Build a Python App with CockroachDB and psycopg2


Take our newest Cockroach University course, CockroachDB for Python Developers.

This tutorial shows you how build a simple Python application with CockroachDB and the psycopg2 driver. For the CockroachDB back-end, you'll use a temporary local cluster.

Step 1. Install the psycopg2 driver

To install the Python psycopg2 driver, run the following command:

$ pip install psycopg2-binary

For other ways to install psycopg2, see the official documentation.

Step 2. Start CockroachDB

Choose whether to run a temporary local cluster or a free CockroachDB cluster on CockroachCloud. The instructions below will adjust accordingly.

Create a free cluster

  1. If you haven't already, sign up for a CockroachCloud account.
  2. Log in to your CockroachCloud account.
  3. On the Clusters page, click Create Cluster.
  4. On the Create your cluster page, select the Free Plan.


    This cluster will be free forever.

  5. (Optional) Select a cloud provider (GCP or AWS) in the Additional configuration section.

  6. Click Create your free cluster.

Your cluster will be created in approximately 20-30 seconds.

Set up your cluster connection

Once your cluster is created, the Connection info dialog displays. Use the information provided in the dialog to set up your cluster connection for the SQL user that was created by default:

  1. Click the name of the cc-ca.crt to download the CA certificate to your local machine.
  2. Create a certs directory on your local machine:

    $ mkdir certs
  3. Move the downloaded cc-ca.crt file to the certs directory:

    $ mv <path>/<to>/cc-ca.crt <path>/<to>/certs

    For example:

    $ mv Users/maxroach/Downloads/cc-ca.crt Users/maxroach/certs
  4. Copy the connection string provided, which will be used in the next steps (and to connect to your cluster in the future).


    This connection string contains your password, which will be provided only once. If you forget your password, you can reset it by going to the SQL Users page.

  1. If you haven't already, download the CockroachDB binary.
  2. Run the cockroach demo command:

    $ cockroach demo \

    This starts a temporary, in-memory cluster and opens an interactive SQL shell to the cluster. Any changes to the database will not persist after the cluster is stopped.

  3. Take note of the (sql/tcp) connection string in the SQL shell welcome text:

    # Connection parameters:
    #   (console)
    #   (sql)     postgres://demo:demo11762@?host=%2Fvar%2Ffolders%2Fc8%2Fb_q93vjj0ybfz0fz0z8vy9zc0000gp%2FT%2Fdemo382139081&port=26257
    #   (sql/tcp) postgres://demo:demo11762@

Step 3. Create a database

  1. In the SQL shell, create the bank database that your application will use:

  2. Create a SQL user for your app:

    > CREATE USER <username> WITH PASSWORD <password>;

    Take note of the username and password. You will use it in your application code later.

  3. Give the user the necessary permissions:

    > GRANT ALL ON DATABASE bank TO <username>;
  1. If you haven't already, download the CockroachDB binary.
  2. Start the built-in SQL shell using the connection string you got from the CockroachCloud Console earlier:

    $ cockroach sql \
    --url='postgres://<username>:<password>@<global host>:26257/<cluster_name>.defaultdb?sslmode=verify-full&sslrootcert=<certs_dir>/cc-ca.crt'

    In the connection string copied from the CockroachCloud Console, your username, password and cluster name are pre-populated. Replace the <certs_dir> placeholder with the path to the certs directory that you created earlier.

  3. In the SQL shell, create the bank database that your application will use:


Step 4. Run the Python code

Now that you have a database, you'll run the code shown below to:

  • Create an accounts table and insert some rows.
  • Transfer funds between two accounts inside a transaction.
  • Delete the accounts from the table before exiting so you can re-run the example code.

To handle transaction retry errors, the code uses an application-level retry loop that, in case of error, sleeps before trying the funds transfer again. If it encounters another retry error, it sleeps for a longer interval, implementing exponential backoff.

Get the code

Download the file, or create the file yourself and copy the code into it.

If you prefer, you can also clone a version of the code:

$ git clone
#!/usr/bin/env python3
Test psycopg with CockroachDB.

import time
import random
import logging
from argparse import ArgumentParser, RawTextHelpFormatter

import psycopg2
from psycopg2.errors import SerializationFailure

def create_accounts(conn):
    with conn.cursor() as cur:
            "CREATE TABLE IF NOT EXISTS accounts (id INT PRIMARY KEY, balance INT)"
        cur.execute("UPSERT INTO accounts (id, balance) VALUES (1, 1000), (2, 250)")
        logging.debug("create_accounts(): status message: %s", cur.statusmessage)

def delete_accounts(conn):
    with conn.cursor() as cur:
        cur.execute("DELETE FROM bank.accounts")
        logging.debug("delete_accounts(): status message: %s", cur.statusmessage)

def print_balances(conn):
    with conn.cursor() as cur:
        cur.execute("SELECT id, balance FROM accounts")
        logging.debug("print_balances(): status message: %s", cur.statusmessage)
        rows = cur.fetchall()
        print(f"Balances at {time.asctime()}:")
        for row in rows:

def transfer_funds(conn, frm, to, amount):
    with conn.cursor() as cur:

