Losing data whenwriting at high throughput

We’re planning on using InfluxDB 2.0 for writing data at a very high frequency. I have noticed that, when writing data using C#'s “InfluxDB.Client” library, most of the data simply gets dropped and never actually gets written to the timeseries. When I lower the rate at which is written, everything seems to work fine.
I’m not sure if this is a library issue, or a server issue. Either way, what’s the correct way to make sure none of the data gets dropped?

It turns out to be an issue where the library doesn’t flush correctly. I can probably work around this :slight_smile:

1 Like

Hello @twim,
Can you please share the workaround? I’m sure the community would benefit from your knowledge. Thank you.

I actually ended up not using the library and simply making a HTTP connection to InfluxDB manually, en just writing line protocol data to it. Since we only have write datapoints, this was the quickest way to fix our problem.

2 Likes