Failure injection / rate limit
We have a service that restrict the network links. Having the same for compute resources is highly demanded (e.g cpu/ram limit). Similarly failure injection (e.g crash of specific process).
In this direction I looked at os-fault, which first deal with the latter
- Get the location of running process to act on
- Allow to script action on those processes (can be docker, process, machines ...).
One step would be to ease the integration of os-fault: generate its inputs.