Add instructions for k8s audit support in >= 1.13

In K8s 1.13, there's a new mechanism for k8s audit logs using Audit
Sinks, which can be created and managed like other k8s objects.

Add instructions for enabling k8s audit logging for 1.13. The patching
script is still required, as dynamic audit is not a GA feature and needs
to be enabled. Also, the audit sink config is a template and needs to be
filled in with the cluster ip address, like the webhook config for 1.11.
This commit is contained in:
Mark Stemm 2019-05-15 11:56:19 -07:00 committed by Lorenzo Fontana
parent bd0ca4f5a7
commit 6a6342adc6
4 changed files with 113 additions and 23 deletions

View File

@ -1,6 +1,8 @@
# Introduction
This page describes how to get K8s Audit Logging working with Falco. For now, we'll describe how to enable audit logging in k8s 1.11, where the audit configuration needs to be directly provided to the api server. In 1.13 there is a different mechanism that allows audit confguration to be managed like other k8s objects, but these instructions are for 1.11.
This page describes how to get K8s Audit Logging working with Falco for either K8s 1.11, using static audit policies/sinks, or 1.13, with dynamic audit policies/sinks using AuditSink objects.
## K8s 1.11 Instructions
The main steps are:
@ -9,11 +11,11 @@ The main steps are:
1. Restart the API Server to enable Audit Logging
1. Observe K8s audit events at falco
## Deploy Falco to your K8s cluster
### Deploy Falco to your K8s cluster
Follow the [K8s Using Daemonset](../../integrations/k8s-using-daemonset/README.md) instructions to create a falco service account, service, configmap, and daemonset.
## Define your audit policy and webhook configuration
### Define your audit policy and webhook configuration
The files in this directory can be used to configure k8s audit logging. The relevant files are:
@ -26,11 +28,11 @@ Run the following to fill in the template file with the ClusterIP ip address you
FALCO_SERVICE_CLUSTERIP=$(kubectl get service falco-service -o=jsonpath={.spec.clusterIP}) envsubst < webhook-config.yaml.in > webhook-config.yaml
```
## Restart the API Server to enable Audit Logging
### Restart the API Server to enable Audit Logging
A script [enable-k8s-audit.sh](./enable-k8s-audit.sh) performs the necessary steps of enabling audit log support for the apiserver, including copying the audit policy/webhook files to the apiserver machine, modifying the apiserver command line to add `--audit-log-path`, `--audit-policy-file`, etc. arguments, etc. (For minikube, ideally you'd be able to pass all these options directly on the `minikube start` command line, but manual patching is necessary. See [this issue](https://github.com/kubernetes/minikube/issues/2741) for more details.)
It is run as `bash ./enable-k8s-audit.sh <variant>`. `<variant>` can be one of the following:
It is run as `bash ./enable-k8s-audit.sh <variant> static`. `<variant>` can be one of the following:
* "minikube"
* "kops"
@ -40,15 +42,65 @@ When running with variant="kops", you must either modify the script to specify t
Its output looks like this:
```
$ bash enable-k8s-audit.sh minikube
$ bash enable-k8s-audit.sh minikube static
***Copying apiserver config patch script to apiserver...
apiserver-config.patch.sh 100% 1190 1.2MB/s 00:00
***Copying audit policy/webhook files to apiserver...
audit-policy.yaml 100% 2519 1.2MB/s 00:00
webhook-config.yaml 100% 248 362.0KB/s 00:00
***Modifying k8s apiserver config (will result in apiserver restarting)...
***Done!
$
```
### Observe K8s audit events at falco
K8s audit events will then be routed to the falco daemonset within the cluster, which you can observe via `kubectl logs -f $(kubectl get pods -l app=falco-example -o jsonpath={.items[0].metadata.name})`.
## K8s 1.13 Instructions
The main steps are:
1. Deploy Falco to your K8s cluster
1. Restart the API Server to enable Audit Logging
1. Deploy the AuditSink object for your audit policy and webhook configuration
1. Observe K8s audit events at falco
### Deploy Falco to your K8s cluster
Follow the [K8s Using Daemonset](../../integrations/k8s-using-daemonset/README.md) instructions to create a falco service account, service, configmap, and daemonset.
