InfluxQL parsing error when grouping by tag named `name`

[hiltontj]

Problem

A user has reported the following error when performing the InfluxQL query:

Query

SELECT time, name, read_bytes, write_bytes FROM diskio GROUP BY name LIMIT 1

Error

error in InfluxQL statement: parsing error: invalid InfluxQL statement at pos 0. Parsing Error: Nom("SELECT time, name, read_bytes, write_bytes FROM diskio GROUP BY name LIMIT 1", Tag)

This is probably appropriate, but worth verifying.

When quoting the name column, which in their case, is a tag, the response structure includes the tag as a field, instead of a tag for grouping:

Query

SELECT time, \"name\", read_bytes, write_bytes FROM diskio GROUP BY \"name\" LIMIT 1

Response

{"results":[{"statement_id":0,"series":[{"name":"diskio","columns":["time","name","read_bytes","write_bytes"],"values":[[1743544020000000000,"nvme0n1",9269587968,180620936192],[1743544020000000000,"nvme0n1p1",7719936,1024],[1743544020000000000,"nvme0n1p2",5608448,139264],[1743544020000000000,"nvme0n1p3",9254359040,180620795904],[1743544020000000000,"zram0",1458176,4653056]]}]}]}

Purpose of this issue

Initially we should reproduce the above behaviour with a test.

Following that, figure out where the problem may be. It is likely due to the fact that core uses some custom logic for GROUP BY. If there is an easy fix in there we should probably make it but if not, we may revisit the work that has been done recently on GROUP BY in IOx, and whether a sync could provide a solution.

[hiltontj]

Added a reproducer in 2869449

[hiltontj]

The problem seems to be that the InfluxQL parser parses the tag name in the GROUP BY clause as \"name\", including the leading and trailing \", which does not match the field name in the schema, which is just name.

I will see about whether or not we can pass on having to handle the parsed GROUP BY representation directly and rely on what the internal iox_query planner is doing to determine the GROUP BY tags.