input field \"uptime\" on measurement \"system\" is type float, already exists as type integer

[joeplaa]

I'm new to Grafana and Influxdb, so I was very happy to find this repository to get me started. So thanks for putting this together and sharing it.

However, after a few days, the dashboards stop displaying guest specific information and all nodes have status "offline" with "N/A" uptime. The only suspicious thing I can find are these errors in the Influxdb logs:

[httpd] 172.19.0.4 - - [19/Jul/2021:07:40:13 +0000] "POST /write?consistency=any&db=telegraf HTTP/1.1 " 204 0 "-" "Telegraf/1.19.1 Go/1.16.5" 8dcd86c4-e864-11eb-93a6-0242ac130006 93317
ts=2021-07-19T07:40:17.654949Z lvl=info msg="Failed to write point batch to database" log_id=0VO~LV70000 service=udp db_instance=db_proxmox error="partial write: field type conflict: input field \"uptime\" on measurement \"system\" is type float, already exists as type integer dropped=12"

I'm using the docker-compose file from this repo pretty much unmodified and my nodes are on Proxmox 6.4-13.

[joeplaa]

Edit: thought I fixed it, but the error is back.

Fixed it by manually converting the column/metric type to float.

• Open Chronograf client <ip>:8888
• Go to Explore tab
• Run query (this can take a while):

SELECT "uptime"::float FROM "db_proxmox"."autogen"."system" 

I think this is Proxmox metric server when you're also running telegraf anywhere. I think Telegraf correctly creates an int if it gets there first, but proxmox is sending floats which are incompatible.

[franklupo]

ok, what can we do to resolve the problem?

best regards