### Restart the API Server to enable Audit Logging
A script [enable-k8s-audit.sh](./enable-k8s-audit.sh) performs the necessary steps of enabling dynamic audit support for the apiserver by modifying the apiserver command line to add `--audit-dynamic-configuration`, `--feature-gates=DynamicAuditing=true`, etc. arguments, etc. (For minikube, ideally you'd be able to pass all these options directly on the `minikube start` command line, but manual patching is necessary. See [this issue](https://github.com/kubernetes/minikube/issues/2741) for more details.)
It is run as `bash ./enable-k8s-audit.sh <variant> dynamic`. `<variant>` can be one of the following:
* "minikube"
* "kops"
When running with variant="kops", you must either modify the script to specify the kops apiserver hostname or set it via the environment: `APISERVER_HOST=api.my-kops-cluster.com bash ./enable-k8s-audit.sh kops`
Its output looks like this:
```
$ bash enable-k8s-audit.sh minikube dynamic
***Copying apiserver config patch script to apiserver...
apiserver-config.patch.sh 100% 1190 1.2MB/s 00:00
***Modifying k8s apiserver config (will result in apiserver restarting)...
***Done!
$
```
## Observe K8s audit events at falco
### Deploy AuditSink objects
[audit-sink.yaml.in](./audit-sink.yaml.in), in this directory, is a template audit sink configuration that defines the dynamic audit policy and webhook to route k8s audit events to Falco.
Run the following to fill in the template file with the ClusterIP ip address you created with the `falco-service` service above. Although services like `falco-service.default.svc.cluster.local` can not be resolved from the kube-apiserver container within the minikube vm (they're run as pods but not *really* a part of the cluster), the ClusterIPs associated with those services are routable.
```
FALCO_SERVICE_CLUSTERIP=$(kubectl get service falco-service -o=jsonpath={.spec.clusterIP}) envsubst < audit-sink.yaml.in > audit-sink.yaml
```
### Observe K8s audit events at falco
K8s audit events will then be routed to the falco daemonset within the cluster, which you can observe via `kubectl logs -f $(kubectl get pods -l app=falco-example -o jsonpath={.items[0].metadata.name})`.

View File

@ -4,10 +4,18 @@ IFS=''
FILENAME=${1:-/etc/kubernetes/manifests/kube-apiserver.yaml}
VARIANT=${2:-minikube}
AUDIT_TYPE=${3:-static}
if grep audit-webhook-config-file $FILENAME ; then
echo audit-webhook patch already applied
exit 0
if [ $AUDIT_TYPE == "static" ]; then
if grep audit-webhook-config-file $FILENAME ; then
echo audit-webhook patch already applied
exit 0
fi
else
if grep audit-dynamic-configuration $FILENAME ; then
echo audit-dynamic-configuration patch already applied
exit 0
fi
fi
TMPFILE="/tmp/kube-apiserver.yaml.patched"
@ -26,19 +34,29 @@ do
echo "$LINE" >> "$TMPFILE"
case "$LINE" in
*$APISERVER_LINE*)
echo "$APISERVER_PREFIX --audit-log-path=/var/lib/k8s_audit/audit.log" >> "$TMPFILE"
echo "$APISERVER_PREFIX --audit-policy-file=/var/lib/k8s_audit/audit-policy.yaml" >> "$TMPFILE"
