Enable TLS in Vertica on Kubernetes

To protect database communications and enforce certificate verification, the VerticaDB operator supports Transport Layer Security (TLS) for the following:

  • HTTPS/NMA service: Secures the REST interface used by vclusterops and the Node Management Agent (NMA).
  • Client-Server authentication: Secures connections between external client applications (vsql, JDBC, ODBC, ADO.NET) and Vertica server nodes.
  • Internode traffic: Encrypts and validates communications between database cluster nodes.

The Vertica database server does not store private keys or certificates in its catalog or on local disk. Instead, it dynamically reads TLS credentials from Kubernetes secrets or AWS Secrets Manager.

TLS configuration modes and parameters

The following table describes the TLS configuration parameters and their supported modes:

Parameter Description Supported modes Default mode
httpsNMATLS TLS for the HTTPS service and Node Management Agent (NMA). DISABLE, ENABLE, TRY_VERIFY, VERIFY_CA, VERIFY_FULL TRY_VERIFY
clientServerTLS TLS for SQL client connections (vsql, drivers). DISABLE, ENABLE, TRY_VERIFY, VERIFY_CA, VERIFY_FULL TRY_VERIFY
interNodeTLS TLS for internal cluster communication (version 26.1.0-0 and higher). DISABLE, ENABLE, VERIFY_CA VERIFY_CA

You can enable TLS when you:

Enable TLS when creating the database

When you initialize a new database (initPolicy: Create), you can configure TLS parameters directly in the initial VerticaDB custom resource (CR) manifest. The operator provisions the cluster with TLS configured and enabled from the start.

  1. (Optional) If you are using custom certificates instead of operator-generated certificates, you can create your own in AWS Secrets Manager or create the Kubernetes secrets in the same namespace where the VerticaDB CR is deployed:

    # Create custom secret for HTTPS/NMA
    $ kubectl create secret generic custom-https-tls \
      --from-file=tls.key=/path/to/https.key \
      --from-file=tls.crt=/path/to/https.crt \
      --from-file=ca.crt=/path/to/ca.crt
    
    # Create custom secret for Client-Server TLS
    $ kubectl create secret generic custom-client-tls \
      --from-file=tls.key=/path/to/client.key \
      --from-file=tls.crt=/path/to/client.crt \
      --from-file=ca.crt=/path/to/ca.crt
    
    # Create custom secret for Internode TLS
    $ kubectl create secret generic custom-internode-tls \
      --from-file=tls.key=/path/to/internode.key \
      --from-file=tls.crt=/path/to/internode.crt \
      --from-file=ca.crt=/path/to/ca.crt
    
  2. Define the initial VerticaDB CR manifest. Include the TLS annotations and specifications in metadata.annotations and spec:

    apiVersion: vertica.com/v1
    kind: VerticaDB
    metadata:
      name: vertica-db
      annotations:
        vertica.com/tls-cache-duration: "86400"
        vertica.com/remove-tls-secret-on-vdb-delete: "false"
    spec:
      initPolicy: Create
      dbName: vertdb
      licenseSecret: vertica-license
      passwordSecret: su-passwd
      communal:
        path: "s3://vertica-communal-storage/vertdb"
        endpoint: "https://s3.amazonaws.com"
        credentialSecret: s3-creds
        region: us-east-1
      subclusters:
        - name: primary
          size: 3
          serviceType: ClusterIP
      shardCount: 6
      httpsNMATLS:
        enabled: true
        mode: TRY_VERIFY
        secret: custom-https-tls       # Omit to allow the operator to auto-generate
      clientServerTLS:
        enabled: true
        mode: TRY_VERIFY
        secret: custom-client-tls      # Omit to allow the operator to auto-generate
      interNodeTLS:
        enabled: true
        mode: VERIFY_CA
        secret: custom-internode-tls   # Omit to allow the operator to auto-generate
      dbTlsConfig:
        tlsVersion: 2                  # 2 for TLS 1.2, 3 for TLS 1.3
        cipherSuites: ""
    
