AWS Lambda function 连接到 SQL 服务器与 ZA7F5F35426B9274173FC9231B563

Question

I've been stuck trying to connect to an SQL Server using AWS Lambda functions for a long while now.

To do so i'm trying to use any library (tried with pyodbc, pypyodbc, etc), packaging everything into a zip file and uploading the code.

The code is pretty much the same for every library, but the errors are different.

The code:

import pypyodbc

def lambda_handler(event, context):
    conn = pypyodbc.connect('DRIVER={SQL Server};'
                      'SERVER=1.1.1.1;'
                      'DATABASE=dbname;'
                      'UID=user;'
                      'PWD=pwd')

    cur = conn.cursor()

    cur.execute("SELECT * FROM Table")

    item_count = 0

    for row in cur:
        item_count += 1

    print(item_count)

    cur.close()
    conn.close()

    return item_count

Common issues that i have covered: - I'm adding to the zip the project contents, not the folder. - I'm also adding to the zip file the libraries needed for the code to run.

If i try to use pyodbc , the zip i'm uploading looks like this:

.idea (dir)
pyodbc (dir)
lambda_function.py
pyodbc.pyd

The error i get:

Unable to import module 'lambda_function': No module named pyodbc

After searching for quite a while about this, i couldn't find anything that helps. Only one comment that said that pyodbc needed to be instaled on a linux enviroment for the lambda function to work. But i don't have that Enviroment available, also i don't know if that will fix this.

If i try to use pypyodbc , the zip i'm uploading looks like this:

The error i get:

module initialization error: 'ODBC Library is not found. Is LD_LIBRARY_PATH set?'

For this one i tried to install multiple python packages suggested by other stackoverflow posts (python-pyodb, unixodbc), but i failed every time.

Then there was one comment around saying "Make sure to put native ODBC libraries under the lib folder in your zip deployment package"

Maybe that is some help? I don't know how to get native ODBC libraries..

Oh and one last thing. Both libraries work if i run them from my local machine. I can get access to the target server. It fails if i do it from the lambda function.

Hopefully someone can help me and, apparently, the whole internet with this.

Answer 1

• you need to know Lambda copy your function in local /var/task/
• create a instance using Lambda official AMI https://docs.aws.amazon.com/lambda/latest/dg/current-supported-versions.html
• start instance, login
• yum install gcc gcc-c++
• go in to /home/ec2-user
• Download the last unixodbc manager from: ftp://ftp.unixodbc.org/pub/unixODBC/
• wget ftp://ftp.unixodbc.org/pub/unixODBC/unixODBC-2.3.5.tar.gz
• tar xvzf unixODBC-2.3.5.tar.gz
• cd unixODBC-2.3.5

• configure it with the correct sysconfdir value

  ./configure --sysconfdir=/var/task --disable-gui --disable-drivers --enable-iconv --with-iconv-char-enc=UTF8 --with-iconv-ucode-enc=UTF16LE --prefix=/home

• make install
• Go to /home dir and copy bin,include,lib,share directory on your computer where the Lambda project is (ex: C:\AWS\Lambda\sql_query)
• install on your EC2 instance the the Microsoft driver libmsodbcsql-13.1.so.9.1 and then copy the driver file on your PC local directory (ex: C:\AWS\Lambda\sql_query\msodbcsql\msodbcsql\lib64 )
• Take a look https://blogs.msdn.microsoft.com/sqlnativeclient/2017/02/04/odbc-driver-13-1-for-linux-released/
• On your computer, in the same root directory create file odbcinst.ini

[ODBC Driver 13 for SQL Server] Description=Microsoft ODBC Driver 13 for SQL Server Driver=/var/task/msodbcsql/msodbcsql/lib64/libmsodbcsql-13.1.so.9.1 UsageCount=1

• On your computer, in the same root directory create file odbc.ini

  [ODBC Driver 13 for SQL Server] Driver = ODBC Driver 13 for SQL Server Description = My ODBC Driver 13 for SQL Server Trace = No

• on your python program use pyodbc:

  import pyodbc def lambda_handler(event, context): server = "xxxxxxxxxxxxxxxxxxxx" database = "xxxxxxxxxxxxxxxxxxxx" username = "xxxxxxxxxxxxxxxxxxxx" password = "xxxxxxxxxxxxxxxxxxxx" cnxn = pyodbc.connect('DRIVER={ODBC Driver 13 for SQL Server};SERVER='+server+';DATABASE='+database+';UID='+username+';PWD='+ password) cursor = cnxn.cursor() ...other things....