echo "$APISERVER_PREFIX --audit-webhook-config-file=/var/lib/k8s_audit/webhook-config.yaml" >> "$TMPFILE"
echo "$APISERVER_PREFIX --audit-webhook-batch-max-wait=5s" >> "$TMPFILE"
if [ $AUDIT_TYPE == "static" ]; then
echo "$APISERVER_PREFIX --audit-log-path=/var/lib/k8s_audit/audit.log" >> "$TMPFILE"
echo "$APISERVER_PREFIX --audit-policy-file=/var/lib/k8s_audit/audit-policy.yaml" >> "$TMPFILE"
echo "$APISERVER_PREFIX --audit-webhook-config-file=/var/lib/k8s_audit/webhook-config.yaml" >> "$TMPFILE"
echo "$APISERVER_PREFIX --audit-webhook-batch-max-wait=5s" >> "$TMPFILE"
else
echo "$APISERVER_PREFIX --audit-dynamic-configuration" >> "$TMPFILE"
echo "$APISERVER_PREFIX --feature-gates=DynamicAuditing=true" >> "$TMPFILE"
echo "$APISERVER_PREFIX --runtime-config=auditregistration.k8s.io/v1alpha1=true" >> "$TMPFILE"
fi
;;
*"volumeMounts:"*)
echo " - mountPath: /var/lib/k8s_audit/" >> "$TMPFILE"
echo " name: data" >> "$TMPFILE"
if [ $AUDIT_TYPE == "static" ]; then
echo " - mountPath: /var/lib/k8s_audit/" >> "$TMPFILE"
echo " name: data" >> "$TMPFILE"
fi
;;
*"volumes:"*)
echo " - hostPath:" >> "$TMPFILE"
echo " path: /var/lib/k8s_audit" >> "$TMPFILE"
echo " name: data" >> "$TMPFILE"
if [ $AUDIT_TYPE == "static" ]; then
echo " - hostPath:" >> "$TMPFILE"
echo " path: /var/lib/k8s_audit" >> "$TMPFILE"
echo " name: data" >> "$TMPFILE"
fi
;;
esac

View File

@ -0,0 +1,15 @@
apiVersion: auditregistration.k8s.io/v1alpha1
kind: AuditSink
metadata:
name: falco-audit-sink
spec:
policy:
level: RequestResponse
stages:
- ResponseComplete
webhook:
throttle:
qps: 10
burst: 15
clientConfig:
url: "http://$FALCO_SERVICE_CLUSTERIP:8765/k8s_audit"

View File

@ -3,6 +3,7 @@
set -euo pipefail
VARIANT=${1:-minikube}
AUDIT_TYPE=${2:-static}
if [ $VARIANT == "minikube" ]; then
APISERVER_HOST=$(minikube ip)
@ -23,14 +24,18 @@ if [ $VARIANT == "kops" ]; then
fi
fi
echo "***Copying audit policy/webhook files to apiserver..."
echo "***Copying apiserver config patch script to apiserver..."
ssh -i $SSH_KEY $SSH_USER@$APISERVER_HOST "sudo mkdir -p /var/lib/k8s_audit && sudo chown $SSH_USER /var/lib/k8s_audit"
scp -i $SSH_KEY audit-policy.yaml $SSH_USER@$APISERVER_HOST:/var/lib/k8s_audit
scp -i $SSH_KEY webhook-config.yaml $SSH_USER@$APISERVER_HOST:/var/lib/k8s_audit
scp -i $SSH_KEY apiserver-config.patch.sh $SSH_USER@$APISERVER_HOST:/var/lib/k8s_audit
if [ $AUDIT_TYPE == "static" ]; then
echo "***Copying audit policy/webhook files to apiserver..."
scp -i $SSH_KEY audit-policy.yaml $SSH_USER@$APISERVER_HOST:/var/lib/k8s_audit
scp -i $SSH_KEY webhook-config.yaml $SSH_USER@$APISERVER_HOST:/var/lib/k8s_audit
fi
echo "***Modifying k8s apiserver config (will result in apiserver restarting)..."
ssh -i $SSH_KEY $SSH_USER@$APISERVER_HOST "sudo bash /var/lib/k8s_audit/apiserver-config.patch.sh $MANIFEST $VARIANT"
ssh -i $SSH_KEY $SSH_USER@$APISERVER_HOST "sudo bash /var/lib/k8s_audit/apiserver-config.patch.sh $MANIFEST $VARIANT $AUDIT_TYPE"
echo "***Done!"