| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| How do HTTP Input/Reply nodes work internally? |
View previous topic :: View next topic |
| Author |
Message
|
| murdeep |
 Posted: Thu Jun 22, 2006 8:25 am Post subject: How do HTTP Input/Reply nodes work internally? |
 |
|
Master
Joined: 03 Nov 2004
Posts: 211
|
Hi, I have some doubts on how WMB http nodes work. I've searched the web for details but haven't had much luck. Here is what I have discovered.
Each HTTP Input node monitors the arrival of requests associated to specific URLs.
WMB has an internal configuration file (or an internal WMQ queue in WMB V6 Fix Pack1) that contains the list of URLs that are monitored by the different HTTP Input nodes in the different message flows deployed in any of the execution groups.
biphttplistener accesses this internal file (or queue) to match the URLs received in HTTP requests with the URLs that the HTTP Input nodes that are waiting for. The configuration file (or queue) has a unique correlation ID associated with each URL.
The HTTP Input node uses this correlation ID to get the messages from the internal queue (“SYSTEM.BROKER.WS.INPUT”) and the biphttplistener uses the same correlation ID to put the messages on this queue. This is the mechanism used to correlate the incoming HTTP requests and the HTTP Input nodes in the message flows.
The HTTP Input node copies the MQ input Message ID in the LocalEnvironment.Destination.HTTP.RequestIdentifier to be used by the HTTP Reply node for MQ output message Correlation ID. The biphttplistener does a selective get by correlation ID to the reply
queue (“SYSTEM.BROKER.WS.REPLY”) and it receives the response messages.
Is this correct?
I see messages building on SYSTEM.BROKER.WS.INPUT when the flow doesn't reach a HTTP reply node. The messages on WS.INPUT have an expiry time (300 miuntes - what sets this expiry time?) the WS.INPUT queue is defined with NPMCLASS(normal) so this means that they will be purged on qmgr restart. But what if I cycle the broker. Will messages already on WS.INPUT be reporcessed? Also, does the Dataflowengine only browse messages on WS.INPUT? If so, why wouldn't it do a GET under sysncpoint? |
|
| Back to top |
|
 |
| vennela |
| Posted: Thu Jun 22, 2006 1:42 pm Post subject: |
|
|
Jedi Knight
Joined: 11 Aug 2002
Posts: 4055
Location: Hyderabad, India
|
| Quote: |
| I see messages building on SYSTEM.BROKER.WS.INPUT when the flow doesn't reach a HTTP reply node. The messages on WS.INPUT have an expiry time (300 miuntes - what sets this expiry time?) |
300 minutes is awfully long for a HTTP request given the fact that HTTP is a synchronous protocol. Most likely this is in seconds. Maybe it is the time that is specified on the HTTPInput node.
| Quote: |
| But what if I cycle the broker. Will messages already on WS.INPUT be reporcessed? |
Again, I would expect after a broker recycle, a broker most likely will not process any requests that were made before it was started.
These are just my thoughts and I have no idea of what's going on inside though. |
|
| Back to top |
|
|
| elvis_gn |
| Posted: Thu Jun 22, 2006 9:33 pm Post subject: Re: How do HTTP Input/Reply nodes work internally? |
|
|
Padawan
Joined: 08 Oct 2004
Posts: 1905
Location: Dubai
|
Hi murdeep,
I agree with vennela
| murdeep wrote: |
| I see messages building on SYSTEM.BROKER.WS.INPUT when the flow doesn't reach a HTTP reply node |
This I cannot understand...the HTTPInput node should have picked up all the messages UNLESS the Reply node was within the same flow...
You can have the reply node in another flow and provide the HTTP reply parameters...this way all ur messages from the queue will be picked...
Regards. |
|
| Back to top |
|
|
|
|
|
|
|
Page 1 of 1 |
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
