Install with Python3

[initrd]

Affected Puppet, Ruby, OS and module versions/distributions

• Puppet: 5.5
• Ruby: 2.0.0
• Distribution: CentOS 7.7
• Module version: master/8b828b695ce1322ee9857ff3f1dcc2eaabb4afe2 with puppet-python 3.0.1

How to reproduce (e.g Puppet code you use)

How do I make the module use the python3/python36 on CentOS 7? With manage_virtualenv, the module uses the 'system' version of Python - 2.7.

Thanks!

[cleverlight]

I can confirm this is an issue for me too. Since Puppetboard upgraded from release tag v1.0.0 to v1.1.0 and dropped support for Python 2.7, this module no longer support CentOS / RHEL.

Support is notionally possible using SCL rh-python3.6 but there are a number of interlinked issues:

1. the virtualenv command, which originates from the puppetlabs-python module, prefixes the passed virtualenv_version parameter with 'python'. This makes it impossible to pass a path to /opt/rh/rh-python36/root/usr/bin/python.

2. I did try symlinking /usr/bin/python3 -> /opt/rh/rh-python36/root/usr/bin/python3.6. The command line python3 --version then correctly reports 3.6.3. I then set the virtualenv_version parameter to 3, which meant that the virtualenv call passed -p python3. /srv/puppetboard/virtenv-puppetboard seems to build out correctly. It contains chardetect, cmark, flask, easy_install, pip etc. and a python3 binary, which reports version 3.6.3.

3. RHSCL installs Python in a non-standard location then, if enabled, appends that new location to the PATH environment variable for all users. Even though Update module dependencies #2 means it runs the right binary, I think Puppetboard is still struggling to see its package directory, because HTTP requests against Puppetboard return a 500 Internal Error. The error log reports:

mod_wsgi (pid=6227): Target WSGI script '/srv/puppetboard/puppetboard/wsgi.py' cannot be loaded as Python module.
mod_wsgi (pid=6227): Exception occurred processing WSGI script '/srv/puppetboard/puppetboard/wsgi.py'.
Traceback (most recent call last):
  File "/srv/puppetboard/puppetboard/wsgi.py", line 15, in <module>
    from puppetboard.app import app as application
  File "/srv/puppetboard/puppetboard/puppetboard/app.py", line 5, in <module>
    from urllib.parse import unquote, unquote_plus, quote_plus
ImportError: No module named parse

In answer to the OP question, @initrd it is possible to get Puppetboard running on CentOS if you force the Puppetboard web interface (not this Puppetboard Puppet module) back to revision v1.0.0:

  class { '::puppetboard':
    manage_git => false,
    manage_virtualenv => true,
    # doesn't like using symlinked python 3.6, I suspect because of the environment variables/path
    # virtualenv_version => '3',
    # lock to puppetboard 1.0.0 because 1.1.0 breaks virtualenv/parse module
    revision => 'v1.0.0',
  }

There are several ways to resolve this issue:
A. Grapple with an SCL install on CentOS7 and work out how to get Puppetboard v1.1.0 running using /opt/rh/rh-python36/root/usr/bin/python3.6.
B. Make Puppetboard v1.0.0 the default for CentOS/RedHat and flag in the docs that v1.1.0 is incompatible because it requires Python > 3.5.
C. Remove the "CentOS" and "RedHat" tags from this module.

I think these are broadly in order of preference, but I'm sorry to say I haven't been able to find a way to implement fix A.

[dd5154]

I also have this issue and used the workaround to force the web interface back to revision v1.0.0

[vegaaz]

I also have this issue.
OS: Debian 10 (Buster)
Back to revision v1.0.0 solved the problem.

[nmaludy]

+1 having this same issue.