You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
| TCP | 10251/10259 | kube-scheduler | 10251 could be insecure before 1.13, after that only the secure port 10259 is available |
72
+
| TCP | 10252/10257 | kube-controller-manager | 10252 could be insecure before 1.13, after that only the secure port 10257 is available |
73
+
| TCP | 10255 | Read-Only Kubelet API | HTTP port for GET requests |
74
+
| TCP | 30000-32767 | NodePort Services | Service endpoints, could be HTTP |
82
75
83
76
#### Authentication and Authorization
84
77
@@ -101,7 +94,7 @@ could manipulate resources that are managed with the Kubelet.
101
94
102
95
To disable anonymous requests, the Kubelet should be started with `--anonymous-auth=false`.
103
96
Authentication can be provided either through x509 client certificates or API bearer tokens.
104
-
How to setup both approaches can be found in the [Kubelet Authentication and Authorization](https://kubernetes.io/docs/reference/access-authn-authz/kubelet-authn-authz/).
97
+
How to set up both approaches can be found in the [Kubelet Authentication and Authorization](https://kubernetes.io/docs/reference/access-authn-authz/kubelet-authn-authz/).
105
98
106
99
Kubelet authorization is set to `AlwaysAllow` as a default mode. This can be quite problematic,
107
100
since all authenticated users can do all actions. To mitigate this, it is possible to delegate
@@ -118,9 +111,8 @@ Multiple policies can be setup to limit and control the capabilities of workload
118
111
especially in order to prevent malicious actors from exploiting obvious faults or even to
119
112
just prevent incorrectly configured workloads from overusing resources.
120
113
121
-
An easy way to do this would be Resource Limiting on a cluster. This can be done on a per
122
-
namespace basis and can prevent the overuse of resources or even prevent the creation of
123
-
too many pods, services or volumes.
114
+
An easy way to do this would be Resource Limiting on a cluster. This can be done on a per-namespace
115
+
basis and can prevent the overuse of resources or even prevent the creation of too many pods, services or volumes.
124
116
125
117
It is also possible to change the security context of a pod, which changes different settings of a pod, like
126
118
its underlying Linux user, the access to the host network or different other controls that
0 commit comments