Create StreamNode with groupBy(time, some_tags) and mean/max/sum of all fields

InfluxDB 1 Kapacitor
, , , ,
1

Hi everyone,

my goal is to downsample values from 1 minute to 5 minutes and compute mean, max or sum of some fields. Given the Continuous Query

CREATE CONTINUOUS QUERY downsampling_cpumemspace_5min ON mydb RESAMPLE EVERY 5m FOR 60m
BEGIN SELECT mean(cpu) AS cpu, mean(mem) AS mem, max(space) AS space
  INTO "5min".cpumemspace FROM "1min".cpumemspace
  GROUP BY time(5m), tagA, tagC
END

I created a downsampling TICKscript using BatchNode. Notice all fields are selected and tagB is omitted. This was pretty straightforward:

batch
   |query('SELECT mean(cpu) AS cpu, mean(mem) AS mem, max(space) AS space FROM "mydb"."1min".cpumemspace')
       .period(60m)
       .every(5m)
       .groupBy(time(5m), 'tagA', 'tagC')
   |influxDBOut()
       .database('mydb')
       .retentionPolicy('5min')
       .measurement('cpumemspace')
       .precision('s')

I have now great difficulties to create a TICKscript with a StreamNode. This is the best I have so far:

stream
    |from()
        .database('mydb')
        .retentionPolicy('1min')
        .measurement('cpumemspace')
        .groupBy(time(5m), 'tagA', 'tagC')
    |window()
        .period(60m)
        .every(5m)
        .align()
    |influxDBOut()
        .database('mydb')
        .retentionPolicy('5min')
        .measurement('cpumemspace')
        .precision('s')

I ran into different problems, e.g. if I don’t specify time(5m) in .groupBy() the script generates values every minute - despite of .every(5m).

.groupBy(time(5m), 'cpu') gives me
> invalid TICKscript: invalid dimension object of type kapacitor.TimeDimension

.groupBy(time(5min), "some_tag") gives me a strange message
> invalid TICKscript: parser: unexpected identifier line 6 char 25 in "By(time(5min), “some”. expected: “)” where the tag is truncated after 4 chars. This is obviously a syntax error because I’m missing something in the docs.

I’m doing this in Chronograf. I have no direct access to log messages other than here with Editor + Logs.

My second problem is: By omitting Mean, Max or Sum of each field I managed to read from all fields. But how can I really convert the original continuous query to a Stream TICKscript including the computations?

Can somebody help me? I really appreciate any help.

2

Sorry, no help here but apparently the same problem.
Maybe we can share some ideas.

Were you able to group by time?
Best Regards