7 minute read.Last Modified 2021-07-23 12:49 EDT
TrueNAS supports different encryption options for critical data.
Users are responsible for backing up and securing encryption keys and passphrases! Losing the ability to decrypt data is similar to a catastrophic data loss.
Data-at-rest encryption is available with:
- Self Encrypting Drives (SEDs) using OPAL or FIPS 140.2 (Both AES 256)
- Encryption of specific datasets (AES-256-GCM in TrueNAS 12.0)
Keys for data-at-rest are managed on the local TrueNAS system. The user is responsible for storing and securing their keys. The Key Management Interface Protocol (KMIP) is included in TrueNAS 12.0.
Always consider the following drawbacks/considerations when encrypting data:
- Losing encryption keys and passwords means losing your data.
- Unrelated encrypted datasets do not support deduplication.
- We do not recommend using GELI or ZFS encryption with deduplication because of the sizable performance impact.
- Be cautious when using many encryption and deduplication features at once since they will all be competing for the same CPU cycles.
Encrypting the root dataset of a new storage pool further increases data security. Create a new pool and set Encryption in the Pool Manager. TrueNAS shows a warning.
Read the warning, set Confirm, and click I Understand.
The default encryption Cipher is recommended, but there are other ciphers available.
New datasets within an existing storage pool can also be encrypted without having to encrypt the entire pool. To encrypt a single dataset, go to Storage > Pools, open the more_vert for an existing dataset, and click Add Dataset.
Look at the Encryption Options and, when the parent dataset is unencrypted, unset Inherit and set Encryption.
Now choose which Type of authentication to use: a Key or a Passphrase. The remaining options are the same as a new pool. Datasets with encryption enabled show additional icons in the Storage > Pools list.
The dataset status is determined from an icon:
- The dataset unlocked icon: lock_open.
- The dataset locked icon: lock.
- A Dataset on an encrypted pool with encryption properties that don’t match the root dataset have this icon: .
NOTE: An unencrypted pool with an encrypted dataset will also show this icon:.
Encrypted datasets can only be locked and unlocked if they are secured with a passphrase instead of a keyfile. Before locking a dataset, verify that it is not currently in use, then click (Options) and Lock.
Use the Force unmount option only if you are certain that no one is currently accessing the dataset. After locking a dataset, the unlock icon changes to a locked icon. While the dataset is locked, it is not available for use.
To unlock a dataset, click more_vert and Unlock.
Enter the passphrase and click Submit. If there are child datasets that are locked with the same passphrase you can unlock them all at the same time by setting Unlock Children. Confirm unlocking the datasets and wait for a dialog to confirm the unlock is successful.
The dataset listing changes to show the unlocked icon.
There are two ways to manage the encryption credentials: with Key Files or Passphrases:
Creating a new encrypted pool automatically generates new key file and prompts to download it. Always back up the key file to a safe and secure location.
To manually back up a root dataset keyfile by opening the pool settings menu and selecting Export Dataset Keys.
To change the key, click the dataset more_vert and Encryption Options.
Enter your custom key or click Generate Key.
To use a passphrase instead of a keyfile, click the dataset more_vert and Encryption Options. Change the Encryption Type from Key to Passphrase.
Set the rest of the options:
- Passphrase : User-defined string used to decrypt the dataset.
Can be used instead of an encryption key.
Must be longer than 8 characters.
The passphrase is the only means to decrypt the information stored in this dataset. Be sure to create a memorable passphrase or physically secure the passphrase.
- pbkdf2iters : Number of password-based key derivation function 2 (PBKDF2) iterations to use for reducing vulnerability to brute-force attacks. Entering a number larger than 100000 is required.
TrueNAS Enterprise users may connect a Key Management Interoperability Protocol (KMIP) server to centralize keys when passphrases are not used to unlock a dataset or zvol.
Users with TrueNAS CORE or Enterprise installations without KMIP should either replicate the dataset or zvol without properties to disable encryption at the remote end, or construct a special json manifest to unlock each child dataset/zvol with a unique key.
- Replicate every encrypted dataset you want to replicate with properties.
- Export key for every child dataset which has a unique key.
- For each child dataset construct a proper json with poolname/datasetname of the destination system and key from the source system like this:
- Save this file with the extension
- On remote system unlock the dataset(s) using properly constructed
Uncheck properties when replicating so that the destination dataset will not be encrypted on the remote side and will not require a key to unlock.
- Go to Tasks > Replication Tasks and click ADD.
- Click ADVANCED REPLICATION CREATION.
- Fill out the form as needed and make sure Include Dataset Properties is NOT checked.
- Click SUBMIT.
NOTE: This does not affect TrueNAS Enterprise installs with KMIP.
GELI encryption is deprecated in TrueNAS and no longer supported.
Data can be migrated from the GELI-encrypted pool to a new ZFS-encrypted pool. Be sure to unlock the GELI-encrypted pool before attempting any data migrations. The new ZFS-encrypted pool must be at least the same size as the previous GELI-encrypted pool. Do not delete the GELI dataset until you have verified the data migration.
There are a few options to migrate data from a GELI-encrypted pool to a new ZFS-encrypted pool:
This method does not preserve file ACLs.
The web interface supports using Tasks > Rsync Tasks to transfer files out of the GELI pool.
In the Shell,
rsync and other file transfer mechanisms (
rdiff-backup) are available for copying data between pools.
These instructions are an example walkthrough. It is not an exact step-by-step guide for all situations. Research ZFS send/receive before attempting this. There are many edge cases that cannot be covered by a simple example.
GELI Pool = pool_a Origin Dataset = dataset_1 Latest Snapshot of GELI Pool = snapshot_name ZFS Native Encrypted Pool = pool_b Receieving Dataset = dataset_2
- Create a new encrypted pool in Storage > Pools.
- Open the Shell.
Make a new snapshot of the GELI pool and dataset with the data to be migrated:
zfs snapshot -r pool_a/dataset_1@snapshot_name.
- Create a passphrase:
echo passphrase > /tmp/pass.
- Use ZFS send/receive to transfer the data between pools:
zfs send -Rv pool_a/dataset_1@snapshot_name | zfs recv -o encryption=on -o keyformat=passphrase -o keylocation=file:///tmp/pass pool_b/dataset_2.
- When the transfer is complete, go to Storage > Pools and lock the new dataset.
After locking the dataset, immediately unlock it.
TrueNAS prompts for the passphrase.
After entering the passphrase and the pool is unlocked, you can delete the
/tmp/passfile used for the transfer.
- If desired, you can convert the dataset to use a keyfile instead of a passphrase. To use a key file, click the dataset (Options) and click Encryption Options. Change the Encryption Type from Passphrase to Key and save. Back up your key file immediately!
- Repeat this process for every dataset in the pool that needs to be migrated.