I have the need to monitor performance of HPC systems running on centos7. Juju makes the process of deploying SLURM easy. But how about performance monitoring for centos?
First, I thought to port the telegraf charm, but since it turned out to use charm-helpers and layer-basic (that are broken for anything thats not using apt) I ended up writing a new, reduced, in bash. It also avoids all cloudinit hacks needed for centos since python3 is not needed at all.
In other words: just deploy
This subordinate charm can be used with charms for centos7 and relates to prometheus. The docs include an example deployment of this.
If you run centos charms out there, here is your basic performance monitoring: telegraf prometheus | Juju