        # Check the current balance.
        cur.execute("SELECT balance FROM accounts WHERE id = %s", (frm,))
        from_balance = cur.fetchone()[0]
        if from_balance < amount:
            raise RuntimeError(
                f"Insufficient funds in {frm}: have {from_balance}, need {amount}"

        # Perform the transfer.
            "UPDATE accounts SET balance = balance - %s WHERE id = %s", (amount, frm)
            "UPDATE accounts SET balance = balance + %s WHERE id = %s", (amount, to)

    logging.debug("transfer_funds(): status message: %s", cur.statusmessage)

def run_transaction(conn, op, max_retries=3):
    Execute the operation *op(conn)* retrying serialization failure.

    If the database returns an error asking to retry the transaction, retry it
    *max_retries* times before giving up (and propagate it).
    # leaving this block the transaction will commit or rollback
    # (if leaving with an exception)
    with conn:
        for retry in range(1, max_retries + 1):

                # If we reach this point, we were able to commit, so we break
                # from the retry loop.

            except SerializationFailure as e:
                # This is a retry error, so we roll back the current
                # transaction and sleep for a bit before retrying. The
                # sleep time increases for each failed transaction.
                logging.debug("got error: %s", e)
                logging.debug("EXECUTE SERIALIZATION_FAILURE BRANCH")
                sleep_ms = (2 ** retry) * 0.1 * (random.random() + 0.5)
                logging.debug("Sleeping %s seconds", sleep_ms)

            except psycopg2.Error as e:
                logging.debug("got error: %s", e)
                logging.debug("EXECUTE NON-SERIALIZATION_FAILURE BRANCH")
                raise e

        raise ValueError(f"Transaction did not succeed after {max_retries} retries")

def test_retry_loop(conn):
    Cause a seralization error in the connection.

    This function can be used to test retry logic.
    with conn.cursor() as cur:
        # The first statement in a transaction can be retried transparently on
        # the server, so we need to add a dummy statement so that our
        # force_retry() statement isn't the first one.
        cur.execute("SELECT now()")
        cur.execute("SELECT crdb_internal.force_retry('1s'::INTERVAL)")
    logging.debug("test_retry_loop(): status message: %s", cur.statusmessage)

def main():
    opt = parse_cmdline()
    logging.basicConfig(level=logging.DEBUG if opt.verbose else logging.INFO)

    conn = psycopg2.connect(opt.dsn)

    amount = 100
    fromId = 1
    toId = 2

        run_transaction(conn, lambda conn: transfer_funds(conn, fromId, toId, amount))

        # The function below is used to test the transaction retry logic.  It
        # can be deleted from production code.
        # run_transaction(conn, test_retry_loop)
    except ValueError as ve:
        # Below, we print the error and continue on so this example is easy to
        # run (and run, and run...).  In real code you should handle this error
        # and any others thrown by the database interaction.
        logging.debug("run_transaction(conn, op) failed: %s", ve)



    # Close communication with the database.

def parse_cmdline():
    parser = ArgumentParser(description=__doc__,
database connection string

For cockroach demo, use
with the username and password created in the demo cluster, and the hostname
and port listed in the (sql/tcp) connection parameters of the demo cluster
welcome message.

For CockroachCloud Free, use

If you are using the connection string copied from the Console, your username,
password, and cluster name will be pre-populated. Replace
<your_certs_directory> with the path to the 'cc-ca.crt' downloaded from the


    parser.add_argument("-v", "--verbose",
                        action="store_true", help="print debug info")

    opt = parser.parse_args()
    return opt

if __name__ == "__main__":

Run the code

The Python code is a command-line utility that accepts the connection string to CockroachDB as a command-line argument:

$ python3 \

Before running the command, update the connection string as follows:

  • Replace <username> and <password> with the SQL username and password that you created earlier.
  • Replace <hostname> and <port> with the hostname and port in the (sql/tcp) connection string from SQL shell welcome text.
$ python3 \

Before running the command, update the connection string that you copied earlier from the Connection info dialog as follows:

  • Replace <username> and <password> with the SQL username and password that you created earlier.
  • Replace <globalhost> with the name of the CockroachCloud Free (beta) host (e.g.,
  • Replace <cluster-name> with the name of your cluster.
  • Replace <certs_directory> with the path to the cc-ca.crt file that you downloaded from the CockroachCloud Console.

If you are using the connection string that you copied from the Connection info dialog, your username, password, and cluster name will be pre-populated.

The output should show the account balances before and after the funds transfer:

Balances at Fri Oct 30 18:27:00 2020:
(1, 1000)
(2, 250)
Balances at Fri Oct 30 18:27:00 2020:
(1, 900)
(2, 350)

What's next?

Read more about using the Python psycopg2 driver.

You might also be interested in the following pages:

YesYes NoNo