Set Up Physical Cluster Replication

• The system virtual cluster for administration tasks in both clusters, and starting the replication stream from the standby cluster.
• The application virtual cluster on the primary cluster to work with databases, tables, workloads, and so on.

​

Overview

1. Create and start the primary cluster.
2. Configure and create a user on the primary cluster.
3. Create and start the standby cluster.
4. Configure and create a user on the standby cluster.
5. Securely copy certificates.
6. Start the replication stream from the standby cluster.

​

Before you begin

• You need two separate CockroachDB clusters (primary and standby), each with a minimum of three nodes. The standby cluster should be the same version or one version ahead of the primary cluster. The primary and standby clusters must be configured with similar hardware profiles, number of nodes, and overall size. Significant discrepancies in the cluster configurations may result in degraded performance.
  • To set up each cluster, you can follow . When you initialize the cluster with the command, you must pass the --virtualized or --virtualized-empty flag. Refer to the cluster creation steps for the primary cluster and for the standby cluster for details.
  • The tutorial creates a self-signed certificate for each self-hosted cluster. To create certificates signed by an external certificate authority, refer to .
• All nodes in each cluster will need access to the Certificate Authority for the other cluster. Refer to Manage cluster certificates.
• The primary and standby clusters can have different . However, behavior for features that rely on multi-region primitives, such as Region by Row and Region by Table, may be affected. For more information, refer to .

​

Step 1. Create the primary cluster

​

Initialize the primary cluster

cockroach init --certs-dir=certs --host={address of any node on --join list} --virtualized

​

Connect to the primary cluster system virtual cluster

1. To connect to the system virtual cluster, pass the options=-ccluster=system parameter in the URL:
   cockroach sql --url \
   "postgresql://root@{node IP or hostname}:26257?options=-ccluster=system&sslmode=verify-full" \
   --certs-dir "certs"
   
   The prompt will include system when you are connected to the system virtual cluster.

1. Set the kv.rangefeed.enabled cluster setting to true. The replication job connects to a long-lived request, a rangefeed, which pushes changes as they happen:
   SET CLUSTER SETTING kv.rangefeed.enabled = true;
   
2. Confirm the status of your virtual cluster:
   SHOW VIRTUAL CLUSTERS;
   
   The output will include the system virtual cluster and the main virtual cluster:
     id |  name  | data_state | service_mode
   -----+--------+------------+---------------
      1 | system | ready      | shared
      3 | main   | ready      | shared
   (2 rows)
   
   Because this is the primary cluster rather than the standby cluster, the data_state of all rows is ready, rather than replicating or another .

​

Create a user with replication privileges

1. From the primary’s system virtual cluster SQL shell, create a user and password:
   CREATE USER {your username} WITH PASSWORD '{your password}';
   
   If you need to change the password later, refer to .
2. Grant the to your user:
   GRANT SYSTEM REPLICATIONSOURCE TO {your username};
   

​

Connect to the primary virtual cluster (optional)

1. If you would like to run a sample workload on the primary’s virtual cluster, open a new terminal window and use to run the workload. For example, to initiate the workload:
   cockroach workload init movr "postgresql://root@{node_advertise_address}:{node_advertise_port}?options=-ccluster=main&sslmode=verify-full&sslrootcert=certs/ca.crt&sslcert=certs/client.root.crt&sslkey=certs/client.root.key"
   
   Replace and with a node’s IP address or hostname and port. The cockroach workload command does not support connection or security flags like other . Instead, you must use a at the end of the command. As a result, for the example in this tutorial, you will need:
   • options=-ccluster=main
   • sslmode=verify-full
   • sslrootcert={path}/certs/ca.crt: the path to the CA certificate.
   • sslcert={path}/certs/client.root.crt: the path to the client certificate.
   • sslkey={path}/certs/client.root.key: the path to the client private key.
   For additional detail on the standard CockroachDB connection parameters, refer to .
2. Run the movr workload for a set duration using the same connection string:
   cockroach workload run movr --duration=5m "postgresql://root@{node_advertise_address}:{node_advertise_port}?options=-ccluster=main&sslmode=verify-full&sslrootcert=certs/ca.crt&sslcert=certs/client.root.crt&sslkey=certs/client.root.key"
   
3. To connect to the primary cluster’s virtual cluster, use the options=-ccluster={virtual_cluster_name} parameter:
   cockroach sql --url \
   "postgresql://root@{node IP or hostname}:26257?options=-ccluster=main&sslmode=verify-full" \
   --certs-dir "certs"
   
   The prompt will include main when you are connected to the virtual cluster.
4. Create a user for your primary cluster’s main virtual cluster:
   CREATE USER {your username} WITH PASSWORD '{your password}';
   
5. You can connect to the with this user to observe activity on the primary cluster. Open a web browser at https://{node IP or hostname}:8080/ and enter your credentials.