  3. Apply the manifest:

    $ kubectl apply -f verticadb-initial.yaml
    
  4. Monitor the operator reconciliation events:

    $ kubectl describe vdb vertica-db
    

    Expected events include:

    Normal  TLSConfigurationStarted    3m  verticadb-operator  Starting to configure TLS for httpsNMA
    Normal  TLSConfigurationSucceeded  3m  verticadb-operator  Successfully set HTTP tls config
    Normal  TLSConfigurationStarted    3m  verticadb-operator  Starting to configure TLS for clientServer
    Normal  TLSConfigurationSucceeded  2m  verticadb-operator  Successfully set Server tls config
    Normal  TLSConfigurationStarted    2m  verticadb-operator  Starting to configure TLS for interNode
    Normal  TLSConfigurationSucceeded  2m  verticadb-operator  Successfully set Internode tls config
    
  5. Verify the generated Kubernetes secrets (if auto-generated):

    $ kubectl get secrets -l app.kubernetes.io/name=vertica
    
  6. Verify the TLS authentication records created in the database:

    => SELECT auth_name, auth_host_type, auth_method, is_fallthrough_enabled FROM client_auth WHERE auth_name LIKE '%k8s%';
                       auth_name                   | auth_host_type | auth_method | is_fallthrough_enabled
    ----------------------------------------------+----------------+-------------+------------------------
     k8s_remote_ipv4_tls_builtin_auth             | HOSTSSL        | TLS         | False
     k8s_remote_ipv6_tls_builtin_auth             | HOSTSSL        | TLS         | False
     k8s_local_tls_builtin_auth                   | LOCALSSL       | TLS         | False
     k8s_remote_ipv4_reject_non_tls_builtin_auth  | HOSTNOSSL      | REJECT      | False
     k8s_remote_ipv6_reject_non_tls_builtin_auth  | HOSTNOSSL      | REJECT      | False
    (5 rows)
    

Enable TLS when reviving the database

When you revive an existing database (initPolicy: Revive) from data files on communal storage, the database catalog already contains references to the TLS configuration established when the database was running.

  1. Ensure the original TLS secrets exist in the cluster, based on how the original database stored its certificates:

    • If the original VerticaDB CR used auto-generated secrets and vertica.com/remove-tls-secret-on-vdb-delete: "false" (the default), the auto-generated secrets are preserved and can be referenced directly when reviving in the same Kubernetes namespace. If you are reviving to a different namespace, copy the secrets to the new namespace.

    • If the original cluster used custom certificates in Kubernetes secrets, reference the secrets if reviving in the same namespace, or recreate the original TLS secrets from backup files if reviving on a new cluster or namespace:

      # Recreate the original HTTPS TLS secret
      $ kubectl create secret generic vertica-db-https-tls-original \
        --from-file=tls.key=./backup/https.key \
        --from-file=tls.crt=./backup/https.crt \
        --from-file=ca.crt=./backup/ca.crt
      
      # Recreate the original Client-Server TLS secret
      $ kubectl create secret generic vertica-db-clientserver-tls-original \
        --from-file=tls.key=./backup/clientserver.key \
        --from-file=tls.crt=./backup/clientserver.crt \
        --from-file=ca.crt=./backup/ca.crt
      
      # Recreate the original Internode TLS secret
      $ kubectl create secret generic vertica-db-internode-tls-original \
        --from-file=tls.key=./backup/internode.key \
        --from-file=tls.crt=./backup/internode.crt \
        --from-file=ca.crt=./backup/ca.crt
      
    • If the original secrets were stored in AWS Secrets Manager, no extra steps are required; reuse them directly.

