When executing a step, commander agent records output from the commands of the step to the step log file. During the execution of a step, commander agent does the following steps:
Creates the step log file in the workspace
Appends output from step commands to and save the step log file (this step happens multiple times in case the output is big)
Close the step log file
Commander agent relies on the file service of the operating system in the above steps. It is possible that the operating system returns errors when commander agent invokes file service system calls. Here’re two example cases:
The step log file is deleted accidentally, so the operating system cannot find the step log file anymore
The workspace is in a network shared folder and the network is temporarily down
In such a case, commander agent retries the file service system 10 times in 500 millisecond intervals. Commander agent reports error and fails the step if all retries failed.
The retry times and retry interval is configurable. You can add the following two lines in the agent configuration file (/conf/agent/wrapper.conf) and set desired value.