Description
When a CloudBees CD (CloudBees Flow) step is aborted or force aborted, what mechanism is being used in the OS?
Solution
For force abort, the server:
-
sends an abort command to the agent
-
marks the step aborted (without waiting for the agent to finish aborting the step)
After the agent receives the message, it:
-
sends a SIGTERM signal to ecwrapper
-
waits for ecwrapper to complete
-
ecwrapper in turn sends a SIGTERM to the actual running command even though this is a force abort.
-
If the command does not complete in 2 minutes, ecwrapper sends a kill -9 to all child processes associated with this step.
-
In other words, regardless of regular abort or force abort, the agent waits two minutes before kill -9’ing the step.
This article is part of our Knowledge Base and is provided for guidance-based purposes only. The solutions or workarounds described here are not officially supported by CloudBees and may not be applicable in all environments. Use at your own discretion, and test changes in a safe environment before applying them to production systems.