​

Step 2. Create the standby cluster

​

Initialize the standby cluster

cockroach init --certs-dir=certs --host={address of any node on --join list} --virtualized-empty

​

Connect to the standby cluster system virtual cluster

1. To connect to the system virtual cluster, pass the options=-ccluster=system parameter in the URL:
   cockroach sql --url \
   "postgresql://root@{node IP or hostname}:{26257}?options=-ccluster=system&sslmode=verify-full" \
   --certs-dir "certs"
   
2. Set the kv.rangefeed.enabled cluster setting to true. The replication job connects to a long-lived request, a rangefeed, which pushes changes as they happen:
   SET CLUSTER SETTING kv.rangefeed.enabled = true;
   
3. Confirm the status of your virtual cluster:
   SHOW VIRTUAL CLUSTERS;
   
   The output will show the system virtual cluster, but no main virtual cluster:
   id |   name   | data_state | service_mode
   ---+----------+------------+---------------
   1  | system   | ready      | shared
   (1 rows)
   

​

Create a user with replication privileges on the standby cluster

1. Create a user:
   CREATE USER {your username} WITH LOGIN PASSWORD {'your password'};
   
2. To observe the replication activity, your user will need privileges:
   GRANT SYSTEM REPLICATIONDEST, MANAGEVIRTUALCLUSTER TO {your username};
   
   Open the DB Console in your web browser: https://{node IP or hostname}:8080/, where you will be prompted for these credentials. Refer to for more detail on tracking relevant metrics for your replication stream.

​

Step 3. Manage cluster certificates and generate connection strings

cockroach convert-url --url "postgresql://{replication user}:{password}@{node IP or hostname}:26257" --format crdb --ca-cert {path to certs directory}/certs/ca.crt --inline

postgresql://{replication user}:{password}@{node IP or hostname}:26257/defaultdb?options=-ccluster%3Dsystem&sslinline=true&sslmode=verify-full&sslrootcert=-----BEGIN+CERTIFICATE-----{encoded_cert}-----END+CERTIFICATE-----%0A

​

Step 4. Start replication

1. In the SQL shell on the standby cluster, create an external connection using the primary cluster’s connection string. Prefix the postgresql:// scheme to the connection string and replace with your external connection name:
   CREATE EXTERNAL CONNECTION {source} AS 'postgresql://{replication user}:{password}@{node IP or hostname}:{26257}/defaultdb?options=-ccluster%3Dsystem&sslinline=true&sslmode=verify-full&sslrootcert=-----BEGIN+CERTIFICATE-----{encoded_cert}-----END+CERTIFICATE-----%0A';
   
   If the primary and standby cluster nodes are on different networks, you can route the replication stream through the primary cluster’s load balancer. Add &crdb_route=gateway to the connection string:
   CREATE EXTERNAL CONNECTION {source} AS 'postgresql://{replication user}:{password}@{node IP or hostname}:{26257}/defaultdb?options=-ccluster%3Dsystem&crdb_route=gateway&sslinline=true&sslmode=verify-full&sslrootcert=-----BEGIN+CERTIFICATE-----{encoded_cert}-----END+CERTIFICATE-----%0A';
   
   For an optimally performant replication stream, all nodes on the primary and standby clusters should share the same virtual network. The gateway route option should only be used when this network configuration is not possible due to firewall or IP allocation constraints. Once the external connection has been created, create the replication stream from the standby cluster:
   CREATE VIRTUAL CLUSTER main
   FROM REPLICATION OF main
   ON '{source}';
   
   Once the standby cluster has made a connection to the primary cluster, the standby pulls the topology of the primary cluster and distributes the replication work across all nodes in the primary and standby.

