cancel
Showing results for 
Search instead for 
Did you mean: 

Central Instance in a Windows2008 MSCS with Oracle

Former Member
0 Kudos

Hello,

I have just finished the installation of a SAP Solution Manager with MSCS on Windows2008 and Oracle. This is my first MSCS, so I'm a little bit confused. I think almost all the things work right except the central instance. i explain:

I have the first node (node1) and the second node (node2) working fine. To check the MSCS installation, I shutdown node1 and make remote logon in node2 to see what's happening. And there are two remarkable things:

1.- Service/Instance SAP<SID>0 allways fails. If I startup it manually it works fine --> I suppose I have to add somewhere a retry after a delay.

2.- The central instance <hostname>20 doesn't exist in node2. So I cannot logon because there is no central instance. I only have the central instance in node1. I have read several times the documentation of the installation and I haven't found anything about installing the central instance in the additional MSCS node. It only says about the two options of the central instance installation: On a MSACS node or in a host outside MSCS (Not our case). Have I understood right?

Besides, I think the java instance is wrong too, I only have it in the first MSCS node.

Any help?

Thanks and best regards,

Ana

Accepted Solutions (0)

Answers (10)

Answers (10)

Former Member
0 Kudos

Hello,

Krishna, I almost has solved my problem. Now I have the CI running in node A and the DI has been running in node B for some minutes. Then I have stopped it for making some profiles changes and after these changes the sapmmc get hangs and I know the instance is up, because I can logon, but I cannot start/stop/see the traffic lights with the sapmmc. SAPMMC is allways not responding.

I daren't run the failover tests in this situation.

Thanks and best regards,

Ana.

Former Member
0 Kudos

If MMC hanging issue looks like OS related problem, check Windows event viewer and see what is the problem and resolve it.

Can you see both instances running in SM51? it is good start, otherwise, see what logs are complaining about.

Regards,

Krishna

Former Member
0 Kudos

Hello,

Varadhu, your answer clarify mi ideas about the cluster operation. But I still have some black points:

The DI installed in the second MSCS node must be allways running? In this is not the case, when the CI fails, the MSCS failover mechanism will not start the DI, as the DI is not a cluster resorce, so, it is a SPOF, isn't it? I'll try to explain:

why SAP does not deal with the CI as a SPOF and includes it in a cluster group? Could this be done? I say this because in the second node I have another non SAP application running and SAP will only be running in this second node in case of disaster o maintenance.

Krishna, thank you for your answer. I only have problems with the CI/DI. The database, ASCS, SCS, ERS and so, switch perfectly from the first node to the second. The only thing is that I cannot logon in SAP because the CI does not switch and the DI that I have installed in the second node does not start --> I have an error. I thought the error was caused because the installation wasn´t right, but, through all your answers and the documentation I have read, the installation is well done and the error is produced by another reason.

Thanks and best regards,

Ana.

PD.: I'm going mad with this HA installation.

Former Member
0 Kudos

Hi,

You mentioned:

"The DI installed in the second MSCS node must be allways running? I"

Ans: Yes. Both instances will be running. It is not the case that only one instance runs at a time. When either of the instances fail other instances will continue to run.

As mentioned by Varadharajan, when Primary instance fails ( Node A or CI), ASCS will move to DI node or Node B) and it will act as central instance.

Looks like, you should revisit your installation steps, if not atleast why is Secondory application instance is not running.. ? Check the logs in dev_disp and post them from DI instance.

regards,

Krishna

Former Member
0 Kudos

Hello,

Yes, I have followed the guide and it finished successfully.

The sap instance works fine in the first node. The problem is the failover.

Must I install the CI also in the second node?

Thanks,

Ana.

Former Member
0 Kudos

Hi,

No need to install the CI in the second node but u need to have a DI. The correct cluster configuration makes sure that ur DI to act as a CI when the actual CI fails. Go through the installation guide for the MSCS configuration and then proceed as described.

Regards,

Varadharajan M

Former Member
0 Kudos

Hi,

No need to install the CI in the second node but u need to have a DI. The correct cluster configuration makes sure that ur DI to act as a CI when the actual CI fails. Go through the installation guide for the MSCS configuration and then proceed as described.

Regards,

Varadharajan M

Former Member
0 Kudos

a) As Manoj advised usual known problem is as mentioned in note 1345206. --> This is the first thing to investigate.

b) If that is not issue, as mentioned in other replies, generally no issue around CI or DI , its worth verifying installation check list again and probably next place to look at is error log and posting details from error log.

c) Area to focus is ASCS and SCS as that is the part which moves, when you do fail over ( I guess both instances are starting and working fine on their own and you have trouble when you fail over.

