Online Tutorials & Training Materials |
Register Login

SAP and MS Cluster Server Interview Questions and Answers

|| || 0

SAP and MS Cluster Server Interview Questions and Answers

FAQ: SAP and MS Cluster Server

1. What should I do if I receive a message that the system has not been installed correctly?

A message with this text is issued after an installation on a cluster each time there is an attempt to log on. With transaction SE16 maintain table INSTVERS. For every possible physical database host, (such as ORACLE) INSTVERS must contain either the virtual database host name (such as MSSQL7) or an entry with status 0. That means that INSTVERS must have two entries that contain the respective physical host names for a two-node cluster. The way it is written in the SPAGUI menu System -> Status... -> Database data -> Host must correspond with the syntax of these names (upper/lowercase).

There is a separate R3SETUP function for adjusting INSTVERS as of Release 4.5A: From the start menu choose "Completing cluster installation (Instvers)" and answer the questions. The step must be executed twice for Oracle clusters:

  • Once with Oracle and SAP on node A.
  • Once with Oracle and SAP on node B.

2. How should I use SAPDBA in a Microsoft cluster environment?

3.1I and higher are the affected releases
114287 can provide more information on this
ORACLE is the database that is used.
How SAPDBA can be used in an MS cluster environment is described in detail in 114287.

3. Which additional profile parameters are required?

The following additional profile parameters are required to operate the SAP system in an MSCS cluster environment. During the installation by R3Setup or SAPinst, these are set partly.

The Default profile includes:

gw/netstat_once = 0
rsdb/reco_trials = 3 (only ORACLE)
rsdb/reco_sosw_for_db = OFF (only ORACLE)
rsdb/reco_sleep_time = 5 (only ORACLE)
SAPDBHOST = <virt. database host name> (this must be manually changed partially after installation.)
Instance profile:
SAPLOCALHOSTFULL = <virt. R3 host name>
rsdb/reco_add_error_codes = +12500,+12571 (only ORACLE)
Since some parameters are not set correctly in all SAP releases automatically, after the installation the parameters for SAP releases 3.1I to 6.40 should be checked.

4. What should I do if work processes go into the "reconnect state" and/or the performance is worse in normal operation (database and SAP system on different nodes) than in emergency operation?

The automatic configuration of the network cards is the typical cause behind this issue. To set automatically, most Fast Ethernet adapters allow speed (10/100 MBit/s) and operation mode (full/half duplex). With "twisted" cables particularly that are often used in clusters the automatic configuration does not work properly. All network cards should be set to the same setting, for instance 100 MBit/full duplex. Caution: This setting should be supported by switches/hubs as well.

5. How to make a correct IP configuration?

For the cluster to function properly, a correct IP configuration is essential. The following settings are recommended by SAP:

  • Per host, two network cards are used. More than two network cards can increase the complexity; therefore, it is not sensible to use more network cards.
  • For access to SAP clients and the "rear" network for the heartbeat attribute and database communication (see point 10) the "front" network is used by a typical configuration.
  • Via host files or DNS servers name resolution for all seven IP numbers
  • "Enable DNS for Windows Resolution" to activate check box
  • Enter the DNS domain under "Domain Suffix Search Order" if you use a DNS server or host file that works with "fully qualified names" (such as

6. What should I do if there is redundancy for the heartbeat attribute?

In case, of a network failure, make sure that the cluster server for its "heartbeat" can use another network. The two network cards should be set to "Use for all communication” in the recommended configuration. In the cluster administrator, the preferred network in the list "Network Priority" should be set which can be reached by using "Properties" of the cluster name. (On the root of the hierarchy, click with the right mouse button.)

7. How to make sure that the shared disks are operating perfectly?

The "spine" of a cluster is formed by the shared disks. In case of a power outage or a node error, make sure that the shared disks can continue functioning. Individual shared disks should not break down under any circumstances with others continuing to operate. The hardware partner should be contacted to make sure the UPS software and hardware are set appropriately.

8. How to save cluster database?

A registry hive that stores the entire cluster configuration is the cluster database. It should be saved regularly as it is a critical component of the cluster. In 140990, the procedure has been described.

