Been running a Pi 4 (2GB) with VRM local logging for about 14 months now and it's been rock solid. I've got it pulling data from my Cerbo GX via MQTT every 30 seconds and pushing it into InfluxDB, then visualising everything through Grafana. The dashboard shows SOC, AC loads, PV yield, and battery temps all on one screen — far more useful than the VRM portal for day-to-day tweaking.
The main reason I went local rather than relying purely on VRM cloud is obvious if you're off-grid — internet goes down and you're flying blind otherwise. Having everything stored on a local NAS means I've got years of data I can actually query properly.
A few things worth mentioning if you're setting this up:
- Use a good quality SD card or boot from USB/SSD — SD cards die. Lost three months of data early on learning that lesson.
- The Victron MQTT topics are well documented —
N/<VRM ID>/vebus/etc. Once you understand the structure it's straightforward. - Set your InfluxDB retention policy sensibly — I keep full resolution for 90 days, then downsampled for long-term storage.
One thing I'm currently working on is correlating my EV charging sessions (I'm on a Type 2 setup pulling from the off-grid system) against solar yield and battery state. Trying to automate charging windows when SOC is above 85% and PV is genuinely surplus. The data's all there in Grafana, just need to close the automation loop with Node-RED.
Anyone else doing something similar with EV load management through Pi-based automation? Keen to see how others have approached it.