Source often starts failing to write to influx and never recovers until manually restarted

[w-le]

Source often starts failing to write to influx and never recovers until manually restarted

https://gist.github.com/w-le/7a53c4696babe5b692c8ea0e39f2b5c0

[chillfox]

likely fixed by #46

[stakach]

We'll need to write a migration to convert integers to floats in influxdb
https://community.influxdata.com/t/chage-field-type-from-string-to-float/16222/15

although I think it'll be harder than this post makes out as we're using Flux v2

[stakach]

this looks like the only way to do a migration: https://stackoverflow.com/questions/70248103/flux-influxdb-conversion-of-a-table-from-float-to-integer

just need to work out how to automate all tables and fields

[w-le]

likely fixed by #46

actually I believe they are seperate issues because we see instances where Source experiences the type mismatch and keeps pushing other data fine, and also instances where Source doesn't have any type mismatch issues and just randomly stops pushing all data until manually restarted.

it would be great if it can detect when it's not successfully pushing data when it should be and either just exit with an error (service managers like docker or k8s will restart it) or try to start pushing data again