9. Why is the quorum disk never switched offline?

Information about replicating the cluster database between both nodes is contained in the quorum disk. As in the case of a network failure, the quorum disk prevents error functions, due to which quorum disk is never switched offline. For this reason, in one of the SAP groups (SAP or database) it is not desired that the quorum disk be contained. After installation, if the quorum disk is part of the ORACLE group, this may be because during installation, the repository database have been created and experimentally clustered on the quorum disk as suggested.

10. How to carry out database communication using the "rear" network?

During installation, SAP recommends that the entire communication should run using the "front" (public) network by restricting all settings. In this way, the documentation is also formulated. To ensure that the database communication is carried out using the "rear" network, the "finished" cluster should be reconfigured easily.

Proceed as follows to access this information:

  • In the rear network choose a new IP address
  • For the virtual database name change the name resolution so that the new IP address (in all hosts files or in DNS server) is pointed.
  • To ensure that it contains the new IP address, the database group's IP resource should be changed. Modify the linked network.
  • The DB and R3 cluster groups should be stopped and started

11. How to create additional IP addresses for SAP and database?

In the R/3 cluster group, additional IP resources can be created without any problems. To access the SAP system over the "rear" network, this can be done as well for example. From all existing IP addresses, since the SAP system receives the incoming data automatically; therefore, additional steps are not required.

To administrate the database using the other network, the additional IP resources to the database group can be added as well for example. To certain IP addresses, some databases (such as ORACLE) lock their entry ports. One should proceed according to the Knowledge Base Article Q254321 "INF: Clustered SQL Server Do's, Don'ts and Basic Warnings" for Microsoft SQL Server 7.0.

12. What will happen if incorrect database performance values using MSSQL 7.0?

On a second cluster node (i.e., on a node on which the SQL server has not been installed directly), if the MSSQL Server 7.0 database is running, incorrect values are displayed by transaction ST04. For example, under SQL Requests the memory values and all values may be 0. It is an SQL Server 7.0 problem in this case. On the second node, the files for the Windows performance counter are not updated.

13. How to install the SAP cluster group manually?

To the cluster configuration (such as if the cluster database has been damaged), sometimes an SAP cluster group needs to be created or similar changes are needed to be made. No new installation is necessary in this case. Programs that perform the necessary tasks can be found in the installation directory. Unpack archive NTCLUST.CAR or NTCLUSDB.CAR into an empty directory, if an installation directory no longer exists.

  • The R/3 group needs to be deleted: The Cluster Administrator should be used. Disk resources should not be deleted. Move them to a temporary group, instead.
  • The SAP Resource Type (Resource and Extension DLL) need to be unregistered: "DISAPRCT" should be called.
  • The SAP Resource Type needs to be registered: In the directory WINNTSYSTEM32 of both nodes, ensure that files SAPRC.DLL and SAPRCEX.DLL are there. Then "INSAPRCT" should be called.
  • The SAP cluster group: Call "CRCLGRP /sid <SID>" is created. Then as the dialog box appears fill it. Group "SAP-R/3 <SID>" is then created by the program, which except the R/3 resource contains all resources.
  • The R/3 resource needs to be created: Call "COCLGRP /sid <SID>". Then as the dialog box appears fill it. Resource "SAP-R/3 <SID>" in group "SAP-R/3 <SID>" is then created by the program.
  • Write any error message in all four programs that may arise to file R3CLUS.LOG.

14. Do the additional application servers (central or dialog instance) in a cluster are supported by SAP?

Up to and including 640, SAP kernel versions: No

The installation of additional application servers in a cluster is not supported by SAP. Clustered SAP central instance and the clustered database are the only supported configuration that it consists of. On other hosts, the application servers (dialog instances) are installed.

As of 700, SAP kernel versions: Yes.

Only the SCS instances (ABAP SCS and Java SCS, if required) are configured as of version 700 (only for new installations or homogeneous system copies currently) in the cluster. On one of the cluster nodes with a second, identically configured dialog instance on the remaining cluster node, the ABAP central instance (now without the message and enqueue servers) is generally installed locally. In the installation guide, more information can be found. If SAP is of the opinion that such a configuration makes sense and there is no alternative, a purely clustered central instance (without a database) can be supported.

