container_engine_selinux − Security Enhanced Linux Policy for the container_engine processes
Security-Enhanced Linux secures the container_engine processes via flexible mandatory access control.
The container_engine processes execute with the container_engine_t SELinux type. You can check if you have these processes running by executing the ps command with the −Z qualifier.
For example:
ps -eZ | grep container_engine_t
SELinux defines process types (domains) for each process running on the system
You can see the context of a process using the −Z option to ps
Policy governs the access confined processes have to files. SELinux container_engine policy is very flexible allowing users to setup their container_engine processes in as secure a method as possible.
The following process types are defined for container_engine:
container_engine_t
Note: semanage permissive -a container_engine_t can be used to make the process type container_engine_t permissive. SELinux does not deny access to permissive process types, but the AVC (SELinux denials) messages are still generated.
The SELinux process type container_engine_t is an MCS (Multi Category Security) constrained type. Sometimes this separation is referred to as sVirt. These types are usually used for securing multi-tenant environments, such as virtualization, containers or separation of users. The tools used to launch MCS types, pick out a different MCS label for each process group.
For example one process might be launched with container_engine_t:s0:c1,c2, and another process launched with container_engine_t:s0:c3,c4. The SELinux kernel only allows these processes can only write to content with a matching MCS label, or a MCS Label of s0. A process running with the MCS level of s0:c1,c2 is not allowed to write to content with the MCS label of s0:c3,c4
SELinux policy is customizable based on least access required. container_engine policy is extremely flexible and has several booleans that allow you to manipulate the policy and run container_engine with the tightest access possible.
If you want to deny any process from ptracing or debugging any other processes, you must turn on the deny_ptrace boolean. Disabled by default.
setsebool -P deny_ptrace 1
If you want to allow sandbox containers to use all capabilities, you must turn on the virt_sandbox_use_all_caps boolean. Enabled by default.
setsebool -P virt_sandbox_use_all_caps 1
If you want to allow sandbox containers to use netlink system calls, you must turn on the virt_sandbox_use_netlink boolean. Disabled by default.
setsebool -P virt_sandbox_use_netlink 1
The SELinux process type container_engine_t can manage files labeled with the following file types. The paths listed are the default paths for these file types. Note the processes UID still need to have DAC permissions.
bpf_t
/sys/fs/bpf
cifs_t
container_file_t
/srv/containers(/.*)?
/var/lib/origin(/.*)?
/var/lib/rkt/cas(/.*)?
/var/lib/nerdctl/[^/]*/volumes(/.*)?
/var/lib/buildkit/[^/]*/snapshots(/.*)?
/var/srv/containers(/.*)?
/var/lib/containerd/[^/]*/snapshots(/.*)?
/var/lib/kubernetes/pods(/.*)?
/opt/local-path-provisioner(/.*)?
/var/local-path-provisioner(/.*)?
/var/lib/containers/storage/volumes/[^/]*/.*
/var/lib/kubelet/pod-resources/kubelet.sock
/home/[^/]+/.local/share/containers/storage/volumes/[^/]*/.*
ecryptfs_t
/home/[^/]+/.Private(/.*)?
/home/[^/]+/.ecryptfs(/.*)?
fs_t
fusefs_t
/var/run/user/[0-9]+/gvfs
hugetlbfs_t
/dev/hugepages
/usr/lib/udev/devices/hugepages
nfs_t
onload_fs_t
semanage fcontext can also be used to manipulate default file context mappings.
semanage permissive can also be used to manipulate whether or not a process type is permissive.
semanage module can also be used to enable/disable/install/remove policy modules.
semanage boolean can also be used to manipulate the booleans
system-config-selinux is a GUI tool available to customize SELinux policy settings.
This manual page was auto-generated using sepolicy manpage .
selinux(8), container_engine(8), semanage(8), restorecon(8), chcon(1), sepolicy(8), setsebool(8)