awatertrevi/infisical
1
0
Fork 0
mirror of https://github.com/awatertrevi/infisical.git synced 2026-09-22 13:39:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update hsm-integration.mdx

Browse Source
This commit is contained in:
Daniel Hougaard
2025-10-13 23:42:56 +04:00
parent 9e60350b89
commit 96ebe558ba
1 changed files with 354 additions and 165 deletions
Download Patch File Download Diff File Expand all files Collapse all files

519
docs/documentation/platform/kms/hsm-integration.mdx
View File

@@ -999,157 +999,135 @@ Enabling HSM encryption has a set of key benefits:
<Steps>
<Step title="Configure CloudHSM client">
Before using the CloudHSM client with Kubernetes, it must be configured properly so Infisical can use it for cryptographic operations.
**1. Download and configure the AWS CloudHSM client**
You can download the AWS CloudHSM client from [the AWS documentation](https://docs.aws.amazon.com/cloudhsm/latest/userguide/pkcs11-library-install.html).
<Note>
Note that the AWS CloudHSM client is only available for Linux and Windows.
If you're on a different operating system, you'll need to access a Linux machine to configure the client, such as an AWS EC2 Debian instance.
</Note>
After installing the CloudHSM client, you should see all related files in the `/opt/cloudhsm/` directory on your machine.
You need to run the `configure-pkcs11` binary which will configure the client to connect with your AWS CloudHSM cluster. Depending on if you have multiple HSM's inside your cluster, you'll need to run the command with different arguments. Below you'll find the appropriate command for your use case:
<AccordionGroup>
<Accordion title="Single HSM">
```bash
sudo /opt/cloudhsm/bin/configure-pkcs11 -a <HSM_ENI_IPV4_ADDRESS> --disable-key-availability-check
```
<Info>
To use a single HSM, you must first manage client key durability settings by setting `disable_key_availability_check` to true by passing the `--disable-key-availability-check` flag. For more information read the [Key Synchronization](https://docs.aws.amazon.com/cloudhsm/latest/userguide/manage-key-sync.html) section in the AWS CloudHSM documentation.
</Info>
</Accordion>
<Accordion title="Multiple HSM's">
```bash
sudo /opt/cloudhsm/bin/configure-pkcs11 -a <HSM_ENI_IPV4_ADDRESS_1> <HSM_ENI_IPV4_ADDRESS_2> ... --disable-key-availability-check
```
</Accordion>
</AccordionGroup>
At this point you should have:
1. [Activated the CloudHSM cluster](https://docs.aws.amazon.com/cloudhsm/latest/userguide/activate-cluster.html)
2. [Created a Crypto User HSM user](https://docs.aws.amazon.com/cloudhsm/latest/userguide/cloudhsm_cli-user-create.html)
3. Downloaded and configured the CloudHSM client as described in the previous steps.
**2. Verify the configuration file**
After configuring the CloudHSM client, you should notice that the PKCS11 configuration file has been updated to include the HSM's ENI IP address. You can find this file in the `/opt/cloudhsm/etc/cloudhsm-pkcs11.cfg` directory, and it should look like this:
```json cloudhsm-pkcs11.cfg
{
"clusters": [
{
"type": "hsm1",
"cluster": {
// Your issuing CA certificate.
// As per AWS documentation, this defaults to `/opt/cloudhsm/etc/customerCA.crt`.
"hsm_ca_file": "/opt/cloudhsm/etc/customerCA.crt",
"servers": [
{
"hostname": "<HSM_ENI_IPV4_ADDRESS_1>",
"port": 2223,
"enable": true
},
{
"hostname": "<HSM_ENI_IPV4_ADDRESS_2>",
"port": 2223,
"enable": true
}
],
// Only relevant if you passed the --disable-key-availability-check flag
"options": {
"disable_key_availability_check": true
}
}
}
],
"logging": {
"log_type": "file",
"log_file": "/opt/cloudhsm/run/cloudhsm-pkcs11.log",
"log_level": "info",
"log_interval": "daily"
}
}
```
**3. Copy the CloudHSM client files to a staging directory**
Create a directory to stage the CloudHSM client files for Kubernetes:
```bash
mkdir -p /etc/cloudhsm-k8s
```
Copy the entire `/opt/cloudhsm` directory to your staging location:
```bash
cp -r /opt/cloudhsm/* /etc/cloudhsm-k8s/
```
Ensure the configuration file paths are correct. The `cloudhsm-pkcs11.cfg` file should reference `/opt/cloudhsm/etc/customerCA.crt` as shown above, since this is where files will be mounted inside the Kubernetes container.
