This tutorial shows you how build a simple Python application with CockroachDB using a PostgreSQL-compatible driver or ORM.
We have tested the Python psycopg2 driver and the SQLAlchemy ORM enough to claim beta-level support, so those are featured here. If you encounter problems, please open an issue with details to help us make progress toward full support.
Before You Begin
Make sure you have already installed CockroachDB.
Step 1. Install the psycopg2 driver
To install the Python psycopg2 driver, run the following command:
$ pip install psycopg2
For other ways to install psycopg2, see the official documentation.
Step 2. Start a single-node cluster
For the purpose of this tutorial, you need only one CockroachDB node running in insecure mode:
$ cockroach start \
--insecure \
--store=hello-1 \
--host=localhost
Step 3. Create a user
In a new terminal, as the root user, use the cockroach user command to create a new user, maxroach.
$ cockroach user set maxroach --insecure
Step 4. Create a database and grant privileges
As the root user, use the built-in SQL client to create a bank database.
$ cockroach sql --insecure -e 'CREATE DATABASE bank'
Then grant privileges to the maxroach user.
$ cockroach sql --insecure -e 'GRANT ALL ON DATABASE bank TO maxroach'
Step 5. Run the Python code
Now that you have a database and a user, you'll run code to create a table and insert some rows, and then you'll run code to read and update values as an atomic transaction.
Basic Statements
First, use the following code to connect as the maxroach user and execute some basic SQL statements, creating a table, inserting rows, and reading and printing the rows.
Download the basic-sample.py file, or create the file yourself and copy the code into it.
# Import the driver.
import psycopg2
# Connect to the "bank" database.
conn = psycopg2.connect(database='bank', user='maxroach', host='localhost', port=26257)
# Make each statement commit immediately.
conn.set_session(autocommit=True)
# Open a cursor to perform database operations.
cur = conn.cursor()
# Create the "accounts" table.
cur.execute("CREATE TABLE IF NOT EXISTS accounts (id INT PRIMARY KEY, balance INT)")
# Insert two rows into the "accounts" table.
cur.execute("INSERT INTO accounts (id, balance) VALUES (1, 1000), (2, 250)")
# Print out the balances.
cur.execute("SELECT id, balance FROM accounts")
rows = cur.fetchall()
print('Initial balances:')
for row in rows:
print([str(cell) for cell in row])
# Close the database connection.
cur.close()
conn.close()
Then run the code:
$ python basic-sample.py
The output should be:
Initial balances:
['1', '1000']
['2', '250']
Transaction (with retry logic)
Next, use the following code to again connect as the maxroach user but this time execute a batch of statements as an atomic transaction to transfer funds from one account to another, where all included statements are either committed or aborted.
Download the txn-sample.py file, or create the file yourself and copy the code into it.
SERIALIZABLE isolation level, CockroachDB may require the client to retry a transaction in case of read/write contention. CockroachDB provides a generic retry function that runs inside a transaction and retries it as needed. You can copy and paste the retry function from here into your code.# Import the driver.
import psycopg2
import psycopg2.errorcodes
# Connect to the cluster.
conn = psycopg2.connect(database='bank', user='maxroach', host='localhost', port=26257)
def onestmt(conn, sql):
with conn.cursor() as cur:
cur.execute(sql)
# Wrapper for a transaction.
# This automatically re-calls "op" with the open transaction as an argument
# as long as the database server asks for the transaction to be retried.
def run_transaction(conn, op):
with conn:
onestmt(conn, "SAVEPOINT cockroach_restart")
while True:
try:
# Attempt the work.
op(conn)
# If we reach this point, commit.
onestmt(conn, "RELEASE SAVEPOINT cockroach_restart")
break
except psycopg2.OperationalError as e:
if e.pgcode != psycopg2.errorcodes.SERIALIZATION_FAILURE:
# A non-retryable error; report this up the call stack.
raise e
# Signal the database that we'll retry.
onestmt(conn, "ROLLBACK TO SAVEPOINT cockroach_restart")
# The transaction we want to run.
def transfer_funds(txn, frm, to, amount):
with txn.cursor() as cur:
# Check the current balance.
cur.execute("SELECT balance FROM accounts WHERE id = " + str(frm))
from_balance = cur.fetchone()[0]
if from_balance < amount:
raise "Insufficient funds"
# Perform the transfer.
cur.execute("UPDATE accounts SET balance = balance - %s WHERE id = %s",
(amount, frm))
cur.execute("UPDATE accounts SET balance = balance + %s WHERE id = %s",
(amount, to))
# Execute the transaction.
run_transaction(conn, lambda conn: transfer_funds(conn, 1, 2, 100))
with conn:
with conn.cursor() as cur:
# Check account balances.
cur.execute("SELECT id, balance FROM accounts")
rows = cur.fetchall()
print('Balances after transfer:')
for row in rows:
print([str(cell) for cell in row])
# Close communication with the database.
conn.close()
Then run the code:
$ python txn-sample.py
The output should be:
Balances after transfer:
['1', '900']
['2', '350']
However, if you want to verify that funds were transferred from one account to another, use the built-in SQL client:
$ cockroach sql --insecure -e 'SELECT id, balance FROM accounts' --database=bank
+----+---------+
| id | balance |
+----+---------+
| 1 | 900 |
| 2 | 350 |
+----+---------+
(2 rows)
What's Next?
Read more about using the Python psycopg2 driver.
You might also be interested in using a local cluster to explore the following CockroachDB benefits:
- Client Connection Parameters
- Data Replication
- Fault Tolerance & Recovery
- Automatic Rebalancing
- Cross-Cloud Migration
- Follow-the-Workload
- Automated Operations
Was this helpful?
