We’re trying to avoid large queries from crashing influxDB. The settings we’ve been looking at are max-select-point and max-select-series. The problem is that I don’t think those are adequate - I prefer a limit on “selected points * selected series” or something to that effect.
For instance, say I want to prevent doing a “select * from” a huge measurement with many fields and points. If I select many points I should only use a few fields, so I set the max-select-series to be a rather modest value (20k). The problem is that it turns out that limit can be hit solely due to series cardinality caused by tags - even something like this can happen:
>show tag values cardinality with key="xxxxxx" ERR: max-select-series limit exceeded: (20001/20000)
So if I want to select only 100k points, but it turns out they have a very diverse set of tags - even though total series cardinality isn’t that high - it sounds like I can hit the max-select-series limit quite easily.
Overall we want to be able to query more fields at least when the number of points is not too large; is there an option for that in the config file? Thanks a lot