Long write times and write hangs with xts and influx_write()

[ckatsulis]

Just looked at a quick profile of influx_write. Seems that there are extremely long operations associated with purrr::map again. That was the culprit last time I had suggested replacing that logic with something quicker.

Don't know if you have time to delve in. In general for xts object with 4 cols and ~1MM rows, it is either hanging or not completing. The problem goes away when I push smaller portions to influx_write() so I imagine it is something to do with the map function, memory allocation, etc.

[dleutnant]

Thanks, I'll take a look. Did you try to optimize (i.e. increase) the max_points parameter? It might significantly result in much faster writes ...(dependend on connection and hardware of the InfluxDB instance)

[ckatsulis]

I have tried; but seems to hang when I give it more than 100k lines to write. I can experiment more with the max lines and revert back.

Chris Sent from my iPhone

On Apr 21, 2018, at 10:11, Dominik Leutnant [email protected] wrote:

Thanks, I'll take a look. Did you try to optimize (i.e. increase) the max_points parameter? It may significantly result in much faster writes ...(dependend on connection and hardware of the InfluxDB instance)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.