Author |
Message
|
rknox |
Posted: Wed Jun 18, 2008 11:21 am Post subject: Tracing message through Broker |
|
|
Apprentice
Joined: 28 Aug 2007 Posts: 34
|
We are moving from ICS to Message Broker. In ICS, we could turn up trace on the connector controllers to actually see the contents of the message before it went into ICS and as it left ICS which greatly helped troubleshooting. This could all be done dynamically.
I have not been able to find an administrative way to perform this action on Message broker dynamically as it is done in ICS. Can someone here point me in the right direction? |
|
Back to top |
|
 |
Gaya3 |
Posted: Wed Jun 18, 2008 9:31 pm Post subject: |
|
|
 Jedi
Joined: 12 Sep 2006 Posts: 2493 Location: Boston, US
|
there is nothing called connector controllers in MB.
you can turn up trace or turn off trace in execution level of broker.
troubleshooting is easier in MB too. _________________ Regards
Gayathri
-----------------------------------------------
Do Something Before you Die |
|
Back to top |
|
 |
sri_csee1983 |
Posted: Wed Jun 18, 2008 11:54 pm Post subject: |
|
|
 Centurion
Joined: 25 Mar 2008 Posts: 125 Location: Chennai,India
|
It will be more easy if u add the trace node and route that to a file whereever in the flow u needed.  _________________ With Cheers,
Sri |
|
Back to top |
|
 |
mqmatt |
Posted: Thu Jun 19, 2008 12:26 am Post subject: |
|
|
 Grand Master
Joined: 04 Aug 2004 Posts: 1213 Location: Hursley, UK
|
I'd certainly recommend trace nodes, as sri_csee1983 says.
Also, with v6.1.0.2 you can audit messages by having the broker publish events as messages enter and leave the broker. These events can be monitored (for example) by WebSphere Business Monitor. |
|
Back to top |
|
 |
rknox |
Posted: Thu Jun 19, 2008 4:53 am Post subject: |
|
|
Apprentice
Joined: 28 Aug 2007 Posts: 34
|
I was under the impression that putting trace nodes into production flows is bad practice since they cause so much overhead in processing. Is this inaccurate? |
|
Back to top |
|
 |
SAFraser |
Posted: Thu Jun 19, 2008 3:19 pm Post subject: |
|
|
 Shaman
Joined: 22 Oct 2003 Posts: 742 Location: Austin, Texas, USA
|
We use trace nodes pretty routinely in production and have found no noticeable impact on performance. We don't have a huge number of transcactions, though, maybe 100,000/day through each flow.
The bigger problem is managing the log files that are outputted, which we do with cron jobs that archive them.
The trace nodes were originally implemented in production because the application people were sure MQ would lose their messages. (I know, that's silly.) We do use them fairly frequently to look for data that has apparently been "lost" by the applications.
On some flows, we have two versions: one with trace nodes and one without. We'll swap them out when troubleshooting is needed. |
|
Back to top |
|
 |
PeterPotkay |
Posted: Thu Jun 19, 2008 5:34 pm Post subject: |
|
|
 Poobah
Joined: 15 May 2001 Posts: 7722
|
In WMB 6.1 you can turn Trace Nodes on and off without redeploying.  _________________ Peter Potkay
Keep Calm and MQ On |
|
Back to top |
|
 |
SAFraser |
Posted: Thu Jun 19, 2008 6:26 pm Post subject: |
|
|
 Shaman
Joined: 22 Oct 2003 Posts: 742 Location: Austin, Texas, USA
|
I have got to make time to look at 6.1. That is a fantastic feature. Thanks, Peter. |
|
Back to top |
|
 |
sri_csee1983 |
Posted: Thu Jun 19, 2008 7:43 pm Post subject: |
|
|
 Centurion
Joined: 25 Mar 2008 Posts: 125 Location: Chennai,India
|
Hi rknox,
If u feel like that then u can turn on the User trace just right clicking the flow and set to debug. U ell get the step by step execution of the flow which u can retrieve using mqsireadlog. And turn off the trace whenever not needed. But anyhow u shud have a proof that ur broker got the message and it has processed the message successfully. I am not sure whether u have any other means to prove that other than a trace node.  _________________ With Cheers,
Sri |
|
Back to top |
|
 |
|