I created a continuous query to downsample readings from temperature sensors in my influxdb to store hourly means for a longer time. There are readings of multiple sensors in one table. Upon executing the query, the sensors ip is missing.
Basic data looks like this:
> SELECT ip,tC FROM ht LIMIT 5
name: ht
time ip tC
---- -- --
1671057540000000000 192.168.0.83 21
1671057570000000000 192.168.0.83 21
1671057750000000000 192.168.0.17 21.38
The continuous query (simplified without CREATE ... END):
SELECT last(ip), mean("tC") AS "mean_temp" INTO "downsampled"."ht_downsampled" FROM "ht" GROUP BY time(1h),ip
The issue is, the value of 'ip' is only a tag, not the value in the table and subsequently is missing in the table the query inserts into:
name: ht
tags: ip=192.168.0.17
time ip mean_temp mean_hum
---- -- --------- --------
1671055200000000000 21.47 42.75
1671058800000000000 21.39428571428571 48.785714285714285
1671062400000000000 21.314999999999998 51.625
Why is last(ip) not producing any value?
Can I get the value from the 'tags' into the table?
Is there a different approach to group data with a constant value?
Could you just try query the ip instead of the last(ip) since you are grouping by the ip in the statement already?
Sample code:
SELECT ip, mean("tC") AS "mean_temp" INTO "downsampled"."ht_downsampled" FROM "ht" GROUP BY time(1h), ip
Related
Is it possible to return just a value, and not a time from influxdb?
I know influxdb is a Time Series Database, so it includes the time regardless, but the tool I'm using requires the first field displayed from the query to be a value, and not a time.
I've also tried switching the order of the arguments and time is still the first argument returned in the query.
select value,time FROM smart_threshold WHERE host =~ /proxmox/ AND instance = 'sda' AND type_instance='reallocated-sector-count' limit 1
Result:
name: smart_threshold
time value
---- -----
1512183206033040777 36
Would it be possible to return:
name: smart_threshold
value time
------ ----
36 1512183206033040777
or just:
name: smart_threshold
value
------
36
instead?
As of now, it seems there's no other way than do simple man-in-the-middle script that would swap two values in the output (in case you don't own "the tool").
What the tool is that, though?
select SUM(value)
from /measurment1|measurment2/
where time > now() - 60m and host = 'hostname' limit 2;
Name: measurment1
time sum
---- ---
1505749307008583382 4680247
name: measurment2
time sum
---- ---
1505749307008583382 3004489
But is it possible to get value of SUM(measurment1+measurment2) , so that I see only o/p .
Not possible in influx query language. It does not support functions across measurements.
If this is something you require, you may be interested in layering another API on top of influx that do this, like Graphite via Influxgraph.
For the above, something like this.
/etc/graphite-api.yaml:
finders:
- influxgraph.InfluxDBFinder
influxdb:
db: <your database>
templates:
# Produces metric paths like 'measurement1.hostname.value'
- measurement.host.field*
Start the graphite-api/influxgraph webapp.
A query /render?from=-60min&target=sum(*.hostname.value) then produces the sum of value on tag host='hostname' for all measurements.
{measurement1,measurement2}.hostname.value can be used instead to limit it to specific measurements.
NB - Performance wise (of influx), best to have multiple values in the same measurement rather than same value field name in multiple measurements.
I'm trying to leverage my moderate SQL-knowledge for InfluxQL, but I'm missing something(s) about the nature of timeseries db.
Use case
I write a measurements from our issue tracker, when an issue is updated:
issue_updated,project=facebook,ticket=fb1,assignee=coolman status="todo"
Problem
Given this returns rows of issues statuses:
SELECT status
FROM "issue_updated"
If this was SQL (fiddle) I would use COUNT(and then add the WHERE time > NOW() - 1Y GROUP BY time(5m)). However the following gives me Mixing aggregate and non-aggregate queries is not supported
SELECT status, count(status) as 'Count'
FROM "issue_updated"
Can someone give some guidance here? ta
Sounds like what you're looking for is the ability to group by a field value which isn't currently supported.
From what I can tell, if you modify your schema a bit, it should be possible to do what you're looking. Instead of
issue_updated,project=facebook,ticket=fb1,assignee=coolman status="todo"
Do
issue_updated,project=facebook,ticket=fb1,assignee=coolman,status=todo value=1
then
SELECT count(value) FROM "issue_updated" WHERE time > now() - 52w GROUP BY status
name: issue_updated
tags: status=other
time count
---- -----
1449523659065350722 1
name: issue_updated
tags: status=todo
time count
---- -----
1449523659065350722 2
should work.
I have a measurement that has the following fields in InfluxDB. The measurement name is data.
Measurement name: data
Time Device Interface Metric Value
2016-10-11T19:00:00Z device1_name int_name In_bits 10
2016-10-11T19:00:00Z device2_name int_name Out_bits 5
.
.
.
I have created a continuous query as follows:
CREATE CONTINUOUS QUERY "test_query" ON "db_name" BEGIN SELECT sum("value") as Sumin INTO "data.copy" FROM "data" where metric = 'In_bits' GROUP BY time(15m), device END
After creating this, how do I see the results saved in new measurement data.copy? After creating this query, I am not able to find the new measurement that has been created. If I'm doing something wrong, I would like to get more input on this matter. Thanks!
I have started playing around with Influxdb v0.13 and I have some dummy values in my test db where id is a tag and value is a field:
> SELECT * FROM dummy
name: dummy
--------------
time id value
1468276508161069051 1234 12345
1468276539152853428 613 352
1468276543470535110 613 4899
1468276553853436191 1234 12
I get no results returned when I run this query:
> SELECT * FROM dummy WHERE id=1234
but I do get the following when querying with the field instead:
> SELECT * FROM dummy WHERE value=12
name: dummy
--------------
time id value
1468276553853436191 1234 12
Am I doing something wrong here? I thought the point of tags were to be queried (since they are indexed and fields are not), but they seem to break my queries.
It appears that Influx will treat every tag key and value we insert as string and this is evidently shown in their official documentation.
See: https://docs.influxdata.com/influxdb/v0.13/guides/writing_data/
When writing points, you must specify an existing database in the db
query parameter. See the HTTP section on the Write Syntax page for a
complete list of the available query parameters. The body of the POST
- we call this the Line Protocol - contains the time-series data that you wish to store. They consist of a measurement, tags, fields, and a
timestamp. InfluxDB requires a measurement name. Strictly speaking,
tags are optional but most series include tags to differentiate data
sources and to make querying both easy and efficient. Both tag keys
and tag values are strings.
Note: the text in bold.
Hence to filter by tag key value - the query must be enquoted.
Example:
SELECT * FROM dummy WHERE id='1234'