Running multiple CVP downloads simultaneously result in errors

[diegolima]

Hi! I'm using the eos_download.py to provision multiple CVP servers (as part of a cluster) simultaneously. When I do that I see the following error message in all but one server:

# /data/eos_download.py --api TOKEN--ver cvp-2021.1.0 --img rpm
cvp-rpm-installer-2021.1.0 is currently downloading....
100%|######################################################################################################################################################################9| 5.64G/5.65G [02:08<00:00, 22.5MB/s]Traceback (most recent call last):
  File "/data/eos_download.py", line 356, in <module>
    sha512_download_link = (sha512_result.json()["data"]["url"])
KeyError: 'data'

As a result, the CVP installer is downloaded but not the md5 checksum file, and ultimately my playbook fails as the exit code is not 0 (and would fail afterward as the next step would be checking the md5sum).

I believe that this is happening as the session gets invalidated when new ones are created and the script never tries to refresh it before getting the .md5 download link (therefore the script working on only one server, whose session was created last).

[diegolima]

I've submitted pull request #6 to deal with that situation.