Rely on | Cgroups to limit and account resource usage of the processes running inside the containers | Namespaces to provide a different view of the system resources for the process inside the container. Currently we have 6 | IPC | Network | Mount | PID | User | UTS. Not everything can be namespace. Privileged can attack and own the system. Power of super user has been divided into distinct units, known as capabilities, which can be independently enabled and disabled.
Linux daemon in charge of managing the lifecycle of containers. Exposes RESTful APIs on a UNIX or HTTP Socket
Public repository for container images
We chose SystemTap since you can extend its script language using C code. We used this feature to query the kernel to get information about the privileged operation daemon and container processes
we can safely grant
/tmp directory that contains temporary files
/usr/share is the place for manuals, documentations, examples
Top 20 of Docker hub
It operates in the testing environment so we don’t have overhead on production
HIDS should be able to notify abnormal activities like if AppArmor has been disabled
Poved to be very effectuve wiht attacks on
Since our AppArmor profiles strictly check the mount flags, the execution of the container is blocked during its early stages
Can not be written
Injected Shocker code by overwriting the application startup script
Any attempt to access the host file system triggers a security check with /shared as first level directory