The engineering marvel that is Percona Monitoring and Management (PMM), has to be, for me, one of the best examples out there of how a large number of vastly different community projects can successfully combine into something whose value and features far exceeds the sum of its component open-source parts.
That’s a lot of words for a simple thing: it’s a metrics mover—it’s PMM!
PMM has exporters that suck system stats from a server or database.
The numbers are safely shuffled through agents across a crowded network toward the sanctuary of the PMM Server.
That’s where your valuable system values are collected and collated, charted and tabled, and where they emerge as fabled knowledge, logically organized and neatly arranged into dashboards.
Dashboards (What you see)
PMM’s dashboards are based on Grafana, and PMM relies on many community-based projects.
Exporters (What you don’t see)
Exporters are specialized programs that run and extract metrics data from a node or database. Every database is different, and so every exporter is too.
Community Meetings (What you may not know)
I tell you what you may already know simply as a prelude to an announcement of what you may not yet know.
We’ve been running monthly community meetings where our Engineering Gurus “talk technical” to share their knowledge of PMM and all its parts.
The format is fairly informal. We use our Discord server’s voice channel. We grab a refreshing region-sensitive hot or cold drink, a very loose agenda, and we talk.
All we need now is someone to listen.
The topics in this month’s meeting:
Exporters: Focusing this month on
node_exporter, and how and why we’re reverting our fork of the Prometheus one.Dashboards: How Percona’s differ from the community’s, and how to migrate them using the scripts in here. Here’s a brief preview.
Convert a dashboard from PMM
This script converts a PMM dashboard so it can be used in an external Prometheus + Grafana installation. It doesn’t need any input from you. It replaces PMM2 labels (
node_name,service_name) that are used in variables with default labels (instance).Convert a dashboard to PMM
This one renames labels and variables interactively because it can’t know in advance the names of the labels you want to use. You must select names for renaming and provide a PMM2 label (
node_name,service_name). The script collects a list of used variables and asks which have to be renamed in variables and expressions.PMM2 labels can be checked in the VM configuration files:
As we’ve already tweeted, our Gurus are not ashamed of showing their consoles. Some are bare and some are full, but they’re always fascinating to watch when they use them with such style.
Why not join us and see for yourself?
