Test framework should sense, indicate or stop tests before system breakdown during a test run

Owned by Vishesh Hadimane
Last updated: Apr 25, 2019

Overview

Systems may go down when disk partitions gets filled up by PBS. Test Framework should keep tab on disk usage and check for corefiles if any present in PBS priv directories.

Interface 1:

Test Framework will periodically for every 5 mins monitor partitions on which pbs home and pbs exec are installed and where the user home directory resides and will display warning messages after 70% of the disk partition is filled up and will kill the benchpress process if disk usage reaches 90%.

  1. The message will be logged on PTL output console at default level after every 5 minutes with the percentage of the disk partition which is filled.

           WARNING: Disk on which <PBS_HOME/PBS_EXEC/USER's HOME> resides is <percent> filled. Please clean the disk

           Example:

           WARNING: Disk on which PBS_HOME resides is 75% filled. Please clean the disk

      2. Once the disk usage reaches 90% in any of the partition which has user's home or PBS_EXEC or PBS_HOME , pbs_benchpress process will be killed with an Error message displayed on the console.

          ERROR: Disk on which <PBS_HOME/PBS_EXEC/USER's HOME> resides is 90% filled test will be stopped.

Interface 2:

Test Framework will periodically for every 5 mins check if any core files are present in server_priv,mom_priv or sched_priv or user's home directory and will log a message on PTL output console at default level.

WARNING: Core files are found in <server_priv/mom_priv/sched_priv/user's home>

Example:

WARNING: Core files are found in server_priv.







OSS Site Map

Project Documentation Main Page

Developer Guide Pages