Author |
Message
|
Sandman |
Posted: Sun Oct 14, 2012 10:42 am Post subject: What happens to publications when error delivering? |
|
|
Centurion
Joined: 16 Oct 2001 Posts: 134 Location: Lincoln, RI
|
If a message flow that sends a message to a Publication node doesn't specify message type of request and a replyToQueue, and the broker fails when trying to send the message (i.e. queue manager not available), is the pub message just discarded?
[edited spelling error]
Last edited by Sandman on Mon Oct 15, 2012 5:43 am; edited 2 times in total |
|
Back to top |
|
 |
Vitor |
Posted: Sun Oct 14, 2012 12:17 pm Post subject: Re: What happens to publications when error delivering? |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
Sandman wrote: |
If a message flow that sends a message to a Publication node doesn't specify messasge type of request and a replyToQueue, and the broker fails when trying to send the message (i.e. queue manager not available), is the pub message just discarded?  |
Why would you think the flow would perform what it does normally when it encounters an error? _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
mqjeff |
Posted: Sun Oct 14, 2012 1:11 pm Post subject: Re: What happens to publications when error delivering? |
|
|
Grand Master
Joined: 25 Jun 2008 Posts: 17447
|
Vitor wrote: |
Sandman wrote: |
If a message flow that sends a message to a Publication node doesn't specify messasge type of request and a replyToQueue, and the broker fails when trying to send the message (i.e. queue manager not available), is the pub message just discarded?  |
Why would you think the flow would perform what it does normally when it encounters an error? |
I believe you forgot a "not" in there somewhere. |
|
Back to top |
|
 |
Sandman |
Posted: Mon Oct 15, 2012 4:12 am Post subject: |
|
|
Centurion
Joined: 16 Oct 2001 Posts: 134 Location: Lincoln, RI
|
I wouldn't expect the flow to throw an error back to the MQInput's catch terminal because the flow didn't fail; the Publication node DID accept the published message - it's just that the broker failed satisfying one of its subscriptions.
Am I wrong in assuming that the actual delivery to subscribers is in a separate unit of work (beyond the pub flow)? |
|
Back to top |
|
 |
nathanw |
Posted: Mon Oct 15, 2012 4:42 am Post subject: |
|
|
 Knight
Joined: 14 Jul 2004 Posts: 550
|
think it would depend on the version of broker and MQ you are using _________________ Who is General Failure and why is he reading my hard drive?
Artificial Intelligence stands no chance against Natural Stupidity.
Only the User Trace Speaks The Truth  |
|
Back to top |
|
 |
Vitor |
Posted: Mon Oct 15, 2012 4:42 am Post subject: |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
Sandman wrote: |
I wouldn't expect the flow to throw an error back to the MQInput's catch terminal because the flow didn't fail; the Publication node DID accept the published message - it's just that the broker failed satisfying one of its subscriptions. |
That's not what you said. You asked about the publication failing. If the broker (or in v7 the queue manager) accepts a message for publication that's one thing. If the queue manager can't then propagate the publication to a subscriber that's another, and brings you into the world of assured delivery as with a point to point message.
Sandman wrote: |
Am I wrong in assuming that the actual delivery to subscribers is in a separate unit of work (beyond the pub flow)? |
It's in a separate unit of work and operated by a different process. Remember that for any publication it's then necessary to match all the potential subscribers to the topic of the new publication before any message movement can take place.
So to answer the question I think you originally intended to ask, a publication isn't lost in the event of a queue manager being unavailable any more than a message on a transmission queue is lost in the same circumstances. Or to the same extent a message is lost in the same circumstances. _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
Sandman |
Posted: Mon Oct 15, 2012 5:46 am Post subject: |
|
|
Centurion
Joined: 16 Oct 2001 Posts: 134 Location: Lincoln, RI
|
I didn't think I had "asked about the Publication failing" when I said "is the pub message just discarded?". But I can see how that could be interpreted as such.
By "pub message", I meant, after successful call to Publication node, the message that the Broker then tries to deliver to subscribers. Perhaps I should've said "subscriber message" to be clearer?
At any rate, if the Broker couldn't deliver the message to a subscriber whose qmgr wasn't available, it would just sit on the transmission queue is what you're saying? |
|
Back to top |
|
 |
Vitor |
Posted: Mon Oct 15, 2012 6:02 am Post subject: |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
Sandman wrote: |
Perhaps I should've said "subscriber message" to be clearer? |
Or "subscription". But comes to the same thing.
Sandman wrote: |
At any rate, if the Broker couldn't deliver the message to a subscriber whose qmgr wasn't available, it would just sit on the transmission queue is what you're saying? |
As indicated above, the mechanics of this changes significantly between WMBv6 & WMBv7 but yes. _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
Sandman |
Posted: Mon Oct 15, 2012 6:31 am Post subject: |
|
|
Centurion
Joined: 16 Oct 2001 Posts: 134 Location: Lincoln, RI
|
|
Back to top |
|
 |
nathanw |
Posted: Mon Oct 15, 2012 7:28 am Post subject: |
|
|
 Knight
Joined: 14 Jul 2004 Posts: 550
|
so this would be in Broker unit of work only as in 6.1 pub sub was handled by Broker.
I believe the answer to this if the target or subscribing queue does not exist then the message is written to the dead letter queue.
The answer is always to ensure that the subscribing queue is always created. _________________ Who is General Failure and why is he reading my hard drive?
Artificial Intelligence stands no chance against Natural Stupidity.
Only the User Trace Speaks The Truth  |
|
Back to top |
|
 |
mqjeff |
Posted: Mon Oct 15, 2012 7:42 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008 Posts: 17447
|
|
Back to top |
|
 |
|