Author |
Message
|
ashtek |
Posted: Wed Dec 21, 2011 7:48 am Post subject: Clustering MQSeries Server for Windows |
|
|
Newbie
Joined: 21 Dec 2011 Posts: 4
|
We are going to install MQSeries server for windows on two VMs (windows 2008 r2). I want to know how to achieve failover automatically by clustering the MQs on these two VMs so if VM1 goes down, other applications will continue sending messages to MQ on VM2 without doing any changes.
Can we do this without windows clustering (MSCS)? If yes, I would like to know what configurations needs to be done to achieve this? (I'm new to MQ and need details on how to cluster MQ, whether we need to cluster queue managers, etc) |
|
Back to top |
|
 |
Vitor |
Posted: Wed Dec 21, 2011 7:56 am Post subject: Re: Clustering MQSeries Server for Windows |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
ashtek wrote: |
I want to know how to achieve failover automatically by clustering the MQs on these two VMs so if VM1 goes down, other applications will continue sending messages to MQ on VM2 without doing any changes. |
Do you mean "sending" or do you mean "processing"? You can have an application using a client connection (which yours may or may not be) send to whichever queue manager is up, but messages on a downed queue manager are stuck.
ashtek wrote: |
Can we do this without windows clustering (MSCS)? |
Maybe. See above.
ashtek wrote: |
If yes, I would like to know what configurations needs to be done to achieve this? (I'm new to MQ and need details on how to cluster MQ, whether we need to cluster queue managers, etc) |
If you'd looked through this potentially misleading forum section you'd have found endless comments from myself and others pointing out that "clustering" has 2 meanings:
a) WMQ Clustering, used for load balancing
b) HA Clustering, used by MSCS and similar to provide failover
Do not confuse the 2.  _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
ashtek |
Posted: Wed Dec 21, 2011 8:10 am Post subject: Re: Clustering MQSeries Server for Windows |
|
|
Newbie
Joined: 21 Dec 2011 Posts: 4
|
Hi Vitor, Thanks for your reply.
Vitor wrote: |
ashtek wrote: |
I want to know how to achieve failover automatically by clustering the MQs on these two VMs so if VM1 goes down, other applications will continue sending messages to MQ on VM2 without doing any changes. |
Do you mean "sending" or do you mean "processing"? You can have an application using a client connection (which yours may or may not be) send to whichever queue manager is up, but messages on a downed queue manager are stuck.
Actually we receive messages from another MQ which is on non-windows.
And we have one box with 2 VMs where we'll install MQSeries server for windows.
If one VM goes down, we should have MQ on other VM receive and process the message (and also send to remote MQ) automatically. This means remote MQ should use single IP but we'll have two MQs on VM1 & VM2 (MQ on VM2 should work if VM1 goes down). Any idea what should be done to achieve this?
ashtek wrote: |
Can we do this without windows clustering (MSCS)? |
Maybe. See above.
ashtek wrote: |
If yes, I would like to know what configurations needs to be done to achieve this? (I'm new to MQ and need details on how to cluster MQ, whether we need to cluster queue managers, etc) |
If you'd looked through this potentially misleading forum section you'd have found endless comments from myself and others pointing out that "clustering" has 2 meanings:
a) WMQ Clustering, used for load balancing
b) HA Clustering, used by MSCS and similar to provide failover
Do not confuse the 2.  |
My requirement is to achieve automatic failover (between VM1 & VM2) and I think we need to do MQ clustering for this? |
|
Back to top |
|
 |
Vitor |
Posted: Wed Dec 21, 2011 8:17 am Post subject: Re: Clustering MQSeries Server for Windows |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
ashtek wrote: |
My requirement is to achieve automatic failover (between VM1 & VM2) and I think we need to do MQ clustering for this? |
Vitor wrote: |
If you'd looked through this potentially misleading forum section you'd have found endless comments from myself and others pointing out that "clustering" has 2 meanings:
a) WMQ Clustering, used for load balancing
b) HA Clustering, used by MSCS and similar to provide failover
|
So not only have you not looked through this section for useful discussions, you're not reading my reply either.
