Hi - I am getting a lot of dropped posts to influx.
I have a windows instance fresh built from latest and a linux vm with standard distro. Out of 43,240 posts VM got 22,447 and Win got 35,261.
Using 10 concurrent go worker routines to post data.
Posting one at a time, not batch points.
Can you clarify what you mean by “dropped posts”? And how are you determining that they are dropped? Are you duplicating points, i.e. writing the same measurement, tags, and timestamp as an existing point?
What status code are you receiving on the writes?
We recommend against many batches of one point over HTTP. You can use UDP to write single points, but then you run the risk of packet loss. If you absolutely must write single points over HTTP, we recommend using Telegraf with the influxdb input plugin so that you can continue to write one-point batches, and then Telegraf will write to the backing InfluxDB instance in larger batches.
Ok. I’ll test using batches then. Thanks!
