Feature request: Perforce Helix Core monitoring with P4Prometheus
Opened this issue · 2 comments
Use case
As part of the deployment of the Helix Core module, users should be able to configure monitoring so that it can be automatically instrumented for them.
Solution/User Experience
- Perforce's p4prometheus project is a suitable option, although we could extend this module to instrument other monitoring solutions as well through configuration.
- Amazon Managed Grafana and Amazon Managed Service for Prometheus could be configured as optional configuration as well for devs that prefer managed services instead of self-managing these.
Alternative solutions
- CloudWatch?
This is a great issue.
According to the p4prometheus documentation we need to install the following on the Helix Core server:
- node_exporter
- p4prometheus
- monitor_metrics.sh
- monitor_wrapper.sh and monitor_metrics.py
On other related servers, e.g. running Swarm, Hansoft, Helix TeamHub (HTH), etc, install:
- node_exporter
We can certainly automate this process inside of the Helix Core module.
p4prometheus also requires the following to be installed on a "monitoring server":
- grafana
- prometheus
- victoria metrics (optional but recommended due to performance and more efficient data storage)
- node_exporter
- alertmanager (optional)
How much of this we can replace with managed prometheus and grafana is TBD.
I'd like to wait until we begin shipping EKS support for modules before picking this up, as I'd like to build for that platform first.