</Step>
<Step title="Find HSM slot number">
On the same machine that you configured the CloudHSM client, you can use `pkcs11-tool` to find the HSM slot number and to verify that the client is working correctly.
First, install the `pkcs11-tool` package:
```bash
sudo apt-get install opensc -y
```
Then, run the following command to find the HSM slot number:
```bash
pkcs11-tool --module /opt/cloudhsm/lib/libcloudhsm_pkcs11.so --list-slots --login
```
It'll prompt you to log in with your PIN, which is your username and password separated by a colon. Example: `testUser:testPassword`.
This will output the HSM slot number like so:
```bash
ubuntu@ec-2:~$ pkcs11-tool --module /opt/cloudhsm/lib/libcloudhsm_pkcs11.so --list-slots
Available slots:
Slot 0 (0x2000000000000001): hsm1
token label : hsm1
token manufacturer : Marvell Semiconductors, Inc.
token model : LS2
token flags : login required, rng, token initialized
hardware version : 66.48
firmware version : 10.2
serial num :
pin min/max : 8/32
```
In this case we see that the HSM has a slot in the position of `0`. This slot number will be used in the later steps to set the `HSM_SLOT` environment variable.
</Step>
<Step title="Obtain the HSM issuing CA certificate">
When you initialized your HSM, you were prompted to download the cluster CSR and sign it.
In order to use the HSM with Infisical, you need to obtain the issuer CA certificate that was used to sign the cluster CSR.
If you followed [the official AWS documentation](https://docs.aws.amazon.com/cloudhsm/latest/userguide/initialize-cluster.html), you should have a CA certificate called `customerCA.crt`.
Copy the CA certificate to your staging directory:
```bash
cp /path/to/customerCA.crt /etc/cloudhsm-k8s/etc/customerCA.crt
```
Ensure the file is at `/etc/cloudhsm-k8s/etc/customerCA.crt` as this is what the configuration file expects.
</Step>
<Step title="Creating Persistent Volume Claim (PVC)">
<Accordion title="Prerequisites for using AWS EKS">
If you're using AWS EKS, you need to specify a storage class for the PVC and ensure that the EBS CSI Driver is installed and running.
By default, EKS exposes `gp2` as the default storage class. Below are the steps required for setting the default storage class and ensuring the EBS CSI Driver is installed and running:
<Steps>
<Step title="Enable OIDC authentication">
Enable OIDC authentication for the EKS cluster:
```bash
eksctl utils associate-iam-oidc-provider \
--region <your-region> \
--cluster <your-cluster-name> \
--approve
```
* Replace `<your-region>` with your AWS region.
* Replace `<your-cluster-name>` with your cluster name.
</Step>
<Step title="Ensure the EBS CSI Driver is installed and running">
1. Check if EBS CSI Driver is installed and running by running the following command:
```bash
kubectl get pods -n kube-system | grep ebs-csi
```
If you see no pods, you need to install the EBS CSI Driver as seen in the next step.
</Step>
<Step title="Install EBS CSI Driver using eksctl">
Create a new IAM service account for the EBS CSI Driver:
```bash
eksctl create iamserviceaccount \
--name ebs-csi-controller-sa \
--namespace kube-system \
--region <your-region> \
--cluster <your-cluster-name> \
--attach-policy-arn arn:aws:iam::aws:policy/service-role/AmazonEBSCSIDriverPolicy \
--approve \
--role-only \
--role-name AmazonEKS_EBS_CSI_DriverRole
```
* Replace `<your-cluster-name>` with your cluster name.
