| Author |
Message
|
| sebastia |
 Posted: Wed Sep 27, 2006 5:49 am Post subject: how does RUNMQTMC work ? |
 |
|
Grand Master
Joined: 07 Oct 2004
Posts: 1003
|
Can you help me to understand RUNMQTMC a little bit more ?
On the Server machine, lets supose we have
Queue Manager Name := QMN ;
Data Queue := DATA_Q ; // it is a Trigger queue, with ...
Initialization Queue := INIT_Q ;
and Process Definition := ABC.exe ;
On the Client machine, the configuration is :
SET MQSERVER=MY.SVRCONN/TCP/1.2.3.4(1490)
and we start the Monitor using (on the Client machine ) :
RUNMQTMC -m QMN -q INIT_Q
? Am I correct to say that putting a data message in DATA_Q on the Server machine will produce the code ABC.exe to be started on the Client machine ?
I guess so.
My question is :
RUNMQTMC did (only) use MY.SVRCONN channel to do its work ?
This is then, my SINGLE POINT OF FAILURE ??
Any (positive) comment is Welcome !
( )
By the way : can I have/get the RUMQTMC source code ?
I am wondering if it uses any Timeout to wait on the Init queue ....
. |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Wed Sep 27, 2006 6:00 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
The job of a trigger monitor is to read messages from the initiation queue and start a process.
RUNMQTMC works just the same as RUNMQTRM, except that it uses a client connection instead of a server connection... 
The process definition is treated as a local process in both cases. So if you put in a path like "/bin/echo", it will run the local /bin/echo, and not the /bin/echo on some other machine. So if you use runmqtrm, it will run the /bin/echo on the MQ server, and if you use runmqtmc it will run the bin/echo on the client machine and NOT on the server machine.
source for runmqtrm/runmqtmc is not available. There is sample trigger monitor code included with the product, but it's not quite the same as what's compiled into runmqtrm/runmqtmc. I don't remember if runmqtrm/tmc uses a get with wait or not. It's mostly irrelevant.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| sebastia |
| Posted: Wed Sep 27, 2006 6:20 am Post subject: |
|
|
Grand Master
Joined: 07 Oct 2004
Posts: 1003
|
OK Jeff - I do agree with you few of the facts are NOT relevant.
.
Let me go into my "real" problem : it is High Availability.
.
If customer uses MQ Client to receive messages, and Client machine A goes down, what shall I do the keep the system running ?
.
First idea is to have some "IP balancer" (sorry I dont know the good english word, but this hardware detects machine A is down and turns machine B on, that has to replace its function )
.
On client machine "B" we think to start the SAME trigger monitor with the same parameters, same application.
>>> DO YOU THINK THIS CONFIG SHALL WORK ??? >>>
Cheers. Sebastian. |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Wed Sep 27, 2006 6:27 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
I guess I tried to answer the question I thought you were asking.
Yes, what you describe should work - you can include a trigger monitor in a HA failover group.
The trigger monitor is really just another MQ application - it reads messages off a queue and processes them - in this case the processing is to start another program to process another queue.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| jeevan |
| Posted: Wed Sep 27, 2006 6:30 am Post subject: |
|
|
Grand Master
Joined: 12 Nov 2005
Posts: 1432
|
| One of the solutions to this problem is to have a Veritas cluster setup of the clients machine so if one machine fails, the second takes up from the same point. It uses SAN storage systems so everythings is the same so it reduces the burden ofcopying data or starting the another machine. the Veritas Clustering does all automatically. Once the first machine is up and running, it resumes it work automatically. |
|
| Back to top |
|
|
| sebastia |
| Posted: Wed Sep 27, 2006 6:41 am Post subject: |
|
|
Grand Master
Joined: 07 Oct 2004
Posts: 1003
|
mr JeeVan : do you have any pointer/url to any "Veritas cluster" info ?
How do you make sure
the second machine takes the job "FROM THE SAME POINT" ???
Sebastian. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Wed Sep 27, 2006 4:33 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| sebastia wrote: |
mr JeeVan : do you have any pointer/url to any "Veritas cluster" info ?
How do you make sure
the second machine takes the job "FROM THE SAME POINT" ???
Sebastian. |
Because that's what a hardware cluster is build for.
The documentation says that both instances use the same shared disk. That's the way that you know the second machine takes the job from the same point. (at least if you've configured it right...) 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| jeevan |
| Posted: Thu Sep 28, 2006 1:39 pm Post subject: |
|
|
Grand Master
Joined: 12 Nov 2005
Posts: 1432
|
|
| Back to top |
|
|
|
|
