Anyone else using a Raspberry Pi to log their Victron data locally?

Been running exactly this setup for about 18 months across my shepherd's hut install and I wouldn't go back to relying on VRM alone.

My current stack is a Pi 4 (2GB) connected to the Cerbo GX via Ethernet, pulling data through the MQTT broker that Victron exposes locally. From there I'm feeding into InfluxDB and visualising everything through Grafana. Takes a bit of initial faff to get the dashboards sorted but once it's running it's rock solid.

The main advantage for me is data retention and granularity. VRM gives you decent historical data but I wanted 1-minute resolution going back years, not just the summary graphs. I can now correlate my Fogstar Drift 200Ah battery's state of charge against ambient temperature readings I'm pulling in from a separate DS18B20 sensor, which has been genuinely useful for understanding capacity loss in winter.

A few things worth mentioning for anyone starting out:

• Use a good quality SD card or better yet boot from a USB SSD — SD cards do die eventually with constant write cycles
• Enable MQTT on the Cerbo under Settings > Services (it's off by default)
• Set InfluxDB to downsample older data or your storage fills up faster than you'd expect

Running the same approach on a minimal Pi Zero 2W for my van conversion as a lightweight logger when I'm off-grid without internet. No Grafana on that one, just logging to CSV for later review.

Anyone else doing anything clever with alerting? I've got basic threshold alerts set up through Grafana's notification channels but I'm curious whether others have pushed data into Home Assistant for more sophisticated automation.

@WD40Wizard78 doing something similar on the narrowboat but running a Pi Zero 2W instead — didn't see the point in paying for a Pi 4 just to shove data into InfluxDB.

Couple of questions though:

• Are you pulling data directly from the VE.Direct port or going via the Cerbo's MQTT broker?
• What happens to your local logging when the Pi drops off the network mid-journey? Do you get gaps or does it buffer somehow?

Main reason I bothered setting this up was VRM going down at the worst possible moment when I was trying to troubleshoot a battery issue with my Fogstar cells. Never again relying on someone else's cloud for critical data.

Grafana dashboards are decent once you've wrestled with them for a week, I'll give it that.

Great thread, @WD40Wizard78 — been meaning to post about this for ages.

Running a Pi 4 (4GB) here on my static caravan setup, pulling data from the Cerbo via MQTT into InfluxDB, then visualising in Grafana. Been solid for about eight months now.

One thing I'd add that nobody's mentioned yet: set your InfluxDB retention policies sensibly from the off. I made the mistake of keeping full-resolution data indefinitely and the SD card was cooked within three months. Now I downsample anything older than a week using continuous queries and the card's been fine since I switched to a decent Samsung Endurance Pro anyway.

@BurnWalker — curious whether vibration from the engine causes you any grief with SD card reliability on the narrowboat? Tempted to suggest a USB SSD instead if that's an issue for you.

@WD40Wizard78 oh brilliant thread — I've been doing exactly this on my static caravan setup and honestly the local logging was the thing that finally convinced me I actually understood what my Fogstar Drift batteries were doing overnight.

Quick question though — are any of you pulling data from multiple VE.Direct devices simultaneously? I've got a Victron SmartShunt AND an MPPT 100/30 both wanting to talk to the Pi and I had a right faff getting the USB-to-VE.Direct cables playing nicely together without one of them randomly dropping off.

Also — @BurnWalker curious whether the Zero 2W struggles at all with keeping up if you've got polling intervals set fairly tight? I went Pi 4 partly because I was worried about processing overhead with Grafana dashboards, but maybe I'm massively overspecced for what amounts to a glorified spreadsheet 😄

@BurnWalker curious how you're finding the Zero 2W for this — I initially went down that route on the narrowboat but hit CPU ceiling pretty quickly once I added InfluxDB retention policies and a Grafana instance running concurrently. Swapped to a Pi 4 (4GB) and the headroom difference is significant.

One thing nobody's mentioned yet: power consumption of the Pi itself matters enormously on a boat. My Pi 4 draws around 3-4W continuously, which is ~90Wh/day — not trivial when you're already managing a tight winter energy budget. I've got it on a scheduled shutdown via cron during overnight low-SOC periods, with the Cerbo continuing to log independently and the Pi syncing on wake-up.

USB-C power comes straight off a Victron Orion 12/5 rather than a cheap converter — learned that lesson the hard way after a ripple-related SD card corruption.

@OBT_Solar replied:

Great thread, been lurking on this topic for a while. Running a similar setup on my off-grid barn conversion — Pi 4 (2GB) pulling data from a Cerbo GX via the MQTT broker, feeding into InfluxDB and Grafana. One thing I'd add that nobody's mentioned yet: set your InfluxDB retention policies sensibly from the start. I made the mistake of keeping full-resolution data indefinitely and filled a 32GB card within four months. Now I downsample anything older than a week using continuous queries and it's been rock solid since.

Also worth enabling the Pi's watchdog timer so it auto-reboots if the system hangs — learned that one the hard way after coming back from a fortnight away to find two weeks of gaps in my data. Small config change, massive peace of mind. Happy to share my configs if anyone wants them.