[
https://issues.apache.org/activemq/browse/AMQNET-258?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=60253#action_60253
]
Jim Gomes commented on AMQNET-258:
----------------------------------
I briefly reviewed the log files, and here are my initial thoughts on this:
1. I would be curious to see what the Java client would do in this fail-over
scenario.
2. What kind of error logs are produced by the broker, if any?
3. It's not clear where the issue lies in this scenario. Is it a problem with
the client, or the broker's handling of failover? (This is where the Java
client and the server logs would be helpful in gaining some insight.)
4. Is there an actual semantic error? Meaning: what is the final disposition
of the messages being sent while the failover is occurring? Are any messages
lost or duplicated, or any other "bad thing"?
> Getting Unknown Response ID errors in client log on failover
> ------------------------------------------------------------
>
> Key: AMQNET-258
> URL: https://issues.apache.org/activemq/browse/AMQNET-258
> Project: ActiveMQ .Net
> Issue Type: Bug
> Components: NMS
> Affects Versions: 1.3.0
> Environment: Client side Windows 2003, NMS 1.3.0.1959, ActiveMQ
> Client 1.3.0.1965
> Broker ActiveMQ 5.3.1-fuse-01-00, Redhat Linux 64 bit
> Reporter: Jonathan Anstey
> Assignee: Timothy Bish
> Attachments: activemq1.xml, activemq2.xml, nms-client.log,
> nms-client.zip
>
>
> So the issue is that I'm getting "Unknown response ID" errors in my NMS
> client log. I've attached the client as nms-client.zip and the log as
> nms-client.log. I was running this client on a Windows server 2003 system
> though any Windows box should do.
> The client is connecting with a failover URI like
> failover:(tcp://mendoza:61616?keepAlive=true&wireFormat.maxInactivityDuration=0,tcp://mendoza:61617?keepAlive=true&wireFormat.maxInactivityDuration=0)?initialReconnectDelay=100&randomize=false&timeout=15000
> The mendoza Linux machine has two AMQ brokers in M/S mode running on
> different ports. Broker configs are attached. The M/S is set up using KahaDB
> pointing to a NFS share on another Linux box. FYI to mount the share on
> mendoza I ran the following (pedro is another Linux box):
> sudo mount -o soft,intr,rsize=8192,wsize=8192 pedro:/local1 /mnt/pedro/local1
> To get this whole set up running, you'll need to change over the brokerUri in
> the client Program.cs to whatever host you are using and change the brokers
> to point to whatever NFS share you have mounted. To reproduce the failure, I
> had this running overnight but could see failures much earlier when I started
> taking down one broker at a time to start the failover process.
> Let me know if you need any more info.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
