Apache NiFi Sensitive Properties: All you need to know

In Apache NiFi, Sensitive Properties as the name indicates are the processor specific properties which should not be available/visible to everyone. In this article, we will see how Apache NiFi caters to these values, how they are stored, what we can do to make it even more secure and all other things we need to know about sensitive properties in Apache NiFi.

If you are new to Apache NiFi and would like to get an introduction to NiFi, check out the article Apache NiFi: An Introduction

1. What is a Sensitive Property?

Apache NiFi processors sometimes need to store some sensitive values like a password, an access token, webhook URL etc. All these types of properties are marked as sensitive values and once we type in the value in the processor properties UI and saves, they will no longer be visible.

Sensitive Property Values

Sensitive Property Values

Now, these values are obviously stored in the flow.xml.gz file so that it can be used when required. To make is secure these values are encrypted


2. How Sensitive Properties are Encrypted?

So now we know that these sensitive values are encrypted and decrypted but how this is done is the question.

PBEWITHMD5AND256BITAES-CBC-OPENSSL is default encryption algorithm which is used. This means that actual cipher used is AES with CBC mode of operation. The default algorithm, as well as the key, are stored in the nifi.properties file under the property names nifi.sensitive.props.algorithm and nifi.sensitive.props.key

nifi.properties file with the encryption properties

nifi.properties file with the encryption properties

The value of the key is used to drive a 256-bit key using Open-SSL EVP_BytesToJey key derivation function(KDF). If the key property is not set in the properties file (as it is in the screenshot above), then a default value will be used by the encryption function.

Once the key is derived, a salt is generated, cipher text is available and the sensitive value is encrypted; it is then followed by concatenating the encrypted value, salt, and cipher and then finally encoding them in hexadecimal. This final value is then stored in flow.xml.gz and is wrapped in “enc{}” so that is will be easily identifiable as encrypted value.

The encryption/decryption process is performed by the core NiFi framework and is not accessible to any extension code. These values are never returned to client side ever again.

3. Setting-up/Migrating encryption key

It is highly advised that the keys are set in the properties file and default key should not be used especially in the production environment. So we will see how to set this encryption key:

If the NiFi instance is freshly installed and there are no dataflow on the canvas yet. Then we can just go ahead and edit nifi.properties file and add a value for the property nifi.sensitive.props.key. For new instance of Apache NiFi, it is as easy as this.

But if this is not the case and there are already dataflows on NiFi canvas, then directly editing in the properties file will not work for existing flows as the sensitive properties in those flows are already encrypted using the default values. In that case, we will need to migrate the existing flow in order to change the key. For this purpose, there is a encrypt-config.sh script available in the nifi-toolkit. This script will help us in migrating and changing to the new key.

This utility will read existing flows in flow.xml.gz and decrypt the sensitive values using the key mentioned in the properties file (or default values if no key value is provided in the properties file). Once decrypted, it encrypts all the values with the new key provided and overwrites the existing flow.xml.gz or create a new one depending on what parameters are passed.

Following are the steps to follow:

  1. First of all, we will need the NiFi toolkit on our system. If you have compiled Apache NiFi from source. It is located in the nifi-toolkit/nifi-toolkit-assembly/target/nifi-toolkit-VERSION-bin/nifi-toolkit-VERSION/ directory. There the encrypt-config.sh script is in the bin directory. But if you have not built it from source, you can also download the toolkit separately from the Apache NiFi Download Page

    Download Toolkit

    Download Toolkit

  2. Once the toolkit is download, unzip it and we are ready to use the command-line utility.
  3. Go to the toolkit directory in the command prompt and use the following command

    Migrating the encryption key

    Migrating the encryption key

    Note: All the parameters are self explanatory in the above command except -x. This is the mandatory parameter in this command as we are telling the utility to encrypt only the sensitive properties in flow.xml.gz and not to decrpyt/encrypt the nifi.properties entries. This is also a good pratice to encrypt the proeprty values which we will lean in the next article.

  4. We can verify the result by opening the nifi.properties file where we will find the nifi.sensitive.props.key is set to testpassword which we passed as a parameter of the command line utility.

    nifi.properties file

    nifi.properties file

  5. Using the utility above will result into overwriting the flow.xml.gz and nifi.properties files. But there might be some cases when we need the new files to be written somewhere else and not overwriting the old files. We can use the same command for that but we need to pass two more parameters telling where we want to store the newly created files.

    This will result in new files being created instead of old once being overwritten as shown in the screenshot below:

    Newly created files

    Newly created files

4. Conclusion

In this article, we understood what sensitive properties are, how they are stored and secured using encryption. How the default values are stored. Followed by how we can set up the encryption keys to make it even more secure and how to migrate the flows using the old keys with new encryption keys. In the coming articles, we will also have a look at securing the properties stored in nifi.properties file which at present is just plaintext.



Enter your email address below to subscribe to our newsletter.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.