  2. Define the revive VerticaDB CR manifest with initPolicy: Revive, and reference the communal storage path and the original secrets:

    apiVersion: vertica.com/v1
    kind: VerticaDB
    metadata:
      name: vertica-db
      annotations:
        vertica.com/tls-cache-duration: "86400"
        vertica.com/remove-tls-secret-on-vdb-delete: "false"
        vertica.com/ignore-cluster-lease: "true"
    spec:
      initPolicy: Revive
      dbName: vertdb                   # Must match the name of the database being revived
      licenseSecret: vertica-license
      passwordSecret: su-passwd
      communal:
        path: "s3://vertica-communal-storage/vertdb"  # Existing non-empty communal path
        endpoint: "https://s3.amazonaws.com"
        credentialSecret: s3-creds
        region: us-east-1
      subclusters:
        - name: primary
          size: 3
      httpsNMATLS:
        enabled: true
        mode: TRY_VERIFY
        secret: vertica-db-https-tls-original
      clientServerTLS:
        enabled: true
        mode: TRY_VERIFY
        secret: vertica-db-clientserver-tls-original
      interNodeTLS:
        enabled: true
        mode: VERIFY_CA
        secret: vertica-db-internode-tls-original
    
  3. Apply the manifest:

    $ kubectl apply -f verticadb-revive.yaml
    
  4. Monitor the operator reconciliation events:

    $ kubectl describe vdb vertica-db
    

    Expected events include:

    Normal  TLSConfigurationStarted    3m  verticadb-operator  Starting to configure TLS for httpsNMA
    Normal  TLSConfigurationSucceeded  3m  verticadb-operator  Successfully set HTTP tls config
    Normal  TLSConfigurationStarted    3m  verticadb-operator  Starting to configure TLS for clientServer
    Normal  TLSConfigurationSucceeded  2m  verticadb-operator  Successfully set Server tls config
    Normal  TLSConfigurationStarted    2m  verticadb-operator  Starting to configure TLS for interNode
    Normal  TLSConfigurationSucceeded  2m  verticadb-operator  Successfully set Internode tls config
    
  5. Verify the generated Kubernetes secrets (if auto-generated):

    $ kubectl get secrets -l app.kubernetes.io/name=vertica
    

    Ensure the revive operation completes and the database pod status transitions to Ready.

Enable TLS after the database is created or revived

If a database was initialized or revived without TLS enabled (or TLS was disabled), you can enable TLS in place by updating the existing VerticaDB custom resource. The operator configures TLS and updates the NMA and HTTPS services without requiring a full cluster restart.

  1. Edit the deployed custom resource:

    $ kubectl edit vdb vertica-db
    

    Modify or add the TLS annotations and specification blocks, then save and exit the editor:

    apiVersion: vertica.com/v1
    kind: VerticaDB
    metadata:
      name: vertica-db
    ...
    spec:
      # ... existing database spec ...
    
      # Add TLS configurations (provide custom secrets or omit secret for auto-generation)
      clientServerTLS:
        enabled: true
        mode: TRY_VERIFY
        secret: custom-client-tls
      httpsNMATLS:
        enabled: true
        mode: TRY_VERIFY
        secret: custom-https-tls
      interNodeTLS:
        enabled: true
        mode: VERIFY_CA
        secret: custom-internode-tls
    
  2. Check the operator progress:

    $ kubectl describe vdb vertica-db
    

    You will observe events confirming TLS configuration across components:

    Normal  TLSConfigurationStarted    3m  verticadb-operator  Starting to configure TLS for httpsNMA
    Normal  TLSConfigurationSucceeded  3m  verticadb-operator  Successfully set HTTP tls config
    Normal  TLSConfigurationStarted    3m  verticadb-operator  Starting to configure TLS for clientServer
    Normal  TLSConfigurationSucceeded  2m  verticadb-operator  Successfully set Server tls config
    Normal  TLSConfigurationStarted    2m  verticadb-operator  Starting to configure TLS for interNode
    Normal  TLSConfigurationSucceeded  2m  verticadb-operator  Successfully set Internode tls config
    

Upgrade scenarios and considerations for TLS

When you upgrade your environment, special considerations apply depending on whether you are upgrading the VerticaDB operator or the database server.

