Warning

Multi-node support has been deprecated with TimescaleDB 2.13. This is the last version that includes multi-node support and it supports multi-node installations with PostgreSQL versions 13, 14, and 15.

Learn more about it here.

When you have your instances set up, you need to configure them to accept connections from the access node to the data nodes. The authentication mechanism you choose for this can be different than the one used by external clients to connect to the access node.

How you set up your multi-node cluster depends on which authentication mechanism you choose. The options are:

  • Trust authentication. This is the simplest approach, but also the least secure. This is a good way to start if you are trying out multi-node, but is not recommended for production clusters.
  • Pasword authentication. Every user role requires an internal password for establishing connections between the access node and the data nodes. This method is easier to set up than certificate authentication, but provides only a basic level of protection.
  • Certificate authentication. Every user role requires a certificate from a certificate authority to establish connections between the access node and the data nodes. This method is more complex to set up than password authentication, but more secure and easier to automate.
Important

Going beyond the simple trust approach to create a secure system can be complex, but it is important to secure your database appropriately for your environment. We do not recommend any one security model, but encourage you to perform a risk assessment and implement the security model that best suits your environment.

Trusting all incoming connections is the quickest way to get your multi-node environment up and running, but it is not a secure method of operation. Use this only for developing a proof of concept, do not use this method for production installations.

Warning

The trust authentication method allows insecure access to all nodes. Do not use this method in production. It is not a secure method of operation.

  1. Connect to the access node with psql, and locate the pg_hba.conf file:

    SHOW hba_file;
  2. Open the pg_hba.conf file in your preferred text editor, and add this line. In this example, the access node is located at IP 192.0.2.20 with a mask length of 32. You can add one of these two lines:

    # Using local loopback TCP/IP connections
    # TYPE DATABASE USER ADDRESS METHOD
    host all all 192.0.2.20/32 trust
    # The same as the previous line, but using a separate netmask column
    # TYPE DATABASE USER IP-ADDRESS IP-MASK METHOD
    host all all 192.0.2.20 255.255.255.255 trust
  3. At the command prompt, reload the server configuration:

    pg_ctl reload

    On some operating systems, you might need to use the pg_ctlcluster command instead.

  4. If you have not already done so, add the data nodes to the access node. For instructions, see the multi-node setup section.

  5. On the access node, create the trust role. In this example, we call the role testrole:

    CREATE ROLE testrole;

    OPTIONAL: If external clients need to connect to the access node as testrole, add the LOGIN option when you create the role. You can also add the PASSWORD option if you want to require external clients to enter a password.

  6. Allow the trust role to access the foreign server objects for the data nodes. Make sure you include all the data node names:

    GRANT USAGE ON FOREIGN SERVER <data node name>, <data node name>, ... TO testrole;
  7. On the access node, use the distributed_exec command to add the role to all the data nodes:

    CALL distributed_exec($$ CREATE ROLE testrole LOGIN $$);
Important

Make sure you create the role with the LOGIN privilege on the data nodes, even if you don't use this privilege on the access node. For all other privileges, ensure they are same on the access node and the data nodes.

Password authentication requires every user role to know a password before it can establish a connection between the access node and the data nodes. This internal password is only used by the access node and it does not need to be the same password as the client uses to connect to the access node. External users do not need to share the internal password at all, it can be set up and administered by the database administrator.

The access node stores the internal password so that it can verify the correct password has been provided by a data node. We recommend that you store the password on the access node in a local password file, and this section shows you how to set this up. However, if it works better in your environment, you can use user mappings to store your passwords instead. This is slightly less secure than a local pasword file, because it requires one mapping for each data node in your cluster.

This section sets up your password authentication using SCRAM SHA-256 password authentication. For other password authentication methods, see the PostgreSQL authentication documentation.

Before you start, check that you can use the postgres username to log in to your access node.

  1. On the access node, open the postgresql.conf configuration file, and add or edit this line:

    password_encryption = 'scram-sha-256' # md5 or scram-sha-256
  2. Repeat for each of the data nodes.

