Author |
Message
|
mqrules |
Posted: Wed Jul 01, 2009 6:56 am Post subject: What happens to cluster queues on FR? |
|
|
Centurion
Joined: 01 Jun 2005 Posts: 100 Location: US
|
I have to remove the FR (full repository) (On Linux MQ 5.3) and rejoin it to the cluster and make it a full repository again. There are cluster queues on this FR. My question is, can I keep the cluster queue defs as they are (i.e. without making them non-cluster and then re-cluster)?
Thx.
MR |
|
Back to top |
|
 |
mqrules |
Posted: Wed Jul 01, 2009 8:37 am Post subject: |
|
|
Centurion
Joined: 01 Jun 2005 Posts: 100 Location: US
|
Let me put the same question in a different way:
The clusqmgr QM1 will be removed from the cluster TEST1 (following the steps in the manual). ANd then it will be joined to the cluster again. My question is, do I need to do the following on the cluster queues:
1- Alter q(xxxx) CLUSTER(' ')
2- Remove the qmgr from the cluster TEST1
3- Add the qmgr to the cluster TEST1
4- Alter q(xxxx) CLUSTER('TEST1')
I don't think the cluster queues need to be altered to make them non-cluster and then cluster queues What do you think?
Your response would be much appreaciated.
Thanks.
MR |
|
Back to top |
|
 |
shashivarungupta |
Posted: Wed Jul 01, 2009 9:08 am Post subject: |
|
|
 Grand Master
Joined: 24 Feb 2009 Posts: 1343 Location: Floating in space on a round rock.
|
mqrules wrote: |
Let me put the same question in a different way:
The clusqmgr QM1 will be removed from the cluster TEST1 (following the steps in the manual). ANd then it will be joined to the cluster again. My question is, do I need to do the following on the cluster queues:
1- Alter q(xxxx) CLUSTER(' ')
2- Remove the qmgr from the cluster TEST1
3- Add the qmgr to the cluster TEST1
4- Alter q(xxxx) CLUSTER('TEST1')
I don't think the cluster queues need to be altered to make them non-cluster and then cluster queues What do you think?
|
Quote: |
The clusqmgr QM1 will be removed from the cluster TEST1 (following the steps in the manual). ANd then it will be joined to the cluster again. |
Do you mean suspend ?
Whats your objective to do that ? _________________ *Life will beat you down, you need to decide to fight back or leave it. |
|
Back to top |
|
 |
shashivarungupta |
Posted: Wed Jul 01, 2009 9:25 am Post subject: |
|
|
 Grand Master
Joined: 24 Feb 2009 Posts: 1343 Location: Floating in space on a round rock.
|
If you remove the queue manager from the cluster, it means it'll be outside of that cluster and would be the part of distributed architecture.
As to follow the best practices, to do THAT you'd remove its cluster components as cluster channels. And when it comes to best practices then you should unshare those queues from the cluster. _________________ *Life will beat you down, you need to decide to fight back or leave it. |
|
Back to top |
|
 |
shashivarungupta |
Posted: Wed Jul 01, 2009 9:32 am Post subject: |
|
|
 Grand Master
Joined: 24 Feb 2009 Posts: 1343 Location: Floating in space on a round rock.
|
This Post should be the part of the Cluster Forum. Isn't it ?  _________________ *Life will beat you down, you need to decide to fight back or leave it. |
|
Back to top |
|
 |
mqrules |
Posted: Wed Jul 01, 2009 10:59 am Post subject: |
|
|
Centurion
Joined: 01 Jun 2005 Posts: 100 Location: US
|
Quote: |
If you remove the queue manager from the cluster, it means it'll be outside of that cluster and would be the part of distributed architecture.
As to follow the best practices, to do THAT you'd remove its cluster components as cluster channels. And when it comes to best practices then you should unshare those queues from the cluster.
|
shashivarungupta, thanks for your reply but you seem to be missing the point here. The qmgr QM1 will be ADDED back to the cluster after being removed. I don't want to get into toomuch detail why we want to do it. So, I think that the CLUSTER attrib of the cluster-queues do not need to be altered in this process. I think, once the qmgr is re-joined the cluster the cluster-queues will work fine.
