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

Been doing this for about 18 months now with my garden office setup. Running a Pi 4 with Grafana + InfluxDB stack, pulling data from my Victron MPPT via VE.Direct USB cable. Honestly transformed how I understand my system — you can actually see when the clouds roll in, watch battery SOC trends overnight, the lot.

I was sceptical about going the DIY route initially versus just using Victron's VRM portal, but having everything local means no dependency on cloud services, no account needed, and the data is mine. VRM is fine but it's another login and another thing that could disappear one day.

A few things I'd flag for anyone considering it:

• The Victron community scripts on GitHub are a decent starting point
• Pi 3B+ works fine if you've got one lying around, no need to splash out on a Pi 4
• Use a good quality SD card or, better yet, boot from a USB SSD — SD cards fail eventually with constant writes

The one headache has been keeping InfluxDB and Grafana playing nicely together after updates. Not for the faint-hearted if you're not comfortable with the command line.

Anyone else doing something similar but with a different inverter? Curious whether there are decent community integrations for Growatt or Solis units — I've seen mixed reports. Also wondering if anyone's gone down the Home Assistant route instead of a pure Grafana stack.

@JackeryNerd interesting setup. Been thinking about doing something similar for my static caravan — currently just relying on the Victron app which obviously requires connectivity, and signal out there is patchy at best.

Few questions if you don't mind:

• Does the Pi pull data continuously or at set intervals? Wondering about SD card wear over 18 months
• What happens to the logged data if the Pi loses power during a bad battery cycle?
• Is there much CPU overhead running both Grafana and InfluxDB on a Pi 4?

I've got a Victron SmartSolar 100/30 and a BMV-712, so presumably both could feed into this via VE.Direct. Tempted to try it but don't want to brick a Pi learning on the job. Is there a straightforward guide you followed or did you piece it together from various sources?

@JackeryNerd been running almost identical kit for two years on my tiny house build — Pi 4, Grafana/InfluxDB, VE.Direct to USB. Couple of things worth adding:

Node-RED sits in the middle of my stack as a data broker — far easier to massage the raw VE.Direct output before it hits InfluxDB than wrestling with custom scripts.

Also strongly recommend adding a UPS HAT for the Pi itself (the Geekworm UPS HAT works a treat). Nothing worse than a grid blip corrupting your SD card and losing months of historical data. Ask me how I know... 🙃

@RetiredEngineer61 for a static caravan this stack is absolutely worth it — even a Pi Zero 2W handles the logging duties fine if power draw is a concern. Pair it with Victron's VRM portal as a backup and you've got belt-and-braces monitoring.

Running this on my narrowboat with one addition worth mentioning — a UPS HAT (I use the Waveshare one) so the Pi survives voltage dips when the inverter kicks in. Lost two SD cards before I sorted that.

Also worth considering venus-os on the Pi itself rather than just pulling VE.Direct data — turns it into a proper Victron GX device. You get the full Victron ecosystem including remote monitoring via VRM portal if you ever want cloud backup alongside your local Grafana stack.

@RetiredEngineer61 for a static caravan with occasional visits, the VRM portal alone might honestly be enough without the Pi complexity — worth considering before going down this rabbit hole.

One practical tip: boot from SSD rather than SD card. SD cards and 24/7 logging do not get along.

@DorsetSolar good shout on the UPS HAT — that's the one thing people consistently overlook until their Pi corrupts an SD card mid-discharge and they lose months of data.

Which is the other thing nobody's mentioned: don't run InfluxDB on an SD card long-term. Constant write cycles will kill it. Boot from SD if you must but mount your InfluxDB data directory on a proper SSD over USB3. Been running mine this way on the shepherd's hut setup for over a year, zero issues.

@RetiredEngineer61 for a static caravan you'd be fine with a Pi Zero 2W honestly — Pi 4 is overkill unless you're doing something else with it. Lower parasitic draw matters when you're off-grid.

@BatteryTim corrupted SD cards are basically the initiation ritual for this hobby — mine took out three before I switched to booting off a USB SSD on the Pi 4, which has been rock solid for eight months now in the shepherd's hut.

Also worth noting for anyone on Victron kit: if you've got a Cerbo GX already, it exposes a proper MQTT broker locally so you can skip the VE.Direct USB faff entirely and just point your InfluxDB telegraf config straight at mqtt://venus.local — massively cleaner setup and you get all your device data in one feed rather than daisy-chaining cables.

Fogstar Drift cells don't half generate satisfying charge curves to stare at on a Grafana dashboard at midnight, ask me how I know.

@CamperCarl three SD cards is almost impressive tbh

Switched mine to a USB SSD boot ages ago on the cabin Pi — night and day for reliability. Running a Fogstar Drift 100Ah so power dips aren't unusual and the SSD just... handles it better somehow.

Also worth mentioning — if you're pulling VE.Direct data, venus-data-collector on GitHub saved me a lot of messing about vs writing something from scratch. Can't remember who pointed me toward it originally.

Still occasionally lose a data point here and there but nothing catastrophic. Grafana dashboards are oddly satisfying once they're working properly.

@Fogstar_Fan USB SSD boot was genuinely one of the best things I did for my setup too — rock solid ever since. I'm also running the Grafana/InfluxDB stack like @JackeryNerd but added a small Python script to push daily summaries to a Telegram bot, so I get a morning digest of yesterday's generation and battery state without needing to log into dashboards. Took an afternoon to set up and I find it far more useful day-to-day than constantly checking graphs. Happy to share the script if anyone's interested.