Supported Options

• Help and Version

  • --help, -h: Display usage instructions

  • --version: Display version information

• Configuration

  • --affinity <cpu>: Set CPU affinity for the SysStat process. Default is no affinity. For example, the sysstat process can execute on any available system core

  • --mem-update-interval <#>: Number of samples between checking memory-related metrics for existing processes. The default value is 1 sample

  • --proc-update-interval <#>: Number of samples between checking proc tree for new processes. The default value is 1 sample

  • --sample-period <ms>: Define the sampling frequency in milliseconds. The default is 1000 ms i.e., 1 second

  • --ignore-existing-procs: Disable tracking of processes existing in the system before SysStat’s start

Attention

Modifying the –sample-period, –proc-update-interval, or –mem-update-interval parameters can affect the monitoring overhead. Increasing the sampling rate can result in a large amount of data log file and high cpu load.

• Collectors

  • --all: Enable all collectors (exception is thread collector which must be explicitly enabled)

  • --cpu: Enable CPU collector

  • --memory: Enable memory collector

  • --process: Enable process collector

  • --thread: Enable thread collector

  • --fs: Enable filesystem collector

• Output and Debugging

  • --output <filename>: Specify the output log file. Default is “sysstat_<timestamp>.log” file in the current directory

  • --pidfile <filename>: Specify the PID file location. This file contains the process ID (i.e., PID) of the sysstat process. The default is the “sysstat.pid” file in the directory containing the output log file mentioned above

  • --parse <sysstat.log>: Parse an existing binary log file and output collected data in human readable format

  • --dump: Dump debug strings to sysstat.log.txt for analysis

  • --split: Split the log file when its size exceeds 50MB