Fargate aws-cli:latest installing Python Pandas

Question

Hi! I am installing Python Pandas into a Fargate instance (aws-cli:latest) defined by the Dockerfile below. 
The goal is to use this instance as a ETL job.  Essentially from python Pandas mostly just used for CSV transformations.

Dockerfile:
```
FROM XXXXXXXXXXXX.dkr.ecr.eu-central-1.amazonaws.com/aws-cli:latest
RUN yum install unzip -y -q
RUN python --version
RUN yum -y install https://dl.fedoraproject.org/pub/epel/epel-release-latest-7.noarch.rpm
RUN rpm -Va --nofiles --nodigest
RUN yum -y install python-wheel
RUN yum -y install python-pandas
...
```
Since Pandas were not available from core repos i started by adding a repo for epel, and installed python-pandas from there. Added also some suggestions (rpm -Va --nofiles --nodigest, yum -y install python-wheel)
But as you can see below it complains because it does not find python-matplotlib, tried to install this package but could not find it.

Since i don't need to plot anything anyways, i tried the suggestion of --skip-broken, this allowed me to complete the installation, but when importing from my python script (import pandas as pd), gives the error ImportError: No module named pandas.

What am i missing here?

Answer

It end up just changing the Dockerfile to:

```
RUN yum install python2-pip -y
RUN pip install pandas
```

Note: by default AMI2 ships with python 2.7

Fargate aws-cli:latest installing Python Pandas

Relevanter Inhalt