IBM Support

DataStage jobs fail with OshExecuter.sh Segmentation fault error

Troubleshooting


Problem

A DataStage jobs fail with logged error similar to: RT_SC52/OshExecuter.sh: line 25: 15572 Segmentation fault $APT_ORCHHOME/bin/osh "$@" 2>&1

Resolving The Problem

A segmentation fault in OshExecuter.sh at line 25 means that the script was unable to launch the primary parallel job process, osh. This condition can occur due to insufficient system or DataStage resources and thus could affect random jobs during peak usage periods, or large jobs that require more resources.

Please check the following possible causes:

  1. Confirm that system resources (memory, cpu, i/o) are not constrained at the time of problem. Also ensure that the disk volume holding the project directories and system default temp directory are not full.
  2. Check the ulimit settings in effect for DataStage jobs. For Unix/Linux environments do not just run ulimit command at OS command prompt. Instead use the following step to confirm the limits that the DataStage engine (and thus all jobs) run with:

    Login to the DataStage Administrator client, select any project then click the Command button and enter command: sh -c "ulimit -a"

    Memory and Data should usually be unlimited or at least 2GB, while number of file handles should usually be at least 1024 but usually higher. For sites with service and engine tiers on same machine a value of 10240 is recommended.
  3. Check the uvconfig settings by logging into the DataStage engine server machine and then from the DSEngine directory issue command:
    bin/smat -t

    Refer to the following technote for a discussion of the default and recommended values for the important settings in this file such as the MFILES setting which if too low can cause a segmentation fault in OshExecuter.sh. These settings can be increased if needed according to the guidelines in the technote, but should not be decreased below default values:
    Use of the tunable parameters in the UVCONFIG file
  4. Ensure that your job and project do not define the following environment variable:
    APT_EXECUTION_MODE=ONE_PROCESS
    The above setting is only intended for debugging and can cause parallel jobs to abort.

[{"Product":{"code":"SSVSEF","label":"IBM InfoSphere DataStage"},"Business Unit":{"code":"BU059","label":"IBM Software w\/o TPS"},"Component":"--","Platform":[{"code":"PF002","label":"AIX"},{"code":"PF010","label":"HP-UX"},{"code":"PF016","label":"Linux"},{"code":"PF027","label":"Solaris"}],"Version":"9.1;11.5;11.3","Edition":"","Line of Business":{"code":"LOB10","label":"Data and AI"}}]

Document Information

Modified date:
16 June 2018

UID

swg21633531