I can't make sense of the kapacitor record query syntax. The documentation
only covers the HTTP API, not the CLI, but the Custom Anomaly Detection
<https://docs.influxdata.com/kapacitor/v1.1/examples/anomaly_detection/>
example
gets one started and the response to kapacitor record query without
arguments prints some guidelines.
Everything suggest that the query is a standard line protocol string. But
there are some confusing differences. Observe:
$ influx
> CREATE DATABASE record
> USE record
> INSERT points a=1
> INSERT points b=2
> SELECT * FROM points
name: points
time a b
---- - -
1483716972843739426 1
1483716980505185982 2
> exit
$ cat record_issue.tick
stream
|from()
.measurement('points')
|log()
.prefix('POINTS_ISSUE')
$ kapacitor define record_issue -type stream -tick record_issue.tick -dbrp
record.autogen
$ rid=$(kapacitor record query -query $'SELECT * FROM
"record"."autogen"."points"' -type stream)
$ kapacitor replay -task record_issue -recording $rid -rec-time
At this point the log file shows:
[record_issue:log2] 2017/01/07 02:40:46 I! POINTS_ISSUE
{"Name":"points","Database":"record","RetentionPolicy":"autogen","Group":"","Dimensions":{"ByName":false,"TagNames":null},"Tags":{},"Fields":{"a":1},"Time":"2017-01-06T15:36:12.843739426Z"}
which indicates that only the points with an 'a' field have been recorded.
Executing the same query at the CLI returns both points.
Other differences:
$ rid=$(kapacitor record query -query $'SELECT a,b FROM
"record"."autogen"."points"' -type stream)
still only returns 'a' field, not 'b' field, when replayed.
$ influx
> USE record
> INSERT points,t1=3 a=3,b=3
> INSERT points,t2=4 a=4,b=4
> SELECT * FROM points
name: points
time a b t1 t2
---- - - -- --
1483716972843739426 1
1483716980505185982 2
1483717340694062989 3 3 3
1483717349891893752 4 4 4
> exit
$ rid=$(kapacitor record query -query $'SELECT a,b FROM
"record"."autogen"."points"' -type stream)
only returns 'a', no tags.
$rid=$(kapacitor record query -query $'SELECT * FROM
"record"."autogen"."points" GROUP BY *' -type stream)
Returns all the tags this time, but not the points without a in it (ie. the
1483716980505185982 point).
Three questions:
1. Why does each field key need to be named in the SELECT statement,
unlike CLI queries which return all the fields listed in the SELECT
statement.
2. How do you return all fields (in order - multiple statements is not
suitable) from a measurement given the * doesn't work.
3. Why is GROUP BY required to return the tags when it's unnecessary for
the CLI.
By the way, the record venture is invaluable. Being able to tweak kapacitor
scripts, and test them on history data before deploring to practice is
crucial.
Regards,
Heath
--
Remember to include the version number!
---
You received this message because you are subscribed to the Google Groups
"InfluxData" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/influxdb.
To view this discussion on the web visit
https://groups.google.com/d/msgid/influxdb/cd1c824d-7a25-4464-b325-d0075b5ebb50%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
