on 05-08-2012 3:11 AM
Hi all,
My company is using SAP Business Objects BI v4 SP02 Patch 11.
Currently we have 1 application server connect to 1 database server
Application server
- running on IBM AIX Machine
- OS 6.1
- 16GB Memory
- 50GB Hard Disk
- installed with WebSphere v7.0.0.21
- installed with BOBJ BI v4 SP02 Patch 11
- BOBJ has been configured as follows :-
a) SIA node name = SIA_PROD1 (with default 16 services)
b) SIA port no = 6410
c) CMS name = CRPRD1
d) CMS port no = 6400
e) CMS database connected to METADATA
f) AUDIT database for CMS connected to AUDIT
Database server
- running on IBM AIX Machine
- OS 6.1
- 16GB Memory
- 150GB Hard Disk
- installed with IBM DB2 server software
- 3 databases have been created which are METADATA , REPORT and AUDIT
Now we would like to add another 2 new physical servers to serve as a backup/high availability purposes. 1 for APPL and another 1 for DBASE.
We've completed clustering the DBASE with IBM PowerHA (active - passive) but we're not sure on how to configure BOBJ BI v4 for the new server.
Questions during installation on the new server :-
a) Do I need to key-in SIA_PROD1 or create a new one called SIA_PROD2 ?
b) Do I need to key-in 6410 or create a new port no for above SIA node ?
c) Do I need to key-in 6400 or create a new port no for CMS ?
d) Is there any different in deploying BOE file ?
For time being only above confused me. I've read "Business Intelligence Platform Administrator Guide" but the explanation is too general.
See attached file on what type of clustering I want to perform.
Also if possible, kindly highlight me any important things that I may left out.
Thank You.
Hi
Found this might help the others trying to do this from scratch:
http://wiki.scn.sap.com/wiki/display/BOBJ/BI+platform+4.0+cluster+configuration
Regards
Emmanuel
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Our new additional server has arrived & ready to be installed.
Below is my checklist :-
Additional application server
- running on IBM AIX Machine
- OS 6.1
- 16GB Memory
- 50GB Hard Disk
- installed with WebSphere v7.0.0.21
- installed with BOBJ BI v4 SP02 Patch 11
- BOBJ has been configured as follows :-
a) SIA node name = SIA_PROD2 (with default 16 services)
b) SIA port no = 6411
c) CMS port no = 6400
e) CMS database connected to METADATA
f) AUDIT database for CMS connected to AUDIT
I need to know whether above checklist (item a,b & c) is correct before proceed.
Kindly advice.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
I dont see any problem with this.
I just don't get why you are changing the default SIA port from 6410 to 6411 but that's not a problem in itself, just a choice to make
As far as Im concerned, the only thing I would do before clustering is to make sure I have a backup of the repository prior to implementation of the secondary server.
The real question is about how you are going to do your cluster:
1- install a standalone box and map it afterward
or
2- expand the existing install
Im more keen on option one as you need to install an SP02 first before patching to patch 11 and I don't like to see that happening on the prod server (because your server is in p2.11 and you would map an SP02 CMS to it).
In case of standalone, I'd run the install and then recreate the SIA in the box to map to the existing repository using serverconfig.sh before configuring the FRS so they are shared between the 2 boxes.
Regards
Hi Romain
You mean I can use same port which is 6410 in PROD2 ? Oh ok, I thought I need to use different port since 6410 has been used by PROD1
I've finished install BOBJ in PROD2.
Can you teach me step by step to perform below in serverconfig.sh ?
ie : Do I need to choose "add node" or "move node" , etc etc...
In case of standalone, I'd run the install and then recreate the SIA in the box to map to the existing repository using serverconfig.sh before configuring the FRS so they are shared between the 2 boxes.
Him
If both env use the same network card (which would imply they are installed on the same OS), you would indeed get a port conflict between the two environments.
However, your schema clearly indicates they are on separate boxes (even if virtual boxes) so there's no risk of port conflict.
In SP02, you should select Recreate Node if available (this was introduced in BI 4.0) or Add Node and reuse the same SIA name you previously used, which would overwrite the existing one, pointing it to the correct system. You can then disable the DB of PROD2 as it won't be used.
As to the exact step by step procedure for serverconfig.sh, you should find guidelines in the administration guides for this.
Hi
Is my below setup correct ?
Physical Box 1
BOBJ has been configured as follows :-
a) SIA node name = PRD1_SIA (with default 16 services)
b) SIA port no = 6415
c) CMS name = BCHRPTPRD01
d) CMS port no = 6405
e) CMS database connected to METADATA
Physical Box 2
BOBJ has been configured as follows :-
a) SIA node name = PRD2_SIA (with default 16 services)
b) SIA port no = 6416
c) CMS name = BCHRPTPRD02
d) CMS port no = 6406
e) CMS database connected to METADATA
http://i26.photobucket.com/albums/c121/aris0azie/CMC_console.jpg
PRD2_SIA resides in 30.134 and PRD1_SIA resides in 30.133
Both connected to the CMS DB but somehow I could'nt start PRD2_SIA
I've run below cmd at 30.134 but failed.
./cms.sh -start PRD2_SIA
Hi Denis,
I am also new to this, I couldn't find any reference doc on how to cluster my BI 4.0 environment here.
