More columns than values in query response

[aviau]

Hello,

I am sure this is a duplicate of something somewhere... but here goes:

Using mean() in a query breaks the response format: there are more columns than values in the query response.

With mean():

Query:
SELECT host, mean(zzz) as hello, host, value as value, zzz, host FROM cpu_value
Response:
{"series": [{"values": [["1970-01-01T00:00:00Z", 34545]], "name": "cpu_value", "columns": ["time", "host", "hello", "host", "value", "zzz", "host"], "tags": {"host": "server01"}}]}

Without mean():

Query:
SELECT host, zzz as hello, host, value as value, zzz, host FROM cpu_value
Response:
{"series": [{"values": [["2015-08-10T14:03:44.833892297Z", 34545, 2.34, 34545]], "name": "cpu_value", "columns": ["time", "zzz", "value", "zzz"], "tags": {"host": "server01"}}]}

easy reproduction:
This is the script I used to test this:

from influxdb import InfluxDBClient
import json

client = InfluxDBClient('localhost', 8086, 'root', 'root', 'example')
resp = client.query('DROP measurement cpu_value')

points = [
    {
        "measurement": "cpu_value",
        "tags": {
            "host": "server01",
            "region": "us-west",
            "type_instance": "idle",
        },
        "fields": {
            "value": 0.64,
            "zzz": 34545,
        }
    },
    {
        "measurement": "cpu_value",
        "tags": {
            "host": "server01",
            "region": "us-west",
            "type_instance": "idle",
        },
        "fields": {
            "value": 2.34,
            "zzz": 34545,
        }
    }

]

client.write_points(points)
QUERY = """SELECT host, zzz as hello, host, value as value, zzz, host FROM cpu_value"""
resp = client.query(QUERY)
print json.dumps(resp.raw)

See the influxdb-python issue: influxdata/influxdb-python#228

[otoolep]

@corylanou

Can you be more precise? What in the output don't you like?

[aviau]

Not sure if you noticed it, but the issue here is that host is a tag. So it should not be included in the columns list.

Not that i'm against it being included. But if it's included, include it in the values too!

[corylanou]

I just fixed a bug with the alias (which is shown in your last query - #3564). I'm also making changes right now to the select behavior that if you put a tag in the select clause, it comes back as a column, not as another grouped set of output (#3549).

This may address your concerns, I'm not sure.

[aviau]

if you put a tag in the select clause, it comes back as a column

Great! Ping me here when this is merged and I will test it. I think that will fix the issue.

[beckettsean]

@aviau the code @corylanou referenced has been checked in and is in the nightly build. Can you try to repro with the latest?

[aviau]

With the same test, this is the new result:

{"series": [{"values": [["2015-08-19T14:18:34.794532536Z", "server01", 34545, "server01", 0.64, 34545, "server01"]], "name": "cpu_value", "columns": ["time", "host", "hello", "host", "value", "zzz", "host"]}]}

Thanks!