| Author |
Message
|
| jmac |
 Posted: Wed Nov 24, 2004 3:09 pm Post subject: Anyone running MQWF in an High Availability environment? |
 |
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
I am trying to understand how this works without much success, I have looked at support pack WC61 and do not find the information I am looking for.
Say I had a 3-tier setup DB2 running on BoxDB2, primary MQWF System running on BoxMQWF1, and secondary MQWF System running on BoxMQWF2. It seems to me that if BoxMQWF1 were to fail the secondary system could handle things except for the Scheduling server (and Cleanup Server). Perhaps I have a major misconception as to how this works, but I was intially under the impression that BoxMQWF2 simply became the primary, but without a Scheduling server, most MQWF productions systems would not function properly.
Anyone with any experience in this area, please let me know how you handle this.
Thanks
_________________
John McDonald
RETIRED |
|
| Back to top |
|
 |
| Andy |
| Posted: Wed Nov 24, 2004 9:27 pm Post subject: |
|
|
Centurion
Joined: 14 May 2003
Posts: 122
|
I have not seen support pack but as far as I know there are few scripts which runs on secodary when failover takes place.
We have our own script which just start all the services(DB2,WAS,MQWF) on boxMQWF2 when HACMP senses boxMQWF1 is failed. Scheduling server and Cleanup server are not a problem as MQWF is started just now.
The disadvantage is server becomes unavailable for few minutes during IP take over by boxMQWF2 and current user sessions become invalid.
_________________
 Andy |
|
| Back to top |
|
|
| jmac |
| Posted: Thu Nov 25, 2004 5:15 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
Andy:
So what you are saying is that your failover system is a HOT standby, it is not processing as part of the normal workload. Is that correct?
Thanks
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| Andy |
| Posted: Thu Nov 25, 2004 10:45 am Post subject: |
|
|
Centurion
Joined: 14 May 2003
Posts: 122
|
yes thats correct. I'll be surprised if it works the way you have described in original post. Lets see what others have to say!!
thanks
_________________
Andy |
|
| Back to top |
|
|
| manoj |
| Posted: Tue Nov 30, 2004 10:51 am Post subject: |
|
|
Master
Joined: 30 Jan 2002
Posts: 237
Location: Virgina
|
John,
In my case we had a similar setup as what you have described ( also No HACMP). I have also observed that when primary goes down the entire MQ Workflow environment is down irrespective of SYSTEM2 is running.
Having said, i have to remaind you that in my case there was no Second Cluster repository on SYSTEM2 (first repository is on SYSTEM1). The Second repository is on a machine running Websphere and MQ Server.
_________________
-manoj |
|
| Back to top |
|
|
| sac |
| Posted: Tue Dec 07, 2004 3:08 am Post subject: |
|
|
Apprentice
Joined: 10 Feb 2003
Posts: 44
|
Going by the concept of queue managers in cluster, i understand that Websphere MQ should be deciding about message delivery i.e message should be delivered to which queue and of which queue manager based on round robin algorithm or anyother. if we go as per this concept then what i understand is that multiple Systems under a System Group are should be actually sharing the Workflow workload among themselves.
As far as high availability is concerened, i am not sure about it, but what i understand about it is that in this kind of environment we have backup resource available for each component (WAS / MQ Server / Mq Workflow).
i.e if MQ Sever goes down, then you have the secondary MQ Server as secondary queue manager in cluster available in your system archietecture. If one WAS system goes down then you have a another WAS instance running in cluster.
But one thing that really does worry is what if Primary Workflow goes down as stated in first query. That is something which does pops question on high availabilty. |
|
| Back to top |
|
|
|
|
