We are looking on installing Harvest 1.6 and using python3.6. We have not starting moving to python3 yet and this is my first earnest attempt. I see most of the installs reference pip and not pip-3, and I am not seeing much reference to python versioning with regards to Harvest. Are there any high level hints that can be provided as I set off on my trial and error install attempt? Maybe someone has a cheat sheet or doc that I can reference?
Harvest 1.6 is based on Perl, the Python packages dependencies are required for a few extensions that are written in Python3. You should indeed use pip3 for those! For more details you can look at the documentation that comes with the package. Otherwise let me know if you have specific issues, I'll try to help (unfortunately I don't always find the time for community...).
One more thing: we have stopped developing the Perl source-code and have recently released an Alpha version of Harvest 2.0 which is completely in Python3 (with the possibility of moving some parts to Go in the future). This new Harvest supports more databases: Prometheus and InfluxDB in addition to Graphite/Whisper. It doesn't not support 7Modes yet, but that will be added in the Beta released which is scheduled this month.
Just as a follow-up to inform you how it went. Since I am using RHEL 7.1 python 2.7 is the default and I needed to run in a virtualenv to use python3. I was able to get the packages installed however I could not force anything into the "normal" directories. Also I could not find any of the config files. Due to time constraints I wound up installing it in the default environment, and requesting a RHEL 8 box to start moving to 2.0 when it's ready.