"user_data" needs to be encoded in base64 without newlines

[ogrand]

Hello,

Bosh Openstack cpi send a POST request for instance creation, with body content user_data base64 encoded row containing newlines \n every 60 characters:

body: {
  "server": {
    "flavorRef": "s3.small.1",
    "name": "vm-d35f15af-9840-44be-8d95-a2b1e1ad5358",
    "imageRef": "97727138-8334-4188-9b1e-4x558599d99b",
    "user_data": "eyJzZXJ2ZXIiOnsibmFtZSI6InZtLWQzNWYxNWFmLTk4NDAtNDRiZS04ZDk1\nLWEyYjFlMWFkNTM1OCJ9LCJuZXR3b3JrcyI6eyJ0Zi1uZXQtY29hYi1kZXBs\ncy1pb

Recent upgrade (openstack 22.3) is less permissive for base64 decoding and refuses newlines \n inside base64 encoded value.

To be compliant, it will be nice to replace encoding format in /var/vcap/packages/bosh_openstack_cpi/vendor/bundle/ruby/*/gems/fog-openstack-*/lib/fog/openstack/compute/models/server.rb ruby script:

Replace [ascii_userdata].pack('m') with [ascii_userdata].pack('m0') which disable newlines assertion.

Thanks.
Olivier

[rkoster]

@ogrand could you create a PR for that?

[ogrand]

Hello Ruben, I create a PR in fog-openstack repository for this issue.
Hope It will be merged soon.

Regards. Olivier

[ogrand]

Hello Ruben, do you know if fog-openstack repository included in bosh openstack cpi is always active.
Nothing new for my PR : fog/fog-openstack#528

[ramonskie]

it seems that the fog-openstack repo is dying as there is no activity since august

[ramonskie]

@ogrand the upstream pr has been merged
could you make a pr so that the the cpi is using the updated version.