Monitoring SAP R/3 with Check_MK


Letzte Aktualisierung: 21. Februar 2013

1. Overview

As of version 1.2.3i1 Check_MK ships an agent plugin and checks for monitoring various aspects of SAP R/3 systems. Currently the following checks are available:

sap.dialog Check response time, number of logged in users
sap.value Process various monitoring information from SAP

Those checks can be used to fetch monitoring information from SAP and make it available to Nagios and give you the option to use your standard visualization, notification and reporting mechanisms. Those checks are very basic at the moment, but might be a good start for future improvements or extensions.

The SAP R/3 (mk_sap) agent plugin can be installed on several systems, either a single seated SAP system, all nodes of a cluster or a solution manager host. It is also possible to install the agent on the Check_MK host an let it connect to a SAP R/3 installation on another host. The agent plugin uses the RFC protocol to connect to SAP. According to the docs of the python module this should work with all R/3 releases since 4.6x.

The agent plugin is implemented in python and uses the sapnwrfc module, which has been used in version 0.19 during development. This module must be installed on the host running the agent plugin. You also need to have the nwrfcsdk installed, it can be downloaded from the SAP download portal.

2. Installing prerequisites

Install the SAP NetWeaver RFCSDK (nwrfcsdk). You need to download this from the SAP service Portal for Software downloads, and follow the path of:

Download -> Support Packages and Patches -> Entry by Application Group -> Additional Components -> SAP NW RFC SDK -> SAP NW RFC SDK 7.xx -> SAP NW RFC SDK 7.xx.

Download this, unpack and install it. Once installed you should have the directory /usr/sap/nwrfcsdk/ on your system. It should contain the needed library /usr/sap/nwrfcsdk/lib/

You need to ensure that this library can be found by ld. To ensure this, create the file /etc/, put /usr/sap/nwrfcsdk/lib into the file and run ldconfig afterwards.

Now you need to install the sapnwrfc python module, which should be shipped with Check_MK and located on your Check_MK server in the agents' sap directory (usually /usr/share/check_mk/agents/sap/sapnwrfc-0.19.tar.gz). Copy this to the host the agent should be running on, unpack, compile and install it. To be able to compile the module you need to have a python versions >= 2.6 and the python-devel and gcc packages of your system installed.

root@linux# cp sapnwrfc-0.19.tar.gz /tmp
cd /tmp
tar xvzf sapnwrfc-0.19.tar.gz
cd sapnwrfc-0.19
python build
python install

Now you can remove the temporary files we created below /tmp.

If you run this plugin on several SAP systems, you do not need to compile it on every system. You must only ensure that you compile it on one system and copy all needed files of the python module to the other systems. For examply, you can run python build --prefix=target and then copy all files and directories from the target directory to e.g. /usr/local on the target system.

Having this plugin installed you should be able to load the module in python like this, without receiving any error message:

root@linux# python -c "import sapnwrfc"

3. Installing the plugin

In order to use the SAP R/3 monitoring you need the agent plugin mk_sap, which you will find in the agents' plugins directory on your Nagios host (usually /usr/share/check_mk/agents/plugins).

Install the plugin on the host of your choice into the plugins directory of the Check_MK agent. Look into the agent's code in order to learn which directory this is. The default is /usr/lib/check_mk_agent/plugins. It may vary in your installation. If you are unsure, you can grep it out of the agent:

root@linux# grep MK_LIBDIR /usr/bin/check_mk_agent
export MK_LIBDIR="/usr/lib/check_mk_agent"

If the directory is not yet existing, simply create it. Then put the plugin mk_sap there and make it executable:

root@linux# mkdir -p /usr/lib/check_mk_agent/plugins
root@linux# cp mk_sap /usr/lib/check_mk_agent/plugins
root@linux# chmod 755 /usr/lib/check_mk_agent/plugins/mk_sap

4. Configuring the plugin

The agent plugin needs to be able to connect to the R/3 system using RFC calls usually you need to configure the connection settings like hostname, system id, username and password. This is done in the sap.cfg, which has to be located in the agent configuration directory. This is usually /etc/check_mk. If unsure, you can again grep this from your agent:

root@linux# grep MK_CONFDIR /usr/bin/check_mk_agent
export MK_CONFDIR="/etc/check_mk"

Now create this directory (if not existing) and copy the sample configuration from the agents' sap directory on your Check_MK host (usually /usr/share/check_mk/agents/sap/sap.cfg).

root@linux# mkdir -p /etc/check_mk
root@linux# cp sap.cfg /etc/check_mk/sap.cfg

Now open the file in your favorite text editor to customize the connection settings. You can execute the mk_sap plugin manually to test your current configuration. It will throw errors if your settings are not correct.

5. Configuration on the Check_MK server

The system will now offer all information you told it to forward in the agent configuration to the Check_MK server. Now, when you create a host named like the SAP System ID provided by the agent, the retrieved monitoring information will automatically be assigned to it.

For example if you have a single SAP Instance named XYZ on the host xyzhost, you first need to add the xyzhost as a regular server to your monitoring. Even after inventory it should not get any SAP related services assigned. Now add the host XYZ to your monitoring and perform a service inventory for this host. It should find some SAP related services now. After a restart of Nagios, you should have the services added to your monitoring.

By default the sap.value check does not add any available monitoring information as service. You can add a rule (for example using WATO) to let the check add all available information to your monitoring. Please refer to the check documentation linked before for details.

6. Handling of data classes

The monitored SAP systems provide their information in different data classes. Currently the agent plugin handles the following classes:

  • MTE_PERFORMACE: Performance value, normally a single integer with a unit of measure.
  • MTE_SINGLE_MSG: Single status message
  • MTE_MSG_CONTAINER: Multiple messages which all have an individual state (like a log)

All of these classes provide a state indicator which simply transported from SAP to Check_MK. There is one special handling for the MTE_MSG_CONTAINER classes: The logs from the message containers are transported as logwatch checks to Check_MK. This makes it possible to handle the single events from this container which are provided by SAP individually within Check_MK, for example acknowledge only some of them while keeping the others unacknowledged.