I’m not using a proxy, but I’m using a “Telegraf Gateway”. I have several Telegraf instances that gathers data and send it to another telegraf instance (the gateway), which uses the “influxdb listener” input plugin and finally sends the data to the database (InfluxDB).
I see the error in the telegraf logs of the “gatherers” instances.
An update on this issue:
Counting the number of points in the end helped me… every measurement had the same number of points for the given interval except one, which is the “sqlserver_requests” measurement.
It captures the query running on the database, so it does not have a constant amount of points by time interval, that said, every other measurement (which have a constant amount of point per interval) had the same amount of points, therefore none of them lost points. or at least this is what I think.
The “sqlserver_requests” measurement stores the query text as a tag (statement_text). My guess is that some texts cause this issue even if I have no proof of that yet.
I will set the log to debug mode, and also output the values to a file, to see if that captures the problematic points. Any other suggestion about how to debug this issue is appreciated.