* Replace `<your-region>` with your AWS region.
Install the EBS CSI Driver:
```bash
eksctl create addon \
--name aws-ebs-csi-driver \
--cluster <your-cluster-name> \
--region <your-region> \
--service-account-role-arn arn:aws:iam::<account-id>:role/AmazonEKS_EBS_CSI_DriverRole \
--force
```
* Replace `<your-cluster-name>` with your cluster name.
* Replace `<your-region>` with your AWS region.
* Replace `<account-id>` with your actual account ID. Can be obtained by running `aws sts get-caller-identity --query Account --output text`.
</Step>
<Step title="Verify the EBS CSI Driver is installed and running">
Verify the EBS CSI Driver is installed and running by running the following command:
```bash
kubectl get pods -n kube-system | grep ebs-csi
```
You should see an output like this:
```bash
kubectl get pods -n kube-system | grep ebs-csi
ebs-csi-controller-6b6bbf996-rvf8r 6/6 Running 0 21s
ebs-csi-controller-6b6bbf996-vk4ng 6/6 Running 0 21s
ebs-csi-node-c6vbb 3/3 Running 0 21s
ebs-csi-node-s9zlr 3/3 Running 0 21s
```
</Step>
<Step title="Find the enabled storage class">
You can find the enabled storage class by running the following command:
```bash
kubectl get storageclass
```
You should see an output like this:
```bash
$ kubectl get storageclass
NAME PROVISIONER RECLAIMPOLICY VOLUMEBINDINGMODE ALLOWVOLUMEEXPANSION AGE
gp2 kubernetes.io/aws-ebs Delete WaitForFirstConsumer false 65m
```
In this case, the enabled storage class is `gp2`.
</Step>
<Step title="Set the default storage class">
You can set the default PVC storage class by patching the storage class with the following command:
```bash
kubectl patch storageclass gp2 -p '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'
```
This will set the `gp2` storage class as the default storage class.
Now when you run `kubectl get storageclass`, you should see that `gp2` is the default storage class.
```bash
$ kubectl get storageclass
NAME PROVISIONER RECLAIMPOLICY VOLUMEBINDINGMODE ALLOWVOLUMEEXPANSION AGE
gp2 (default) kubernetes.io/aws-ebs Delete WaitForFirstConsumer false 68m
```
Notice the `(default)` next to the `gp2` storage class.
</Step>
</Steps>
</Accordion>
You need to create a Persistent Volume Claim (PVC) to mount the HSM client files to the Infisical deployment.
```bash
@@ -1179,8 +1157,8 @@ Enabling HSM encryption has a set of key benefits:
spec:
containers:
- name: setup
image: busybox
command: ["/bin/sh", "-c", "sleep 3600"]
image: debian:bookworm
command: ["/bin/sh", "-c", "sleep 7200"]
volumeMounts:
- name: cloudhsm-data
mountPath: /data
@@ -1191,27 +1169,226 @@ Enabling HSM encryption has a set of key benefits:
EOF
```
The above command will create a pod named `cloudhsm-setup-pod` with a busybox image. The pod will sleep for 3600 seconds _(one hour)_, which is enough time to upload the HSM client files to the PVC.
The above command will create a pod named `cloudhsm-setup-pod` with a Debian image. The pod will sleep for 7200 seconds _(two hours)_, which is enough time to set up the PVC and configure the HSM client.
Ensure that the pod is running and is healthy by running the following command:
```bash
kubectl wait --for=condition=Ready pod/cloudhsm-setup-pod --timeout=60s
kubectl wait --for=condition=Ready pod/cloudhsm-setup-pod --timeout=120s
```
</Step>
Next we need to copy the HSM client files into the PVC.