My scenario : 2 windows 2008 servers, 1 ms sql servers, BI 4.0 Objective:
Install and cluster BI 4.0 so that when end users are connected to node 1 of BI cluster via BI launch pad or other client tools, they should automatically failover to node 2 if node 1 goes down. Also, the users should have a common point of entry to the BI cluster and should not point directly to node 1 or node 2.
I am lost at this point.
Hi Patrick
Denis is right. You're on WINTEL platform while I'm on AIX...
The concept is same but the installation/configuration process are different.
After I read below link, I think I understood on how it works.
http://www.findsol.com/blog/business-objects-31-cluster/
Meaning my clustered setup for PRD2_SIA was totally wrong.
I need to re-install box 2 again with "Expand Install" option.
Or is there any other way that I can avoid to re-install ?
The clustering of SAP Bi4.0 is not different from clusterring in XI, XIR2 or XI3.1, but has some minor differences.
Basicaly, you install the product on your additional servers using expanded install, when it asks for existing CMS name you give them CRPRD1 (it has to run during this and be accessible from machine you're installing on, you should be able to ping by that name), then you provide connection info to the same METADATA and AUDIT databases and cluster key you created during install of your first SIA and you're done.
You should use different SIA names for clarity and you can use same ports.
Process of Clustering s complete when you point 2 or more SAP BI4.0 installs to the same CMS DB and all their FRS to same FRS Share.
Not sure I understand your question about BOE file.
Business Intelligence Platform Planning Guide and Web Applications deployment Guide are great docs that might clarify your questions.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Denis
Ok noted.
I'm not sure whether I know how to use expanded installation because our previous vendor didn't teach us on clustering setup. I'm only familiar with fresh/full installation.
Anyway, worth to try for my own knowledge.
Will download & read the planning the guide for better understanding.
Well, you can use full install on all 3 boxes, using all defaults. Once all 3 installs are working fine by themselfs, with their own CMS DB and FRS, you can proceed to cluster them.
Stop all 3 original SIA
Create new or select existing DB for the cluster CMS DB
If DB you selected is the same as CMS DB used by box 1, skip to step 2, otherwise :
step 1: on box 1 create new SIA pointed to the new DB and with option to create CMS DB (re-create), once succesfull
step 2 : on box 2 create new SIA pointed to the CMS DB from step 1
step 3 on box 3 create new SIA pointed to CMS DB from step 1
step 4 configure all FRS server to the same share, disable original SIA's.... done.
Business Objects data is stored in 2 places:
- the repository (DB) where all the metadata are stored (users, security, information about reports, comments, almost everything)
- the filestore, which contain physical files, as the unv/unx (BI4), the DeskI, WebI, Crystal documents and so forth
In the CMC, you will find the Input and Output File Repository servers (We'll name them FRS). These are the services in charge of accessing the filesystem to retrieve the physical files.
So, apart from the metadata that need to be shared as shown in your schema, you will also need to have a shared folder that both BI server are able to access.
Three things are therefore to be taken into consideration.
1- Your SIA needs to run under a domain account. LocalSystem doesn't have access to network shares
2- Both systems need to be able to access the share in question
3- In the CMC, you need to reconfigure the FRS servers to point to that shared folder.
Warning as a side note on your draft schema:
DO NOT USE VIRTUAL IP IN FRONT OF YOUR BO/BI SERVER. The CORBA layer is very sensitive to the network configuration and is able to address the dispatch of data between the J2EE servers (Web servers) and the BI cluster. This will most likely prove very problematic in your implementation. I reckon this is not only unsupported but that it will give you massive headaches to deal with in the future.
Client tools will probably refuse to connect to the BI back-end and the whole BI system itself could have problems with inter-process communication.
I've seen the system refusing to work properly for less than that (e.g.: a different netmask meaning the two BI servers would not communicate through the same route)
Hope this helps.
And no PowerHA for the BI part please. BI processes in a cluster should give you HA + LB functionalities just fine without it.
Another thing is that you may want to compare your active and passive CMS DB at a time they are in sync. There are non-logged BLOB objects when using DB2 (due to a former DB2 limitation) which may not fully replicate.
This ought to change in the future but is still work in progress.
Edit: PowerHA should do for the DB2 replication, which is more of a system clustering. I got that mixed with HADR (the DB2 clustering functionality).
Still, it's always worth mentioning and checking.
Konovalov,
I saw your advice on "DO NOT USE VIRTUAL IP IN FRONT OF YOUR BO/BI SERVER". So is that mean CMS on MSCS (2-nodes, active standby) will not work properly ?
I have a installation having DB in MSCS, FRS on share storage, and ALSO 2-node CMS cluster with MSCS failover. So there is a virtual IP as an entry point of CMS service. Will the CMS load sharing happen on it ?
If there is no virtual IP for CMS, I'm wondering how the web application (such as BIlaunchpad) know which CMS to connect to ? try one by one every time ?
In addition, how the developer tool know which CMS is alive or not. Should they manually switch the configuration once the main CMS server is down ?
I think there must be some trick under the hood, hope you can shed some light on it.
Thanks
User | Count |
---|---|
75 | |
9 | |
8 | |
6 | |
6 | |
5 | |
5 | |
5 | |
5 | |
5 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.