Cannot remove stale/ghost EVCS devices from VRM-Dashboard

Had a similar headache with my Cerbo GX last month. Was testing a cheap EVCS integration and it half-worked, then I removed it — but the ghost device just sat there on VRM like it owned the place. Tried the usual stuff, restarting the Cerbo, clearing cache on the browser. Nothing shifted it.

Eventually found that doing a full VRM dashboard reset (the "reset to default" option in the dashboard customisation panel) nuked it properly. Lost all my custom widgets in the process which was annoying, had to rebuild everything.

Not ideal but it worked.

Bit different to your setup — I'm not running 3-phase or Pylontech, just a modest Fogstar Drift 200Ah LiFePO4 on a 12V cabin system with a Victron MultiPlus-II and a couple of SmartSolars. But VRM seems to handle ghost entries the same way regardless.

One thing worth checking — if you're integrating a wallbox via MQTT or Modbus, sometimes the device ID gets cached in Venus OS itself rather than just on VRM. Might be worth SSHing into the Cerbo and poking around in /data/conf or similar. There are some stored device configs in there that don't always clean up after themselves.

Anyone else had this with openWB specifically? Curious whether it's a Venus OS v3.7x thing or just how EVCS devices are handled generally. Feels like Victron's EVCS support is still a bit rough around the edges tbh.

@Fogstar_Fan — oh mate, I've been haunted by exactly this with my Cerbo GX setup here in the tiny house.

The nuclear option that actually worked for me: SSH into the Cerbo, navigate to /data/conf/ and manually purge the stale service entries, then do a full VRM device reset from the Remote Console under Settings → VRM online portal → Reset to factory.

Worth checking /etc/venus/dbus-service-list too — ghost EVCS entries can persist there like an unwanted houseguest.

Critical caveat: back up your VEConfig files first. I lost a carefully tuned MPPT profile once doing similar surgery and it was deeply unpleasant.

Also, Victron's community forum has a specific thread on EVCS ghost devices from firmware 3.x — worth cross-referencing before you go digging around in the filesystem.

@Fogstar_Fan had the same ghost EVCS sitting on my VRM for weeks during my van build. What actually cleared it for me was SSHing into the Cerbo and manually deleting the device instance from /data/conf/vrmlogger.conf — then a full reboot.

Worth backing up that config first obviously.

The VRM portal itself has no UI option to remove orphaned devices which is a bit rubbish tbh. Victron support confirmed it's a known limitation — upvote the relevant feature request on their community portal if you haven't already.

If the usual dashboard purge tricks aren't shifting it, check whether the EVCS device still has a valid VRM installation ID cached — VRM sometimes refuses to drop widgets that think they're still "live" even after physical removal. SSH into the Cerbo, run dbus-spy and look for any lingering com.victronenergy.evcharger entries; kill the relevant service and it should vanish from the portal within the next sync cycle. Learned this the hard way when my own cabin setup had a ghost Zaptec haunting my VRM for a fortnight despite never actually working properly in the first place.

Really useful thread this — had something similar crop up on my motorhome setup when I was testing a dodgy integration before settling on a proper Victron stack.

One thing nobody's mentioned yet: if the ghost device is still showing an active portal ID in VRM settings, you'll need to explicitly revoke its installation access from the VRM portal admin panel rather than just removing it from the dashboard view. The dashboard purge won't touch it if VRM still considers it a registered device.

Also worth checking — is the Cerbo's VRM online status actually syncing properly? Sometimes a forced re-sync (Settings → VRM Online Portal → disable/re-enable) kicks the portal into reassessing what's genuinely connected versus what's stale.

@SussexSolar makes a good point about installation IDs — that's probably the root cause here rather than anything on the dashboard side.