<Step title="Configure the PVC">
```bash
kubectl exec cloudhsm-setup-pod -- mkdir -p /data/ # Create the data directory
kubectl cp /etc/cloudhsm-k8s/. cloudhsm-setup-pod:/data/ # Copy the HSM client files into the PVC
kubectl exec cloudhsm-setup-pod -- chmod -R 755 /data/ # Set the correct permissions for the HSM client files
```
We need to configure the PVC to work with the CloudHSM, so Infiscial can consume the HSM client files.
Finally, we are ready to delete the temporary pod, as we have successfully uploaded the HSM client files to the PVC. This step may take a few minutes to complete.
**2.1. Start a shell in the PVC pod:**
This will allow us to run commands directly within the setup pod. We'll use this to configure the CloudHSM client and to validate that it's working correctly.
```bash
kubectl exec -it cloudhsm-setup-pod -- /bin/sh
```
**2.2. Create the data directory:**
This will create a directory called `/data/` within the setup pod. This directory will remain even after the setup pod is deleted, as it's mounted as a volume from the PVC.
```bash
mkdir -p /data/
```
**2.3. Install the necessary packages:**
This will install the necessary packages to allow us to test and install the CloudHSM client.
```bash
apt-get update -y
apt-get install opensc telnet wget -y
```
**2.4. Try to reach the HSM device:**
We need to validate that we're able to reach the HSM device from within Kubernetes. You can use telnet to ping the HSM device like so:
```bash
telnet <HSM_ENI_IPV4_ADDRESS> 2223
```
You should see an output like this:
```bash
$ telnet <HSM_ENI_IPV4_ADDRESS> 2223
Trying <HSM_ENI_IPV4_ADDRESS>...
Connected to <HSM_ENI_IPV4_ADDRESS>.
```
If it gets stuck on `Trying ....`, you may have configured your HSM client's security group incorrectly. Make sure you configure the security group to allow traffic from EKS on port 2223-2225.
**2.5. Install the AWS CloudHSM client:**
The Infisical images run on Debian, so we need to install a Debian-compatible version of the AWS CloudHSM client.
```bash
wget https://s3.amazonaws.com/cloudhsmv2-software/CloudHsmClient/Jammy/cloudhsm-pkcs11_latest_u22.04_amd64.deb
apt-get install ./cloudhsm-pkcs11_latest_u22.04_amd64.deb -y
```
**2.6. Configure the CloudHSM client:**
After installing the CloudHSM client, you should see all related files in the `/opt/cloudhsm/` directory on the CloudHSM setup pod.
You need to run the `configure-pkcs11` binary which will configure the client to connect with your AWS CloudHSM cluster. Depending on if you have multiple HSM's inside your cluster, you'll need to run the command with different arguments. Below you'll find the appropriate command for your use case:
<AccordionGroup>
<Accordion title="Single HSM">
```bash
/opt/cloudhsm/bin/configure-pkcs11 -a <HSM_ENI_IPV4_ADDRESS> --disable-key-availability-check
```
<Info>
To use a single HSM, you must first manage client key durability settings by setting `disable_key_availability_check` to true by passing the `--disable-key-availability-check` flag. For more information read the [Key Synchronization](https://docs.aws.amazon.com/cloudhsm/latest/userguide/manage-key-sync.html) section in the AWS CloudHSM documentation.
</Info>
</Accordion>
<Accordion title="Multiple HSM's">
```bash
/opt/cloudhsm/bin/configure-pkcs11 -a <HSM_ENI_IPV4_ADDRESS_1> <HSM_ENI_IPV4_ADDRESS_2> ... --disable-key-availability-check
```
</Accordion>
</AccordionGroup>
**2.7. Verify the CloudHSM client is configured correctly:**
You can verify the CloudHSM client is configured correctly by running the following command:
```bash
cat /opt/cloudhsm/etc/cloudhsm-pkcs11.cfg
```
You should see an output like this:
```json
{
"clusters": [
{
"type": "hsm1",
"cluster": {
"hsm_ca_file": "/opt/cloudhsm/etc/customerCA.crt",
"servers": [
{
"hostname": "172.31.39.155",
"port": 2223,
"enable": true
}
],
"options": {
"disable_key_availability_check": true
}
}
}
],
"logging": {
"log_type": "file",
"log_file": "/opt/cloudhsm/run/cloudhsm-pkcs11.log",
"log_level": "info",
"log_interval": "daily"
}
}
```
**2.8. Exit the pod:**
Exit the pod by running the following command:
```bash
exit
```
**2.9. Copy your issuer CA certificate to the PVC:**
When you initialized your HSM, you were prompted to download the cluster CSR and sign it.