VerticaDB operator upgrade

Upgrading from operator v2.2.0 or earlier

In VerticaDB operator version 2.2.0, the selector labels for StatefulSets were modified. Kubernetes selector labels are immutable, so the operator reconciler automatically deletes existing StatefulSets and recreates them with updated labels, which triggers a rolling restart of all database pods.

After the rolling restart is complete, you can configure TLS as described in Enable TLS after the database is created or revived.

Upgrading from operator versions prior to v25.3.0 (pre-operator TLS support)

Operator-managed TLS (through CR parameters and Kubernetes secrets) was introduced in version 25.3.0-0. If you are upgrading from an operator version earlier than v25.3.0, consider the following scenarios:

Scenario 1: TLS was previously enabled manually in Vertica

If TLS was configured manually inside the database (for example, through ALTER TLS CONFIGURATION or custom SQL commands), existing TLS configurations remain active and untouched after the operator upgrade.

To manage TLS through the VerticaDB operator going forward, export your existing database certificate, private key, and CA certificate into Kubernetes secrets, and then edit the VerticaDB CR to reference them:

# 1. Create a Kubernetes secret with existing certificates
$ kubectl create secret generic existing-db-tls \
  --from-file=tls.key=/path/to/existing.key \
  --from-file=tls.crt=/path/to/existing.crt \
  --from-file=ca.crt=/path/to/existing_ca.crt
# 2. Update the existing VerticaDB CR
...
spec:
  clientServerTLS:
    enabled: true
    mode: TRY_VERIFY
    secret: existing-db-tls
  httpsNMATLS:
    enabled: true
    mode: TRY_VERIFY
    secret: existing-db-tls

Scenario 2: TLS was not enabled

If TLS was not previously enabled, upgrade the operator to the latest version and configure TLS as described in Enable TLS after the database is created or revived.

Automated TLS spec migration in operator v25.4.0 and higher

Beginning in operator version 25.4.0, the operator automatically backfills and sets the explicit enabled boolean flags in spec.httpsNMATLS.enabled and spec.clientServerTLS.enabled during reconciliation.

If you are upgrading from an operator version earlier than 25.4.0 where the enabled fields are nil in the CR spec, the operator checks the vertica.com/enable-tls-auth annotation and automatically updates the CR spec to match:

  • If vertica.com/enable-tls-auth: "true" is set, the operator sets spec.httpsNMATLS.enabled: true and spec.clientServerTLS.enabled: true.
  • If vertica.com/enable-tls-auth: "false" (or not set), the operator sets both fields to false.

Vertica database server upgrade

When you upgrade the database server image (for example, updating spec.image in the VerticaDB CR), consider the following:

Configuration retention

All database configurations, catalog objects, and TLS settings configured in previous versions are preserved as is.

Automatic TLS fallthrough migration (upgrading from v26.2.0-1 or earlier to v26.2.0-2 or higher)

Database server version 26.2.0-2 introduced stricter built-in client authentication rules (HOSTNOSSL rejections) along with the vertica.com/enable-fallthrough annotation.

Post-upgrade verification

After you complete the database upgrade, verify that the client authentication rules remain the same as they were before the upgrade:

=> SELECT auth_name, auth_host_type, auth_method, is_fallthrough_enabled
   FROM client_auth
   WHERE auth_name LIKE '%k8s%';
             auth_name             | auth_host_type | auth_method | is_fallthrough_enabled
----------------------------------+----------------+-------------+------------------------
 k8s_remote_ipv4_tls_builtin_auth | HOSTSSL        | TLS         | True
 k8s_remote_ipv6_tls_builtin_auth | HOSTSSL        | TLS         | True
 k8s_local_tls_builtin_auth       | LOCALSSL       | TLS         | True
(3 rows)

If you want to enforce strict TLS authentication (disallowing fallthrough to password authentication), edit the CR and set vertica.com/enable-fallthrough: "false".

