How to connect MySQL database using Python+SQLAlchemy remotely?
The classic answer to this issue is to use 127.0.0.1
or the IP of the host or the host name instead of the "special name" localhost
. From the documentation:
[...] connections on Unix to localhost are made using a Unix socket file by default
And later:
On Unix, MySQL programs treat the host name localhost specially, in a way that is likely different from what you expect compared to other network-based programs. For connections to localhost, MySQL programs attempt to connect to the local server by using a Unix socket file. This occurs even if a --port or -P option is given to specify a port number. To ensure that the client makes a TCP/IP connection to the local server, use --host or -h to specify a host name value of 127.0.0.1, or the IP address or name of the local server.
However, this simple trick doesn't appear to work in your case, so you have to somehow force the use of a TCP socket. As you explained it yourself, when invoking mysql
on the command line, you use the --protocol tcp
option.
As explained here, from SQLAlchemy, you can pass the relevant options (if any) to your driver either as URL options or using the connect_args
keyword argument.
For example using PyMySQL, on a test system I've setup for that purpose (MariaDB 10.0.12, SQLAlchemy 0.9.8 and PyMySQL 0.6.2) I got the following results:
>>> engine = create_engine(
"mysql+pymysql://sylvain:passwd@localhost/db?host=localhost?port=3306")
# ^^^^^^^^^^^^^^^^^^^^^^^^^^
# Force TCP socket. Notice the two uses of `?`
# Normally URL options should use `?` and `&`
# after that. But that doesn't work here (bug?)
>>> conn = engine.connect()
>>> conn.execute("SELECT host FROM INFORMATION_SCHEMA.PROCESSLIST WHERE ID = CONNECTION_ID()").fetchall()
[('localhost:54164',)]
# Same result by using 127.0.0.1 instead of localhost:
>>> engine = create_engine(
"mysql+pymysql://sylvain:[email protected]/db?host=localhost?port=3306")
>>> conn = engine.connect()
>>> conn.execute("SELECT host FROM INFORMATION_SCHEMA.PROCESSLIST WHERE ID = CONNECTION_ID()").fetchall()
[('localhost:54164',)]
# Alternatively, using connect_args:
>>> engine = create_engine("mysql+pymysql://sylvain:passwd@localhost/db",
connect_args= dict(host='localhost', port=3306))
>>> conn = engine.connect()
>>> conn.execute("SELECT host FROM INFORMATION_SCHEMA.PROCESSLIST WHERE ID = CONNECTION_ID()").fetchall()
[('localhost:54353',)]
As you noticed, both will use a TCP connection (I know that because of the port number after the hostname). On the other hand:
>>> engine = create_engine(
"mysql+pymysql://sylvain:passwd@localhost/db?unix_socket=/path/to/mysql.sock")
# ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
# Specify the path to mysql.sock in
# the `unix_socket` option will force
# usage of a UNIX socket
>>> conn = engine.connect()
>>> conn.execute("SELECT host FROM INFORMATION_SCHEMA.PROCESSLIST WHERE ID = CONNECTION_ID()").fetchall()
[('localhost',)]
# Same result by using 127.0.0.1 instead of localhost:
>>> engine = create_engine(
"mysql+pymysql://sylvain:[email protected]/db?unix_socket=/path/to/mysql.sock")
>>> conn = engine.connect()
>>> conn.execute("SELECT host FROM INFORMATION_SCHEMA.PROCESSLIST WHERE ID = CONNECTION_ID()").fetchall()
[('localhost',)]
# Alternatively, using connect_args:
>>> engine = create_engine("mysql+pymysql://sylvain:passwd@localhost/db",
connect_args= dict(unix_socket="/path/to/mysql.sock"))
>>> conn = engine.connect()
>>> conn.execute("SELECT host FROM INFORMATION_SCHEMA.PROCESSLIST WHERE ID = CONNECTION_ID()").fetchall()
[('localhost',)]
No port after the hostname: this is an UNIX socket.
In my setup (I'm using mysql-python) just using 127.0.0.1 instead of localhost in the MySQL SQLAlchemy url works. The complete url I'm using exactly for that scenario (tunnel with local port 3307) is:
mysql:/user:[email protected]:3307/
I'm using SQLAlchemy 1.0.5, but I guess that doesn't matter too much...
This worked for me:
import pandas as pd
import pymysql
from sqlalchemy import create_engine
cnx = create_engine('mysql+pymysql://<username>:<password>@<host>/<dbname>')
df = pd.read_sql('SELECT * FROM <table_name>', cnx) #read the entire table
Where credentials are added to mysql database like this:
CREATE USER '<username>' IDENTIFIED BY '<password>';
GRANT ALL PRIVILEGES ON *.* TO '<username>' WITH GRANT OPTION;
FLUSH PRIVILEGES;