Reason: A SAPLOC and, in part, a SAPMNT share is required in each SAP instance. However, it cannot virtualize the windows shares. . In other words, per cluster can only have one SAPLOC or SAPMNT share. All SAP systems would have to run in one cluster group to get by with only one SAPLOC share. However, to the principle of high availability, this runs contrary.

15. How should I fix error message: "irpstacksize" which is too small?

In certain situations, network shares "SAPMNT" and "SAPLOC" may switch to state "Failed" in clusters and the system cannot be started. In the event log, the following message can be found:

For the server to use a local device the server's configuration parameter "irpstacksize" is too small. The value of this parameter needs to be increased.
Solution: Microsoft Knowledge Base Article KB 285089 should be implemented.

16. What should I do if there is an error message during offline backup with Oracle?

During an offline backup with BRBACKUP, if the following error message is received, the batch file BRBACKUP.CMD should be updated according to the latest version.

“Database cannot be shut down as SAPR3 user is connected to the database or R/3 system is running. With warnings BRBACKUP is terminated successfully.”

17. What should be done if the Oracle group does not start since the database requests "recovery" (ORA-01113: file %s needs media recovery)?

If the Oracle group (resource <SID>.World is set to "Failed") cannot be started, follow the below given steps:
* As <sid>adm, the user should log on
* Restart all resources in succession except for "<SID>.World" after moving the Oracle group to your server.
* Manually the Oracle service should be started.
* Enter the following after opening a command prompt:
connect internal
startup mount
recover database (or in accordance with 4162 set data files "Missing
'end backup'" if in backup mode there are data files)
shutdown immediate
* The Oracle service should be stopped manually.
After this, the database resource can be started.

18. What shall I do if there is no display of hardware info and "Operating System Log" in ST06/OS06?

The cluster specific configuration of the SAPOSCOL is the reason for the missing OS06/ST06 display of the information. Proceed as follows to get the information as with non-clustered systems:
o The following profile parameter of the clustered SAP system should be added to the instance profile:
o The SAPOSCOL service on both nodes should be stopped.
o Rename the directory if usrsapPRFCLOG exists on the shared disk of the SAP cluster group.
One after the other execute the following actions on the two cluster nodes
o Open Control Panel -> System.
A new SYSTEM environment variable with the following value should be created:
The node should be rebooted. To apply the changes to the system environment, this is necessary.
Now in the SAPCLUSTER directory, all files of the SAPOSCOL are stored. You can cause the necessary information to be displayed by setting the profile parameter DIR_PERF to this exact directory.

19. The SAP system continues to run on the remaining nodes although one of my nodes has failed completely. What do I have to do to reinstall Windows on the failed node so that SAP functions afterwards?

Answer: Two options are there for you:

  1. Including the operating system, registry and cluster database, you have a functioning backup of all local disks. Get it installed. From the other node (%windir%clusterclusdb) you can also copy the cluster database. See 140990, for more information.
  2. Install a "fresh" copy of Windows and add it to the cluster, if you do not have a functioning backup. Merely execute cluster conversion "Node B" ("Configuring Node B for MSCS"), to get the SAP software to function, as mentioned in the installation/cluster guides. For the database software, configuration steps may still be required. For this, ask the manufacturer or read the database software documentation.

Listed below are the steps that the user can follow to do this:

a) Remove the “lost" host with "Evict Node" if it still appears in the cluster administrator of the remaining node.
b) On the new host install Windows Enterprise Edition and the new host should be added to your domain.
c) The new host is added to the cluster.
d) To let the database software run again in the cluster, execute the required steps.
e) As the Domain Administrator log on to.
f) From the kernel CD the NTCLUST.BAT should be started, directory NTCOMMON. The user should log off and then log on again.
g) To the new host move the database group. Move the SAP group to the new host after setting the SAP resource offline.
h) "Configuring Node B for Microsoft Cluster Server" should be started and the conversion according to the guide should be executed.
i) Use routine SAPINST to build the nodes, for SAP installations with 6.20 and higher.