 _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
ashtek |
Posted: Wed Dec 21, 2011 8:17 am Post subject: Re: Clustering MQSeries Server for Windows |
|
|
Newbie
Joined: 21 Dec 2011 Posts: 4
|
Just to add, infrastructure team wants to avoid MSCS clustering as far as possible. |
|
Back to top |
|
 |
fjb_saper |
Posted: Wed Dec 21, 2011 8:34 am Post subject: Re: Clustering MQSeries Server for Windows |
|
|
 Grand High Poobah
Joined: 18 Nov 2003 Posts: 20756 Location: LI,NY
|
ashtek wrote: |
Just to add, infrastructure team wants to avoid MSCS clustering as far as possible. |
Look up the term Multi Instance Qmgr.  _________________ MQ & Broker admin |
|
Back to top |
|
 |
JasonE |
Posted: Wed Dec 21, 2011 9:11 am Post subject: |
|
|
Grand Master
Joined: 03 Nov 2003 Posts: 1220 Location: Hursley
|
High level summary
Clustering: QM1 and QM2 are both active, and if QM1 dies any messages on one of its queues is therefore unavailable.
MSCS: QM1 is running on machine 1, and in case of problems both its IP address and qmgr + data is moved onto another machine. It can also automatically switch back should another failure occur.
Multiinstance: QM1 is running on machine 1 and in standby on machine 2, with data on shared disk, and in case of failure it will start on machine 2 but under a different ip address to machine 1. It will also involve manual input then to be able to switch back. |
|
Back to top |
|
 |
ashtek |
Posted: Wed Dec 28, 2011 5:22 pm Post subject: |
|
|
Newbie
Joined: 21 Dec 2011 Posts: 4
|
Thank you Jason.
JasonE wrote: |
Multiinstance: QM1 is running on machine 1 and in standby on machine 2, with data on shared disk, and in case of failure it will start on machine 2 but under a different ip address to machine 1. It will also involve manual input then to be able to switch back. |
In this case, is it possible to have a virtual IP assigned to MQ so when QM1 starts on machine2, IP will remain the same? |
|
Back to top |
|
 |
fjb_saper |
Posted: Wed Dec 28, 2011 7:23 pm Post subject: |
|
|
 Grand High Poobah
Joined: 18 Nov 2003 Posts: 20756 Location: LI,NY
|
ashtek wrote: |
Thank you Jason.
JasonE wrote: |
Multiinstance: QM1 is running on machine 1 and in standby on machine 2, with data on shared disk, and in case of failure it will start on machine 2 but under a different ip address to machine 1. It will also involve manual input then to be able to switch back. |
In this case, is it possible to have a virtual IP assigned to MQ so when QM1 starts on machine2, IP will remain the same? |
You probably can, but that would require a networking device with that capability. (additional hardware).  _________________ MQ & Broker admin |
|
Back to top |
|
 |
JasonE |
Posted: Tue Jan 03, 2012 2:31 am Post subject: |
|
|
Grand Master
Joined: 03 Nov 2003 Posts: 1220 Location: Hursley
|
Quote: |
Multiinstance: QM1 is running on machine 1 and in standby on machine 2, with data on shared disk, and in case of failure it will start on machine 2 but under a different ip address to machine 1. It will also involve manual input then to be able to switch back.
Quote: |
In this case, is it possible to have a virtual IP assigned to MQ so when QM1 starts on machine2, IP will remain the same? |
|
[/quote]
Not normally, that's why things were changed to provide the clients with the ability of having multiple destinations on their connames, and autoreconnect etc. You really need to read the docs.. e.g.
http://publib.boulder.ibm.com/infocenter/wmqv7/v7r0/topic/com.ibm.mq.amqzag.doc/fa70155_.htm |
|
Back to top |
|
 |
|