hey - using influx 1.2 on a ubuntu 1604 box w/ dual xeons + 128gb ram.
I just ran a query (using influxdb python DataFrameClient) that pulled all the data from a series from the last 24 hours. It returned 1,307,746 rows x 22 columns (1 tag, 21 fields) in 1min, 33sec. The retention policy is infinity so this should be from one shard, if I understand it right. The db is under load of 800-1000 writes/sec. There are a relatively high proportion null values in the query result. This time is representative of other queries for the same measurement.
My expectation was that queries with aggregation/filtering/etc might take a while but I was surprised that this “load all the data” query is taking such a long time.
My question is - is this expected under the circumstances? Is there something I should be doing differently in terms of schema design? It seems like the write performance I’m getting is great but read is pretty slow.