  3. On each of the data nodes, at the psql prompt, locate the pg_hba.conf configuration file:

    SHOW hba_file
  4. On each of the data nodes, open the pg_hba.conf configuration file, and add or edit this line to enable encrypted authentication to the access node:

    # IPv4 local connections:
    # TYPE DATABASE USER ADDRESS METHOD
    host all all 192.0.2.20 scram-sha-256 #where '192.0.2.20' is the access node IP
  5. On the access node, open or create the password file at data/passfile. This file stores the passwords for each role that the access node connects to on the data nodes. If you need to change the location of the password file, adjust the timescaledb.passfile setting in the postgresql.conf configuration file.

  6. On the access node, open the passfile file, and add a line like this for each user, starting with the postgres user:

    *:*:*:postgres:xyzzy #assuming 'xyzzy' is the password for the 'postgres' user
  7. On the access node, at the command prompt, change the permissions of the passfile file:

    chmod 0600 passfile
  8. On the access node, and on each of the data nodes, reload the server configuration to pick up the changes:

    pg_ctl reload
  9. If you have not already done so, add the data nodes to the access node. For instructions, see the multi-node setup section.

  10. On the access node, at the psql prompt, create additional roles, and grant them access to foreign server objects for the data nodes:

    CREATE ROLE testrole PASSWORD 'clientpass' LOGIN;
    GRANT USAGE ON FOREIGN SERVER <data node name>, <data node name>, ... TO testrole;

    The clientpass password is used by external clients to connect to the access node as user testrole. If the access node is configured to accept other authentication methods, or the role is not a login role, then you might not need to do this step.

  11. On the access node, add the new role to each of the data nodes with distributed_exec. Make sure you add the PASSWORD parameter to specify a different password to use when connecting to the data nodes with role testrole:

    CALL distributed_exec($$ CREATE ROLE testrole PASSWORD 'internalpass' LOGIN $$);
  12. On the access node, add the new role to the passfile you created earlier, by adding this line:

    *:*:*:testrole:internalpass #assuming 'internalpass' is the password used to connect to data nodes
Important

Any user passwords that you created before you set up password authentication need to be re-created so that they use the new encryption method.

This method is a bit more complex to set up than password authentication, but it is more secure, easier to automate, and can be customized to your security environment.

To use certificates, the access node and each data node need three files:

  • The root CA certificate, called root.crt. This certificate serves as the root of trust in the system. It is used to verify the other certificates.
  • A node certificate, called server.crt. This certificate provides the node with a trusted identity in the system.
  • A node certificate key, called server.key. This provides proof of ownership of the node certificate. Make sure you keep this file private on the node where it is generated.

You can purchase certificates from a commercial certificate authority (CA), or generate your own self-signed CA. This section shows you how to use your access node certificate to create and sign new user certificates for the data nodes.

Keys and certificates serve different purposes on the data nodes and access node. For the access node, a signed certificate is used to verify user certificates for access. For the data nodes, a signed certificate authenticates the node to the access node.

  1. On the access node, at the command prompt, generate a private key called auth.key:

    openssl genpkey -algorithm rsa -out auth.key
  2. Generate a self-signed root certificate for the certificate authority (CA), called root.cert:

    openssl req -new -key auth.key -days 3650 -out root.crt -x509
  3. Complete the questions asked by the script to create your root certificate. Type your responses in, press enter to accept the default value shown in brackets, or type . to leave the field blank. For example:

    Country Name (2 letter code) [AU]:US
    State or Province Name (full name) [Some-State]:New York
    Locality Name (eg, city) []:New York
    Organization Name (eg, company) [Internet Widgets Pty Ltd]:Example Company Pty Ltd
    Organizational Unit Name (eg, section) []:
    Common Name (e.g. server FQDN or YOUR name) []:http://cert.example.com/
    Email Address []:

When you have created the root certificate on the access node, you can generate certificates and keys for each of the data nodes. To do this, you need to create a certificate signing request (CSR) for each data node.