1. To view all virtual clusters on the standby, run:
   SHOW VIRTUAL CLUSTERS;
   
   The standby cluster will show the main virtual cluster is in a replicating state.
     id |  name  | data_state  | service_mode
   -----+--------+-------------+---------------
      1 | system | ready       | shared
      3 | main   | replicating | none
   (2 rows)
   
   The standby cluster’s virtual cluster is offline while the replication stream is running. To bring it online, you must explicitly .
2. To manage the replication stream, you can the replication stream as well as the current details for the job:
   ALTER VIRTUAL CLUSTER main PAUSE REPLICATION;
   
   ALTER VIRTUAL CLUSTER main RESUME REPLICATION;
   
   SHOW VIRTUAL CLUSTER main WITH REPLICATION STATUS;
   
   id | name | source_tenant_name |              source_cluster_uri                        |         retained_time         |    replicated_time     | replication_lag | failover_time |   status
   ---+------+--------------------+--------------------------------------------------------+-------------------------------+------------------------+-----------------+--------------+--------------
   3  | main | main               | postgresql://user@{node IP or hostname}:{26257}?redacted | 2024-04-17 20:14:31.952783+00 | 2024-04-17 20:18:50+00 | 00:00:08.738176 |         NULL | replicating
   (1 row)
   
   With the replication stream running, you can monitor the job via the DB Console, SQL shell, or Prometheus. You can also verify data is correct on the standby cluster at a specific point in time. For more detail, refer to .

​

Set up PCR from an existing cluster

• An existing primary cluster. If you need to upgrade your existing cluster, refer to .
• A standby cluster that is at the same version or one version ahead of the primary cluster. To set up the cluster, you can follow . When you initialize the cluster with the command, you must pass the --virtualized-empty flag. For details, refer to the cluster creation steps for the standby cluster.
• Review the remaining prerequisites at the start of this page, which also apply to running PCR from an existing cluster.

​

Example

1. To configure the standby cluster ready for PCR, follow the steps outlined in Connect to the standby cluster system virtual cluster that include enabling the necessary cluster settings.
2. Create a user on both clusters and grant the SYSTEM REPLICATION privilege to the created user on each cluster:
   CREATE USER {your username} WITH PASSWORD '{your password}';
   
   GRANT SYSTEM REPLICATION TO {your username};
   
3. View the virtual clusters on both clusters:
   SHOW VIRTUAL CLUSTERS;
   
   You will find that both clusters only have the system virtual cluster:
     id |  name  | data_state | service_mode
   -----+--------+------------+---------------
   1    | system | ready      | shared
   (1 row)
   
4. To create the replication job, you will need a connection string for the primary cluster containing its CA certificate. For steps to generate a connection string with cockroach convert-url, refer to Step 3. Manage cluster certificates and generate connection strings.
5. If you would like to run a test workload on your existing primary cluster, you can use like the following:
   cockroach workload init movr 'postgresql://root@{node IP or hostname}:{26257}?options=-ccluster=main&sslmode=verify-full&sslrootcert=certs/ca.crt&sslcert=certs/client.root.crt&sslkey=certs/client.root.key'
   
6. To start the replication, you will need to replicate system from the existing primary cluster in order to create a new virtual cluster on the standby cluster (main in this example). On the standby cluster, run:
   CREATE VIRTUAL CLUSTER main FROM REPLICATION OF system ON 'postgresql://{replication user}:{password}@{node IP or hostname}:26257/defaultdb?options=-ccluster%3Dsystem&sslinline=true&sslmode=verify-full&sslrootcert=-----BEGIN+CERTIFICATE-----{encoded_cert}-----END+CERTIFICATE-----%0A';
   
   This statement includes:
   • main: The name of the virtual cluster to create on the standby cluster.
   • system: The system virtual cluster on the existing primary cluster.
   • The connection string for the existing primary cluster.
7. View the virtual clusters on the standby cluster:
   SHOW VIRTUAL CLUSTERS;
   
   The output shows the newly created main virtual cluster and that the replication job is :
     id |  name  |        data_state        | service_mode
   -----+--------+--------------------------+---------------
   1    | system | ready                    | shared
   3    | main   | initializing replication | none
   (2 rows)
   

​

Connection reference

​

What’s next