Why am I seeing this message?
An attack is interrupted when the Gremlin attack process is terminated by an outside actor while Gremlin is running the attack. The outside actor can be supervising software such as an OOMKiller or a Kubernetes kubelet.
The Linux operating system has an Out of Memory Manager (sometimes referred to as the OOMKiller) that kills processes when it detects that the system is out of memory. When a process is reaped by the OOMKiller it receives a SIGKILL signal.
In Kubernetes, the Gremlin attack process runs inside the Linux process namespace of the container that Gremlin is targeting. Resource limit violations and liveness probe failures will trigger the termination of both the target and Gremlin attack processes. Even if the container processes within the target pod have not violated a limit or failed a liveness probe, they may be terminated during a node-pressure eviction.
What hypothesis does this support?
This outcome could confirm one of several hypotheses, all of which are around confirming the proper functioning of some resilience mechanism.