies updated AMQ-1934:
----------------------------
Fix Version/s: 5.4.0
> Message delivery blocks after broker failover
> ---------------------------------------------
>
> Key: AMQ-1934
> URL: https://issues.apache.org/activemq/browse/AMQ-1934
> Project: ActiveMQ
> Issue Type: Bug
> Affects Versions: 5.1.0
> Environment: Windows XP, ActiveMQ 5.1.0, Spring 2.5
> Reporter: Vadim Katz
> Fix For: 5.4.0
>
> Attachments: activemq-failover-test.zip
>
>
> Start two brokers with shared FS. One is locked out because journal is locked by the master as it should be. Then run attached junit test case. The test does the following:
> 1. Sends and receives a message while both master and slave brokers are up.
> 2. Requests manual shutdown of master broker and expects slave to become master within 30 sec
> 3. Sends another message which is never received
> You can observe via admin console that the message sent after broker failover is delivered to the queue; however it is not delivered to a waiting listener. Also, you can observe the logged invocations of TranportListener. Notice, how onCommand() calls are no longer continuously made after failover (except for the first two).
> Eclipse/Maven based test project is attached.
> Please see "sends/receives block during failover" discussion for more information. thanks,
> vadim
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (AMQ-1934) Message delivery blocks after broker
failover
Posted by "Steve Mays (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/activemq/browse/AMQ-1934?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=60020#action_60020 ]
Steve Mays commented on AMQ-1934:
---------------------------------
We are on 5.3.1 and still seeing this issue!
When you have a single hot/cold pair its not a big thing, but we are implementing a network of brokers at Edmunds.com (essentially a stripe across several hot/cold pairs) and this will get VERY yucky as the only way we have gotten fail BACK to work is to shut down all brokers and delete KahaDB and backing store then restart. Yuck!
We'd sure like something on this issue too!
Thanks,
Steve
> Message delivery blocks after broker failover
> ---------------------------------------------
>
> Key: AMQ-1934
> URL: https://issues.apache.org/activemq/browse/AMQ-1934
> Project: ActiveMQ
> Issue Type: Bug
> Affects Versions: 5.1.0
> Environment: Windows XP, ActiveMQ 5.1.0, Spring 2.5
> Reporter: Vadim Katz
> Fix For: NEEDS_REVIEWED
>
> Attachments: activemq-failover-test.zip
>
>
> Start two brokers with shared FS. One is locked out because journal is locked by the master as it should be. Then run attached junit test case. The test does the following:
> 1. Sends and receives a message while both master and slave brokers are up.
> 2. Requests manual shutdown of master broker and expects slave to become master within 30 sec
> 3. Sends another message which is never received
> You can observe via admin console that the message sent after broker failover is delivered to the queue; however it is not delivered to a waiting listener. Also, you can observe the logged invocations of TranportListener. Notice, how onCommand() calls are no longer continuously made after failover (except for the first two).
> Eclipse/Maven based test project is attached.
> Please see "sends/receives block during failover" discussion for more information. thanks,
> vadim
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (AMQ-1934) Message delivery blocks after broker
failover
Posted by "Vadim Katz (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/activemq/browse/AMQ-1934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Vadim Katz updated AMQ-1934:
----------------------------
Could someone please comment on this? It seems very serious because we can't rely on AcriveMQ failover.
> Message delivery blocks after broker failover
> ---------------------------------------------
>
> Key: AMQ-1934
> URL: https://issues.apache.org/activemq/browse/AMQ-1934
> Project: ActiveMQ
> Issue Type: Bug
> Affects Versions: 5.1.0
> Environment: Windows XP, ActiveMQ 5.1.0, Spring 2.5
> Reporter: Vadim Katz
> Attachments: activemq-failover-test.zip
>
>
> Start two brokers with shared FS. One is locked out because journal is locked by the master as it should be. Then run attached junit test case. The test does the following:
> 1. Sends and receives a message while both master and slave brokers are up.
> 2. Requests manual shutdown of master broker and expects slave to become master within 30 sec
> 3. Sends another message which is never received
> You can observe via admin console that the message sent after broker failover is delivered to the queue; however it is not delivered to a waiting listener. Also, you can observe the logged invocations of TranportListener. Notice, how onCommand() calls are no longer continuously made after failover (except for the first two).
> Eclipse/Maven based test project is attached.
> Please see "sends/receives block during failover" discussion for more information. thanks,
> vadim
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.