• and now play the game !

Answer 2

Here's a Python 3.9 solution that uses Docker and Lambda layers .

See this Github gist comment .

Based on:

• Python 3.9
• AWS's base images for Lambda
• unixODBC-2.3.11

1. Create this Dockerfile :

    # Dockerfile # See https://stackoverflow.com/a/72617192/58876 FROM amazon/aws-lambda-python:3.9 ENTRYPOINT [] WORKDIR /root # Get development tools to enable compiling RUN yum -y update RUN yum -y groupinstall "Development Tools" # Get unixODBC and install it RUN yum -y install tar gzip RUN curl ftp://ftp.unixodbc.org/pub/unixODBC/unixODBC-2.3.11.tar.gz -O RUN tar xvzf unixODBC-2.3.11.tar.gz WORKDIR /root/unixODBC-2.3.11 RUN ./configure --sysconfdir=/opt/python --disable-gui --disable-drivers --enable-iconv --with-iconv-char-enc=UTF8 --with-iconv-ucode-enc=UTF16LE --prefix=/home RUN make install WORKDIR /root RUN mv /home/* . RUN mv unixODBC-2.3.11 unixODBC-2.3.11.tar.gz /tmp # Install msodbcsql RUN curl https://packages.microsoft.com/config/rhel/7/prod.repo > /etc/yum.repos.d/mssql-release.repo RUN yum -y install e2fsprogs openssl RUN ACCEPT_EULA=Y yum -y install msodbcsql mssql-tools --disablerepo=amzn* RUN cp -r /opt/microsoft/msodbcsql . # Install pyodbc # Need "unixODBC-devel" to avoid "src/pyodbc.h:56:10: fatal error: sql.h: No such file or directory" during pip install RUN yum -y install unixODBC-devel RUN export CFLAGS="-I/root/include" RUN export LDFLAGS="-L/root/lib" RUN pip install pyodbc adodbapi pyDes --upgrade -t . # Create odbc.ini and odbcinst.ini RUN echo $'[ODBC Driver 17 for SQL Server]\nDriver = ODBC Driver 17 for SQL Server\nDescription = My ODBC Driver 17 for SQL Server\nTrace = No' > /root/odbc.ini RUN echo $'[ODBC Driver 17 for SQL Server]\nDescription = Microsoft ODBC Driver 17 for SQL Server\nDriver = /opt/microsoft/msodbcsql17/lib64/libmsodbcsql-17.9.so.1.1\nUsageCount = 1' > /root/odbcinst.ini # Generate the zipped file that can be uploaded as a Lambda Layer RUN mkdir -p /opt/python RUN cp -r /root/* /opt/python RUN mv /opt/python/lib /opt RUN mv /opt/python/bin /opt WORKDIR /opt RUN zip -r /pyodbc.zip .

2. Build the Dockerfile and extract the resulting zip file:

    # Build the image (this will take a while...) $ docker build --platform=linux/amd64 -t mssql-lambda . # Copy the zipped file to /tmp/pyodbc.zip on your computer $ docker run --platform=linux/amd64 --rm --volume /tmp:/local-folder mssql-lambda cp /pyodbc.zip /local-folder/

3. Upload /tmp/pyodbc.zip from your computer to AWS as a Lambda Layer.

4. Create and run a Python 3.9 Lambda to test things out: (Make sure you configure it to use the Layer you created in the previous step)

    import pyodbc def lambda_handler(event, context): driver = '{ODBC Driver 17 for SQL Server}' server = 'XXX' port = '1433' database = 'XXX' username = 'XXX' password = 'XXX' # ideally put this in a SecretsManager secret instead of directly in the code connection = pyodbc.connect(f'DRIVER={driver};SERVER={server},{port};DATABASE={database};UID={username};PWD={password}') cursor = connection.cursor() query = """ select @@version """ cursor.execute(query) for row in cursor.fetchall(): print(row[0])

Answer 3

I think your problem is uploading the layer without the appropriate folder structure specified in the documentation