In order to use the HSM with Infisical, you need to obtain the issuer CA certificate that was used to sign the cluster CSR.
If you followed [the official AWS documentation](https://docs.aws.amazon.com/cloudhsm/latest/userguide/initialize-cluster.html), you should have a CA certificate called `customerCA.crt`.
Copy the CA certificate from your local machine to the setup pod:
```bash
kubectl cp /path/to/customerCA.crt cloudhsm-setup-pod:/opt/cloudhsm/etc/customerCA.crt
```
Ensure that the file is at `/opt/cloudhsm/etc/customerCA.crt` inside the setup pod by running the following command:
```bash
kubectl exec -it cloudhsm-setup-pod -- cat /opt/cloudhsm/etc/customerCA.crt
```
**2.10. Test the HSM client:**
Finally, after we're done configuring the HSM client, we need to test it to ensure that it's working correctly.
First, start a new shell into the setup pod by running the same shell command as before:
```bash
kubectl exec -it cloudhsm-setup-pod -- /bin/sh
```
Next, try generating a random 32 bytes long string by running the following command:
```bash
pkcs11-tool --module /opt/cloudhsm/lib/libcloudhsm_pkcs11.so \
--login --pin <crypto-user-username>:<crypto-user-password> \
--generate-random 32 | base64
```
You should see an output like this:
```bash
Using slot 0 with a present token (0x2000000000000001)
av1dlhVEsssjpcTNS+ysGUoKWH6+/PCaEDIdal5oQc0=
```
**2.12. Copy the configured client to the PVC:**
Copy from the HSM files into the `/data` directory in the PVC, which is what will be mounted for the Infisical deployment.
```bash
cp -r /opt/cloudhsm/. /data/
```
Verify the files were copied correctly by running the following command:
```bash
ls -la /data/
```
You should see an output like this:
```bash
drwxr-xr-x. 8 root root 4096 Oct 13 18:50 .
drwxr-xr-x. 1 root root 131 Oct 13 18:29 ..
drwxr-xr-x. 2 root root 4096 Oct 13 18:50 bin
drwxr-xr-x. 3 root root 4096 Oct 13 18:50 doc
drwxr-xr-x. 2 root root 4096 Oct 13 18:50 etc
drwxr-xr-x. 3 root root 4096 Oct 13 18:50 include
drwxr-xr-x. 2 root root 4096 Oct 13 18:50 lib
drwxr-xr-t. 2 root root 4096 Oct 13 18:50 run
```
**2.13. Set the correct permissions for the HSM client files:**
```bash
chmod -R 755 /data/
```
**2.14. Exit the pod:**
Exit the pod by running the following command:
```bash
exit
```
**2.15. Delete the setup pod:**
Delete the setup pod by running the following command:
```bash
kubectl delete pod cloudhsm-setup-pod
```
```bash
kubectl delete pod cloudhsm-setup-pod
```
</Step>
<Step title="Updating your environment variables">
@@ -1304,4 +1481,16 @@ To disable HSM encryption, navigate to Infisical's Server Admin Console and set
<Note>
In order to disable HSM encryption, the Infisical instance must be able to access the HSM device. If the HSM device is no longer accessible, you will not be able to disable HSM encryption.
</Note>
</Note>
## Troubleshooting
<AccordionGroup>
<Accordion title="Failing to create PVC for Kubernetes on EKS: no persistent volumes available for this claim and no storage class is set">
If you're using EKS, you may encounter an error like the following: `no persistent volumes available for this claim and no storage class is set`
</Accordion>
</AccordionGroup>