First step to troubleshoot as mentioned in note 1345206 is, stop all SAP instances, Stop DB, copy sapstartsrv.exe from ASCS everywhere you have kernal files ie. ( On NodeA, Node B and Shared) and you may need to take ASCS off line to perform this excercise.

K

former_member189546
Active Contributor
0 Kudos

Hello,

Did you use the correct installation guide to do the install

www.service.sap.com/instguides

Normally sapinst guides you through all the steps

Did the installation complete ok with no errors in sapinst_dev.log?

regards,

John Feely

Former Member
0 Kudos

Hello,

Sorry I haven't updated because I've been ill.

My configuration is like the one that Manoj describes in his message, though I have the CI installed in the first MSCS node and the DI is not installed anywhere, because the second MSCS node has to have nothing running until the first MSCS node fails.

My problem is that when the first MSCS node fails, the second node starts the database (oracle) without problems, The ASCS and the SCS also without problems, but as the CI is only installed in the first node, the SAP instance doesn't work, we cannot logon.

I have also tried to install the DI in the second node, but the DI does not start without a CI running.

Any idea?

Thanks and regards,

Ana.

Former Member
0 Kudos

You haven't mentioned anything about the database. Have you already configured your database in the cluster as well?

A typical configuration looks like;

- ASCS, SCS and Database on the cluster

- CI and DI installed locally (CI can be on Node 1 and DI on Node 2)

- Enqueue Repliocation servers installed locally on both nodes.

Usually there is no need to tweek the start profiles (one exception is in SAP note 1345206)

Former Member
0 Kudos

Hello again,

I have tried this last comment of adding the start ABAP database commands in the START profile and it does not work.

Any other idea? Should I add the central instance as a cluster resource?

I'm desperate. I will necer finish this installation.

Regards

Former Member
0 Kudos

Hello,

Usually, you do not have to tweak Startup profile or any thing apart from routine post install tasks.

When it appears like kernel not able to connect to Database etc, I suspect problem with permissions on the shared drive which is shared between two nodes.

Usually, you should get that part correct before installation. But, you should be able to correct it, seek some help from Windows Admin specialist in your team, if there is none, you have to try giving local admin permissions and/or permissions to <SID>adm to both nodes on shared drive ( ASCS installed drive).

Regards,

Krishna

Former Member
0 Kudos

Hello Krishna,

I have read all you told me and now I understand what I'm talking about. But I have a very very big problem: the DI/Secondary application instance does not start. I have changed the work processes, so that the CI and the DI are equal.

When node 1 fails, all the instances ASCS and SCS switch fine, without any problem, but the Secondary application instance does not start. I think that the problem is that there are not connection with the database, as the steps

  1. Start ABAP database

#----


DB = $(DIRCT_RUN)\strdbs.cmd

Start_Program_02 = immediate $(_DB) OSM

of the start profile are only in the start profile of the Central Instance / Primary application instance.

Should I add these entries/lines in the start profile of the secondary applicacion instance?

Thanks and best regards,

Ana

qingqingwang
Advisor
Advisor
0 Kudos

Hello Ana,

> 1.- Service/Instance SAP<SID>0 allways fails. If I startup it manually it works fine --> I suppose I have to add somewhere a retry after a delay.

check the property of SAP Service/Instance resource, under advanced tab, is "restart" option ticked with threshold and Period

defined. e.g threshold = 3, Period = 900 seconds

> 2.- The central instance <hostname>20 doesn't exist in node2. So I cannot logon because there is no central instance. I only have the central instance in node1. I have read several times the documentation of the installation and I haven't found anything about installing the central instance in the additional MSCS node. It only says about the two options of the central instance installation: On a MSACS node or in a host outside MSCS (Not our case). Have I understood right?

Your understanding is correct, The ABAP central instance (now without the message and enqueue servers) is generally installed locally on one of the cluster nodes with a second, identically configured dialog instance on the remaining cluster node. You can find more information in the installation guide.

As message server and enqueue is running on ASCS, the central instance is not required in order to logon to SAP system.

If you have issue logon to system when central instance is not running, please check you have at least one dialog instance started besides the ASCS instance.

Best regards,

Qingqing Wang

Former Member
0 Kudos

Hello Ana,

Please read the following link and the links provided with in the discussion which should give you reasonable idea.

a) When node A fails ( or if you bring it down manually), ASCS moves to node B and node B will run DB instance and Application and you will be able to continue to work. It may be easy if you make terminology as CI = Primary application instance on nodeA and DI= Secondary application or additional application instance.

b) Java Fail over: If you bring down Node A instance/cluster or server completely, you will still be able to access nwa, ume, portal admin console etc.. but you will not have access to visual admin, config tool , jspm stc. ( I am talking about NW 7.0, SR2, any thing changed recently?.. not sure) .. { if any body got more ideas or correct me.. feel free}.

Hope this helps.

Regards,

K