Verify and validate

After TLS is configured using any of the use cases, verify the TLS status inside the database.

  1. Connect to the database using SSL/TLS through vsql. The initial banner displays the negotiated SSL/TLS cipher and protocol:

    $ vsql -h <VERTICA_SERVICE_IP_OR_HOST> -U dbadmin -w <PASSWORD>
    

    Example output:

    Welcome to vsql, the Vertica Analytic Database interactive terminal.
    
    SSL connection (cipher: ECDHE-RSA-AES256-GCM-SHA384, bits: 256, protocol: TLSv1.2)
    
  2. Verify the database TLS configurations by querying the tls_configurations system table:

    => SELECT name, owner, certificate, ca_certificate, cipher_suites, mode
       FROM tls_configurations;
         name     |  owner  |     certificate     |     ca_certificate     | cipher_suites |    mode
    --------------+---------+---------------------+------------------------+---------------+------------
     https        | dbadmin | https_cert_0        | https_ca_cert_0        |               | TRY_VERIFY
     server       | dbadmin | server_cert_0       | server_ca_cert_0       |               | TRY_VERIFY
     data_channel | dbadmin | data_channel_cert_0 | data_channel_ca_cert_0 |               | VERIFY_CA
     LDAPLink     | dbadmin |                     |                        |               | DISABLE
     LDAPAuth     | dbadmin |                     |                        |               | DISABLE
    (5 rows)
    
  3. Verify the built-in client authentication rules. The operator automatically configures these TLS client authentication records:

    => SELECT auth_name, auth_host_type, auth_method, is_fallthrough_enabled FROM client_auth WHERE auth_name LIKE '%k8s%';
                       auth_name                   | auth_host_type | auth_method | is_fallthrough_enabled
    ----------------------------------------------+----------------+-------------+------------------------
     k8s_remote_ipv4_tls_builtin_auth             | HOSTSSL        | TLS         | False
     k8s_remote_ipv6_tls_builtin_auth             | HOSTSSL        | TLS         | False
     k8s_local_tls_builtin_auth                   | LOCALSSL       | TLS         | False
     k8s_remote_ipv4_reject_non_tls_builtin_auth  | HOSTNOSSL      | REJECT      | False
     k8s_remote_ipv6_reject_non_tls_builtin_auth  | HOSTNOSSL      | REJECT      | False
    (5 rows)
    
  4. Verify the active secret mapping and cache duration to confirm that the database references the expected Kubernetes secret and cache configuration:

    => SELECT name, secret_name, secret_manager_config
       FROM certificates;
    

    Example output:

              name           |            secret_name            |                        secret_manager_config
    ------------------------+-----------------------------------+---------------------------------------------------------------------
     https_ca_cert_0        | vertica-db-https-tls-lhj82        | {"data-key":"ca.crt","namespace":"vertica","cache-duration":86400}
     server_ca_cert_0       | vertica-db-clientserver-tls-j86v9 | {"data-key":"ca.crt","namespace":"vertica","cache-duration":86400}
     data_channel_ca_cert_0 | vertica-db-internode-tls-gbcx8    | {"data-key":"ca.crt","namespace":"vertica","cache-duration":86400}
     https_cert_0           | vertica-db-https-tls-lhj82        | {"data-key":"tls.crt","namespace":"vertica","cache-duration":86400}
     server_cert_0          | vertica-db-clientserver-tls-j86v9 | {"data-key":"tls.crt","namespace":"vertica","cache-duration":86400}
     data_channel_cert_0    | vertica-db-internode-tls-gbcx8    | {"data-key":"tls.crt","namespace":"vertica","cache-duration":86400}
    (6 rows)