InterfaceError (0, '')

I can confirm this is caused by a global cursor which is then later used in some functions. My symptoms were the exact same: intermittent interface errors that would temporarily be cleared up by an apache restart.

from django.db import connection
cursor = connection.cursor() # BAD

def foo():
    cursor.execute('select * from bar')

But, I am using Django on top of Oracle 11.2 so I do not believe this is a bug in the MySQL/python driver. This is probably due to the caching done by apache/mod_wsgi.

You get this error when you have a db.close() call and later try to access the database without creating a new connection. Try to find if you close the connection to the database when you don’t mean to.

I agreed with Moberg. This error is caused when we try to access the database after we have closed the connection. This could be caused by some wrong indentation in the code. Below is my code.

conn = connect()
cur = conn.cursor()
tk = get_tickers(cur)
for t in tk:
    prices = read_price(t, cur)
    if prices != None:
        update_price(t, cur)
        print 'Price after update of ticker ', t, ':'
        p_open, p_high, p_low, p_close = read_price(t, cur)
        print p_open, p_high, p_low, p_close
    else:
        print 'Price for ', t, ' is not available'
    conn.close()

I got the same error as reported by Marian. After dedenting conn.close(), everything worked well. Confirmed that global conn is not an issue.

I had the same problem as for April of 2019 using python 3.7 and Mysql 2.7.

At intermittent intervals, the string (0, ”) would be added at random to my SQL statements causing errors. I have solved the issue by commenting on the closing of the database connection and just leaving the closing of the cursors across my code.

def set_db():
    db = pymysql.connect(host='localhost',
                         user="root",
                         passwd="root",
                         db="DATABASE")
    return db


def execute_sql(cnx, sql_clause, fetch_all):

    if sql_clause and sql_clause is not None:
        try:
            cnx.execute(sql_clause)
        except Exception as e:
            print("Error in sql: " + sql_clause + str(e))
            return 0
            pass

        if fetch_all:
            result = cnx.fetchall()
        else:
            result = cnx.fetchone()

        return result
    else:
        print("Empty sql.")
        return 0

db = set_db()
cnx = db.cursor()
sql = "SELECT * FROM TABLE"
result = execute_sql(cnx, sql, 1)
cnx.close() #close the cursor
#db.close #do not close the db connection

…

I had the same issue with Flask+pymysql, I was getting an empty tuple as a result in the except: block, something like this "("(0, '')",)" to be specific.

It turned out that the connection was getting closed and later the code tried accessing it which resulted into this error.
So I solved it by referring to above solutions and used a function for connection which assured me a conn every time I had to access the db.

You can recreate this issue by inserting conn.close() just before accessing the cursor.

For reference I used this site which helped me solve this issue.

https://hackersandslackers.com/python-mysql-pymysql/

For me, removing the conn.close() from my function worked. I was trying to access the database again after closing.
I am using flask with AWS.
Also you can try to restart your flask application if it has been running for a long time & if you are also using AWS RDS with MYSQL workbench like in my case, then just check whether your session is expired or not and update the access key and id.

Hope this helps.

with connections.cursor() as cursor:
    res=cursor.execute(sql)

I had the same issue using threading with Python3 and Pymysql. I was getting deadlocks and then I would get hit with InterfaceError (0, ”).

My issue was that I was trying to do a rollback on exception of the query- I believe this rollback was trying to use a connection that no longer existed and it was giving me the interface error. I took this rollback out (because I am OK with not doing rollback for this query) and I just let things go. This fixed my issue.

def delete_q_msg(self, assetid, queuemsgtypeid, msgid):
    """
    Given the paramerts below remove items from the msg queue equal to or older than this.
    If appropriate send them into a history table to be processed later

    :param assetid:
    :param queuemsgtypeid:
    :param msgid:
    :return:
    """
    params = (assetid, queuemsgtypeid, msgid,)
    db_connection = self._connect_to_db()
    sp_sql = "{db}.ps_delete_q_msg".format(db=self._db_settings["database"])
    return_value = []
    try:
        with db_connection.cursor() as cursor:
            cursor.callproc(sp_sql, params)
            return_value = cursor.fetchall()
        db_connection.commit()
    except Exception as ex:
        # i think we dont want rollback here
        # db_connection.rollback()
        raise Exception(ex)
    finally:
        db_connection.close()

    return return_value

I had this same problem and what worked for me in Django was what is described in this answer, which consists of:

Replacing

'ENGINE': 'django.db.backends.mysql'

with

'ENGINE': 'mysql_server_has_gone_away'

on

settings.DATABASES['ENGINE']

and installing with pip the package below:

mysql_server_has_gone_away==1.0.0