pyodbc: How to retry to recover from Transient errors?

Question

I've an API hosted on Flask. It runs behind a Tornado server. What is happening is that sometimes changes made on the UI are not reflected in the database. Also a few of the scripts I have running gives any of the 3 following errors:

1. pyodbc.Error: ('08S01', '[08S01] [Microsoft][ODBC SQL Server Driver]Communication link failure (0) (SQLExecDirectW)')
2. pyodbc.Error: ('01000', '[01000] [Microsoft][ODBC SQL Server Driver][DBNETLIB]ConnectionWrite (send()). (10054) (SQLExecDirectW)')
3. pyodbc.Error: ('01000', '[01000] [Microsoft][ODBC SQL Server Driver][DBNETLIB]ConnectionRead (recv()). (10054) (SQLExecDirectW)')

This is the snippet of my Flask API code:

class Type(Resource):

    def put(self):
        parser = reqparse.RequestParser()
        parser.add_argument('id', type = int)
        parser.add_argument('type', type = int)
        args = parser.parse_args()

        query = """
        UPDATE myDb SET Type = ? WHERE Id = ?
        """

        connection = pyodbc.connect(connectionString)
        cursor = connection.cursor()
        cursor.execute(query, [args['type'], args['id']])
        connection.commit()
        cursor.close()
        connection.close()

api.add_resource(Type, '/type')

Is there any retry logic I can add on the cursor.execute line? I've no idea how to deal with transient errors with python. Please help.

Answer 1

Per my experience, I think may be you can try to use the code below to implement the retry logic.

import time

retry_flag = True
retry_count = 0
while retry_flag and retry_count < 5:
  try:
    cursor.execute(query, [args['type'], args['id']])
    retry_flag = False
  except:
    print "Retry after 1 sec"
    retry_count = retry_count + 1
    time.sleep(1)

Hope it helps.