The default names for the key is server.key, and for the certificate is server.crt. They are stored in together, in the data directory on the data node instance.

The default name for the CSR is server.csr and you need to sign it using the root certificate you created on the access node.

  1. On the access node, generate a certificate signing request (CSR) called server.csr, and create a new key called server.key:

    openssl req -out server.csr -new -newkey rsa:2048 -nodes \
    -keyout server.key
  2. Sign the CSR using the root certificate CA you created earlier, called auth.key:

    openssl ca -extensions v3_intermediate_ca -days 3650 -notext \
    -md sha256 -in server.csr -out server.crt
  3. Move the server.crt and server.key files from the access node, on to each data node, in the data directory. Depending on your network setup, you might need to use portable media.

  4. Copy the root certificate file root.crt from the access node, on to each data node, in the data directory. Depending on your network setup, you might need to use portable media.

When you have created the certificates and keys, and moved all the files into the right places on the data nodes, you can configure the data nodes to use SSL authentication.

  1. On each data node, open the postgresql.conf configuration file and add or edit the SSL settings to enable certificate authentication:

    ssl = on
    ssl_ca_file = 'root.crt'
    ssl_cert_file = 'server.crt'
    ssl_key_file = 'server.key'
  2. If you want the access node to use certificate authentication for login, make these changes on the access node as well.

  3. On each data node, open the pg_hba.conf configuration file, and add or edit this line to allow any SSL user log in with client certificate authentication:

    # TYPE DATABASE USER ADDRESS METHOD OPTIONS
    hostssl all all all cert clientcert=1
Note

If you are using the default names for your certificate and key, you do not need to explicitly set them. The configuration looks for server.crt and server.key by default. If you use different names for your certificate and key, make sure you specify the correct names in the postgresql.conf configuration file.

When your data nodes are configured to use SSL certificate authentication, you need to create a signed certificate and key for your access node. This allows the access node to log in to the data nodes.

  1. On the access node, as the postgres user, compute a base name for the certificate files using md5sum, generate a subject identifier, and create names for the key and certificate files:

    pguser=postgres
    base=`echo -n $pguser | md5sum | cut -c1-32`
    subj="/C=US/ST=New York/L=New York/O=Timescale/OU=Engineering/CN=$pguser"
    key_file="timescaledb/certs/$base.key"
    crt_file="timescaledb/certs/$base.crt"
  2. Generate a new random user key:

    openssl genpkey -algorithm RSA -out "$key_file"
  3. Generate a certificate signing request (CSR). This file is temporary, stored in the data directory, and is deleted later on:

    openssl req -new -sha256 -key $key_file -out "$base.csr" -subj "$subj"
  4. Sign the CSR with the access node key:

    openssl ca -batch -keyfile server.key -extensions v3_intermediate_ca \
    -days 3650 -notext -md sha256 -in "$base.csr" -out "$crt_file"
    rm $base.csr
  5. Append the node certificate to the user certificate. This completes the certificate verification chain and makes sure that all certificates are available on the data node, up to the trusted certificate stored in root.crt:

    cat >>$crt_file <server.crt
Note

By default, the user key files and certificates are stored on the access node in the data directory, under timescaledb/certs. You can change this location using the timescaledb.ssl_dir configuration variable.

Your data nodes are now set up to accept certificate authentication, the data and access nodes have keys, and the postgres user has a certificate. If you have not already done so, add the data nodes to the access node. For instructions, see the multi-node setup section. The final step is add additional user roles.

  1. On the access node, at the psql prompt, create the new user and grant permissions:

    CREATE ROLE testrole;
    GRANT USAGE ON FOREIGN SERVER <data node name>, <data node name>, ... TO testrole;

    If you need external clients to connect to the access node as testrole, make sure you also add the LOGIN option. You can also enable password authentication by adding the PASSWORD option.

  2. On the access node, use the distributed_exec command to add the role to all the data nodes:

    CALL distributed_exec($$ CREATE ROLE testrole LOGIN $$);

Keywords

Found an issue on this page?

Report an issue!