20. How to set up new management console after the installation?

An icon is created for the "SAP R3 Management Console" for each of the two nodes during installation. To control only one local SAP system set the MSC file on which the conversion is based. Create a new console (if required on a different host) and change the settings in the cluster. The check box is deactivated to do this "Always show local R/3 instances" and by using the virtual R/3 name, an instance should be entered.

In the event log of the cluster node without these changes, the following message is generated:

An error has been received by DCOM "The system cannot find the file specified.” in order to run the server: {EFC29Cxx-A03C-11D0-A1F2 0000F62893CC} attempting to start the service SAP<SID>_ with arguments "-Service"

21. What shall I do if SAP cluster group changes constantly between the nodes ("ping pong")?

At first, the SAP resource should be set offline; but all other resources should be online in the SAP group. The SAP system will then be ignored by the SAP resource DLL file. In other words, in a non-clustered system as one would work, in the same way the work should be done here. Since, during the "ping pong" you cannot set the SAP resource offline, wait until the SAP resource stops with "Failed". By deactivating the restart of the SAP resource temporarily this can be forced. Select "Properties" by right-clicking on the SAP resource and then switch to "Advanced" and then the option "Do not restart" is selected.

Try to start the system with the Management Console (or with the SAP Service Manager) when the SAP resource has the status "Offline" or "Failed". Check the development traces (in particular dev_disp, dev_ms, dev_w0) and the Windows event log (in particular the application log) if the system does have the status "green". The SAP resource can be restarted if the system switches to status "green" and stays that way only. Normal operation is possible, if the system starts and only the dispatcher stays on "yellow",.

Technical background:

A Windows service (SAP<SID>_<Inst. No.>) is found in the SAP system that starts and monitors the SAP system constantly. By using a named pipe interface (up to Release 4.5A) or using a DCOM interface (as of Release 4.5B), the service itself is controlled. The SAP Service Manager or SAP Management Console should be used to control the service in a non-clustered system.

The SAP resource DLL file uses the same interface on a cluster system to the service as MMC/SAP service Manager. During the "Online Pending" phase, the resource DLL starts the SAP system and waits for the "green" status. The SAP resource DLL will trigger a restart/failover if, due to configuration errors, the status never switches to "green" or switches back from "green" to "yellow". For cluster operation consequently, a stable "green" status is essential. However, with an (offline) SAP resource, you can any time manually operate your SAP system.

22. How should I fix the issue if the SAP resource first starts ("Online Pending") but then jumps to "Failed"?

The option "Run in a separate resource monitor" should be activated and resource DLL SAPRC.DLL to version 6.20 should be upgraded to solve this problem.

23. Why partitions should be avoided?

The shared disks partitioning should be avoided. Since in the same resource multiple partitions of a physical disk are managed and therefore in the same group this yields no advantages. In addition, with partitioned disk resources the SAP installation tools have problems.

24. What should I do if there is an uninstallation of the SAP cluster?

* To node A move the SAP group. Except for the "physical disk" resource, all SAP resources must be offline.
* Choose Settings from the start menu, then select Control Panel and click on Add/Remove Programs.
* The SAP application server should be selected and the option "Remove" should be chosen.
* Then "Custom" option should be selected and the SAP instance that needs to be deleted is selected. In the Add/Remove Programs Wizard if errors occur, from the kernel CD start the uninstallation routine at the following path:
* After you have successfully performed the uninstallation on node A, move the SAP group to node B and start the uninstallation from the kernel CD at the following path:
* On node B, after the uninstallation has been performed successfully, all SAP resources should be removed from the SAP group. The "physical disk" resource should be moved from the SAP group to any other group before the SAP group is deleted. As the SAP group, the target group for the "physical disk" resource must be on the same node.
* The database is uninstalled.

25. What shall I do if the system has an error of system number 43?

For internal communication, Microsoft Cluster Server uses TCP port 3343. If the system has system number 43, this causes a conflict with the SAP gateway port. For this reason, with other system numbers in the cluster install SAP systems.

Related Articles