How To Handle Hana Alert – Host CPU Usage


The HANA system becomes unresponsive.
When checking the Alerts tab, there is an alert with a description such as “suse-hanajl01 runs out of CPU resources! CPU consumption user mode 0%, system mode 0%, wio 0%, idle 0%”

  • In HANA Studio you would find the alert by directory Administration Console -> Alerts -> Show: all alerts.
  • In Solution Manager, you would find the alert using transaction DBACOCKPIT -> choose HANA system -> expand Current Status -> Alerts.
Remember for this “Check host CPU utilization” alert in Solution Manager, it is auto-refreshed every 15 minutes, so the time stamp on the alert may be not the exact time when alert occurs.
To check the exact time, please go to Administration Console -> Alerts -> double click on the alerts -> more occurrence for this alert.



All tests are performed on HANA 1.0 revision 41 and BW on HANA NW7.3 SP8


Normally there are 3 reasons for this alert:
1. The threshold value of “Check host CPU utilization” in “Configure Check Settings” is improper. The default value is Low: 25, Medium: 15, High: 10.

2. Too much CPU resources are consumed.

A high CPU consumption could be caused by the reasons including but not limited to:
  • Execution of certain transactions
  • Data loading
  • Hanging jobs
  • Long running SQL statements
They will generate a high calculation workload on CPU by either frequent access to database or joining between huge tables.

3. Bugs in some revisions

As we experienced in previous customer cases, you may experience bad query performance with BW on HANA cubes (e.g. by executing a query using transaction RSRT).


  1. For the alert triggered due to improper threshold value setting, it could be resolved by resetting it to the default value or a more reasonable threshold value.
  1. For the alert triggered due to high CPU consumption
  • Due to the various kinds of possible causes, the best way is to raise an SAP incident on application component HAN-DB to optimize the process. Please ensure to provide remote connection to HANA DB.
  • If the alert occurs when running a specific job, you could do a trace of this job and send the trace files to SAP for further analysis The trace should include the SQL trace and the performance trace. Both of them could be done in the Trace Configuration tab.
– For the SQL trace, after changing it to “Active” and giving a new file name, click “Finish”.
– For the performance trace, just give a new file name and click “Finish”.
  • If you are not sure which process has caused this alert, then try to figure out the process as described below:

    i. Go totab Landscape to check which server encounters the high memory consumption issue. E.g: The highest CPU usage is on indexserver with Port 30003.

  1. Go to Performance -> Threads, check which threads are running. E.g: a SQL statement is executed in the figure.
  1. Running the SQL again, if the issue is reproduced, we could make sure that the statement is (one of) the main cause of the alert.
  2. Do the trace as previous mentioned, and enclose the trace files in the Customer Message.
  • The trace files can be found in tab Diagnosis Files. The SQL trace file can be opened by double clicking on it.
In order to be able to read the performance trace file, you need to open it in the HDBAdmin tool. Please refer to WIKI page wiki HDBAdmin on Linux on how to execute the tool.
In the SQL trace file (, we could download it to local and open it with Notepad++, then find the SQL tracing info,
After you prepare your environment and execute the HDBAdmin tool, we could load the .tpt file and read it.
Then we could filter the result with known info (e.g Service is indexserver, Method is SqlExecuteStatement, User is SYSTEM) and take a further analysis on the SQL statement.
Even if you could do it by yourself, we still recommend you to send the trace files to SAP unless you are very sure about how to analysis, optimize and configure in further steps.
  1. If the revision of HANA DB is not the latest version, we strongly recommend to upgrade the HANA DB to the latest version.


See Also
  • Are there any Functional Constraints?
    • When the host CPU resource is not enough, scheduled background job running will be influence.
  • Are there any Non-functional Constraints?
    • Insufficient CPU resource may lead to a dump or database reboot, thus the system cannot provide a stable service and a real-time react to user’s action.
  • Are there any side-effects?
    • A high consumption of CPU resource will make the system’ performance very poor, and other user’s processor will be blocked in the query, which will lead to a bad developer experience.
  • Is there any suggestion to avoid this alert?
    • For reason 1, a more reasonable configuration could avoid this alert.
    • For reason 2, an optimization of the process may work.


Note :   Refer sap Note 1909670 – How to handle HANA Alert 5: ‘Host CPU Usage’  for more info

Leave a Reply