| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| can't find reference on how MCAs read from S.C.X.Q |
« View previous topic :: View next topic » |
| Author |
Message
|
| flaufer |
 Posted: Tue Oct 12, 2010 5:01 am Post subject: can't find reference on how MCAs read from S.C.X.Q |
 |
|
Acolyte
Joined: 08 Dec 2004
Posts: 59
|
Folks,
question about clusters... I've heard many doubts that one message in the S.C.X.T to a destination queue manager which is unavailable would block all the messages behind it until the first message again can be moved again.
From my past experience I know this is not true and one cluster queue manager which is temporarily unavailable will NOT block messages for other queue managers in the cluster.
From reading the forum here I gathered that the MCA reads the message by correl-ID and picks out only the ones relevant for the channel/qmgr it has to care about, so the others are untouched and available for other MCAs to pick up.
However.. I can't find any doc on the IBM page that explains this behaviour.. to provide some "written proof" that will take out this doubt from the head of the people I talk to :)
Anybody here knows where I can read something about how a MCA handling a CLUSSDR channel deals with messages in the S.C.X.Q?
Cheers,
Felix |
|
| Back to top |
|
 |
| mvic |
| Posted: Tue Oct 12, 2010 4:08 pm Post subject: Re: can't find reference on how MCAs read from S.C.X.Q |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
A CLUSSDR channel gets by correl id for its channel name.
If there are 1000s of 10-byte messages for channel X, and that channel is moving a 10 Mb message, the 10-byte messages wait behind the 10 Mb message.
Message ordering must be preserved, and this is the right behaviour.
If you want multiple CLUSSDR channels you can do that (NB. you define CLUSRCVRs not CLUSSDRs..!), but they have to have different names from each other. Then you have to cope with any "workload balancing" that MQ does for you.
Can't help you with a ref to a manual page, sorry.
Regardless of the manual page, do you actually have a problem? |
|
| Back to top |
|
|
| flaufer |
| Posted: Tue Oct 12, 2010 11:19 pm Post subject: Re: can't find reference on how MCAs read from S.C.X.Q |
|
|
Acolyte
Joined: 08 Dec 2004
Posts: 59
|
| mvic wrote: |
A CLUSSDR channel gets by correl id for its channel name.
If there are 1000s of 10-byte messages for channel X, and that channel is moving a 10 Mb message, the 10-byte messages wait behind the 10 Mb message.
Message ordering must be preserved, and this is the right behaviour.
Regardless of the manual page, do you actually have a problem? |
Thanks for the explanation... that is what I have expected.
I do not have a technical problem, however have to fight against statemens like "clustering is bad because one unavailable queue manager would keep the cluster from working, because SCXQ is FIFO and only the first message is moved" in an environment with thousands of transmit queues where it actually does not make sense NOT to switch to clustering.
So... I know now how it really works, as mentioned by other board members earlier... however... any statement from IBM in regard to this would be very helpful. Probably I'll just open a PMR and let IBM confirm this.
Felix |
|
| Back to top |
|
|
| WMBDEV1 |
| Posted: Wed Oct 13, 2010 12:28 am Post subject: Re: can't find reference on how MCAs read from S.C.X.Q |
|
|
Sentinel
Joined: 05 Mar 2009
Posts: 888
Location: UK
|
| flaufer wrote: |
have to fight against statemens like "clustering is bad because one unavailable queue manager would keep the cluster from working, because SCXQ is FIFO and only the first message is moved"
|
A simple demo should be easy to setup disproving this statement to help the blind to see  |
|
| 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
|
|
|
|
