|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
Clusters with 3 FR or more - Final |
« View previous topic :: View next topic » |
Author |
Message
|
PeterPotkay |
Posted: Mon May 12, 2008 6:40 pm Post subject: |
|
|
 Poobah
Joined: 15 May 2001 Posts: 7722
|
Well, I played with this. I had a cluster with 3 FRs. Based on my testing all I can say is that once a cluster has 2 or more FRs N/A any changes to your cluster definitions is going to very inconsistent. Even if a PR has manual CLUSSDRs to all the FRs.
When all the FRs were available things were working fine, with each PR having 1 manual CLUSSDR to 1 PR. Any change on a PR was seen immediately on all the FRs. All interested PRs saw the change too. But once 2 or more FRs were N/A all bets are off when it comes to changes.
Manual CLUSSDRs to all the FRs and then issuing the refresh cluster command while 2 of the FRs were down got me 2189 errors on that PR that I'm still trying to chase down.
I agree there could be more documentation on this subject in the manuals. But based on what I (think I) know, and on the current manuals, I am going to only run my clusters with 2 Highly Available FRs and every QM in my cluster will only have one manual CLUSSDR defined to one of the FRs. _________________ Peter Potkay
Keep Calm and MQ On |
|
Back to top |
|
 |
fjb_saper |
Posted: Mon May 12, 2008 7:40 pm Post subject: |
|
|
 Grand High Poobah
Joined: 18 Nov 2003 Posts: 20757 Location: LI,NY
|
Peter,
How fast do the errors resolve after all defined FRs are available again?
 _________________ MQ & Broker admin |
|
Back to top |
|
 |
Nigelg |
Posted: Mon May 12, 2008 10:06 pm Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004 Posts: 1046
|
Quote: |
Manual CLUSSDRs to all the FRs and then issuing the refresh cluster command while 2 of the FRs were down got me 2189 errors on that PR |
This will happen if there are 2 FRs and 1 is down when REFRESH CLUSTER is done. The reason is that the sub request is not answered by the FR that is not working, and 2189 is returned to an app opening a queue if not all sub requests have been replied to. _________________ MQSeries.net helps those who help themselves.. |
|
Back to top |
|
 |
sami.stormrage |
Posted: Tue Jul 08, 2008 1:52 pm Post subject: |
|
|
 Disciple
Joined: 25 Jun 2008 Posts: 186 Location: Bangalore/Singapore
|
Even ur business continuity model suggests the same.. One regular site/server and one backup/disaster recovery.. And I dont think u will ever convince your client by any means, to buy an additional server. Unless he is a lunatic with all sorts of phobias.
Think of a scenerio when somehow some beginner for no reason alters ur repository Qmgr and messages start piling up on XmitQ's everywhere and a Cluster Refresh only results in data about the Q's being lost and worsening with each additional refresh.. its a nightmare scenerio.. with upto 25 different clients testing at the same time.
And Potkey I doubt the propagation time needed when u have a single cluschl defined to ur FR's. In my previous project there were manual channels to each full repository. wonder Y _________________ *forgetting everything * |
|
Back to top |
|
 |
KeeferG |
Posted: Fri Sep 12, 2008 5:41 am Post subject: |
|
|
 Master
Joined: 15 Oct 2004 Posts: 215 Location: Basingstoke, UK
|
Despite my best efforts to reduce the number to 2, we run with 4 FRs. 2 per site for redundancy.
Each FR has three manually defined CLUSSDR to the other FRs.
Each FR has one manually defined CLUSRCVR.
Each PR has manually defined CLUSSDR to it primary FR.
Each PR has one or more manually defined CLUSRCVR (for latency reasons).
We have been running for several years with no problems.
We always build using command line. _________________ Keith Guttridge
-----------------
Using MQ since 1995 |
|
Back to top |
|
 |
|
|
|
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
|
|
|
|