Can someone please correct me if I am wrong with my assumption? |
|
Back to top |
|
 |
Vitor |
Posted: Wed Jul 01, 2009 2:07 pm Post subject: |
|
|
 Grand High Poobah
Joined: 11 Nov 2005 Posts: 26093 Location: Texas, USA
|
mqrules wrote: |
So, I think that the CLUSTER attrib of the cluster-queues do not need to be altered in this process. I think, once the qmgr is re-joined the cluster the cluster-queues will work fine. |
I think you might be right. I think you might be wrong. I think you might get a lot of "cluster not found" errors in the queue manager log.
I think a simple experiment would answer all these questions and more. _________________ Honesty is the best policy.
Insanity is the best defence. |
|
Back to top |
|
 |
bruce2359 |
Posted: Wed Jul 01, 2009 2:09 pm Post subject: |
|
|
 Poobah
Joined: 05 Jan 2008 Posts: 9469 Location: US: west coast, almost. Otherwise, enroute.
|
If you want to make a qmgr a PR, just take REPOS attribute of the qmgr blanks (spaces). Voila, it is no longer an FR; it is a PR.
Cluster queues will still be available to other qmgrs through the remaining FR.
I'm sure you don't want to give much detail as to why. I can't come up with a reason to make an FR a PR, unless you are briefly shutting down the qmgr (to move it, for example); and you plan to make some other qmgr an FR during the brief outage. _________________ I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
Back to top |
|
 |
shashivarungupta |
Posted: Wed Jul 01, 2009 9:43 pm Post subject: |
|
|
 Grand Master
Joined: 24 Feb 2009 Posts: 1343 Location: Floating in space on a round rock.
|
Vitor wrote: |
mqrules wrote: |
So, I think that the CLUSTER attrib of the cluster-queues do not need to be altered in this process. I think, once the qmgr is re-joined the cluster the cluster-queues will work fine. |
I think you might be right. I think you might be wrong. I think you might get a lot of "cluster not found" errors in the queue manager log.
I think a simple experiment would answer all these questions and more. |
It seems you heard my innersole.
By keeping the queue attributes as it is, when you are removing the QM from the cluster (and then including it in ) might not get errors, but you can't ignore the chances when you'll get some errors like that(as pointed by Vitor).
Because in a cluster if you have 2 repositories and it would have info. about the queue managers and its shared objects in cluster. So while removal of one, other can seek some objects that physically were defined on the removed QM. _________________ *Life will beat you down, you need to decide to fight back or leave it. |
|
Back to top |
|
 |
shashivarungupta |
Posted: Thu Jul 02, 2009 12:43 am Post subject: |
|
|
 Grand Master
Joined: 24 Feb 2009 Posts: 1343 Location: Floating in space on a round rock.
|
mqrules wrote: |
Quote: |
If you remove the queue manager from the cluster, it means it'll be outside of that cluster and would be the part of distributed architecture.
As to follow the best practices, to do THAT you'd remove its cluster components as cluster channels. And when it comes to best practices then you should unshare those queues from the cluster.
|
... but you seem to be missing the point here.... |
I have given the point above, that I was earlier pointing to... and you couldn't sense that.
 _________________ *Life will beat you down, you need to decide to fight back or leave it. |
|
Back to top |
|
 |
mqrules |
Posted: Thu Jul 02, 2009 4:11 pm Post subject: |
|
|
Centurion
Joined: 01 Jun 2005 Posts: 100 Location: US
|
I did the testing and I did not have to "alter" the CLUSTER attrib of the queues. I did not get any errors...
Thanks all for your responses.
MR |
|
Back to top |
|
 |
mqrules |
Posted: Thu Jul 02, 2009 4:23 pm Post subject: |
|
|
Centurion
Joined: 01 Jun 2005 Posts: 100 Location: US
|
I should add that, I had no access to my computer or any server to test , and that is why I posed the question to the forum. Thanks again. MR |
|
Back to top |
|
 |
bruce2359 |
Posted: Thu Jul 02, 2009 4:46 pm Post subject: |
|
|
 Poobah
Joined: 05 Jan 2008 Posts: 9469 Location: US: west coast, almost. Otherwise, enroute.
|
Quote: |
So while removal of one, other can seek some objects that physically were defined on the removed QM. |
The point of suspending the FR (any cluster qmgr, acutally) is to tell other repositories (both FR and PR) to stop using the suspended qmgr as a destination for subsequent application mqopens/mqput1's. _________________ I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
Back to top |
|
 |
|