Modeling Conundrum: How is it even possible to store Trade data in InfluxDB?

I'm trying to record all reported trades of a security, but multiple trades can happen at the same time. For example, consider this series:

Time, Market ID, Trade ID, price, volume
13:01, 1, 1, 1.00, 1.00
13:01, 1, 2, 1.01, 0.5
13:01, 1, 3, 1.00, 0.75

The unique key for each row is the compound key (Market ID, Trade ID), but if I create each of these columns as Tags, then we have a very high cardinality in a tag, breaking InfluxDB. Providing finer-grained resolution on the time field also doesn't help. For one thing, the precision in these measurements is not better than one second, and for another, it's possible for two trades to occur at exactly the same time during a multi-fill.

Is there any solution for this?

Hi, anyone know if influxdb implements correctly NOT boolean operation over complex sentences? I'm migrating my requests from Mysql and i'm trying to execute this query but fails:

SELECT time, turbina, Prod_TotAccumulated_ActPwrGen2, Prod_LatestAvg_ActPwrGen1, Prod_LatestAvg_ActPwrGen2, Sys_Logs_FirstActAlarmPar1, Sys_Logs_FirstActAlarmNo FROM datos WHERE turbina = 'WTG01' AND NOT (Sys_Logs_FirstActAlarmNo = -1.0 AND Sys_Logs_FirstActAlarmNo != NULL) ORDER BY time DESC LIMIT 1

It fails on interpreting NOT( ):

"error": "error parsing query: found (, expected ; at line 1, char 205"

I could modify this particular query with some logical equivalence, but I would like to know in advance if I should rewrite all my query logics on now on.

Any help?

Post has attachment
hello everyone,
When I was in use influxdb ,sometimes I want to remove some server from Influxdb, but the Chronograf page "host list" will not be deleted.
I thought kapacitor will give notice of a lot of mail, but did not, why have delete machine won't send E-mail warning?
And what should I do to get deleted the host from the "host list"?

I have a series to capture the state of my heating which is 0 (no heating) or 1 (heating on)
I want now to calculate how long my heating was on on a given period.

select elapsed(*,1s) from heating where Room='Living' does not give me the key value anymore, so I don't know what the elapsed time means

select elapsed(*,1s) from heating where Room='Living' where value=1 gives me wrong elapsed time as it does count time between the periods where heating was on.

Any idea how I can calculate my heating time during a given period (eg day) ?

Note I am using influx 1.2.4 (need to upgrade to 1.3)

Hi, want to ask question, shout if somewhere better to ask then here.

Can you query for a range of dates but a time window inside, like all of June, Monday-Fri, 9:00-17:00?

Influx looks great, but all this Time series data stuff I see seems to focus on continues data. I am trying to find something to replace the database were using, GB of data for a tens of thousands of sensors. and apparently 30 sec for a query of a months of data for a specific set of sensors is too slow.

so like the query above, only interested in query for a time across each day. but that time is select-able, so just discarding data outside that time is not so good.

If any points regarding this, or a better place to ask, thank you in advance.

Post has attachment

Post has attachment

I'd like to share my open source project with the community. I've produced an analytics gathering end-point. This stores metrics in a running InfluxDB

The endpoint use strong secure tokens to protect the service so it can be
safely put on the internet.

The service is a python based project here:


However I've also produce a docker container so you can get up and running quickly:


I've used this for over a year now and thought it might be useful for others.

I'd welcome feedback and pull requests,


I'm using grafana and influxdb(latest)
On grafana graph, I see few series getting hidden. But on unhiding some series of the graph, i get the rest visible.
From some of the issues reported in github, i got to know that it is due to max_row_limit and chunk_size related.

On enabling developer tools, i find partial=true for graphs that have missing series. InfluxDB documentation refers to passing chunk_size and chunked/true or false in curl commands.

I changed max_row_limit in influxDB to zero in conf file and restarted influxDB. Still that does not make any difference.

How can i override the max-limit ( if conf file is not the only place to modify ) and get all data visible ?


I'm currently exploring InfluxDB now for setting up metrics system.

I tried calculating mean of few values to understand the working.

My points are
time distinct
---- --------
1486112400000000000 3
1486198800000000000 0
1486285200000000000 8
1486357200000000000 5
1486371600000000000 2
1486371600000000000 4
1486371600000000000 5
1486458000000000000 2

It starts only from Feb-2-2017
Setting up the date range from 1-Jan-2017 to 28-Feb-2017, with groupby time(30d), i find the mean calculated fine, but it is for the date 24-Jan-2017.
> SELECT mean("VAL") FROM "P1" WHERE time >=1483228800000 AND time >=1486598400000 GROUP BY time(15d) fill(none)
name: P1
time mean
---- ----
1485216000000000000 3.5555555555555554

> SELECT mean("VAL") FROM "P1" WHERE time >=1483228800000 AND time >=1486598400000 GROUP BY time(30d) fill(none)
name: P1
time mean
---- ----
1485216000000000000 3.5555555555555554

Basically, my requirement is to calculate mean for 30days ( for calendar days )

In this case, for Jan, I should see 0 and Feb 3.5

Why does the mean specifically point to Jan-24th ? I have no points in Jan .

What is the way to show mean of calendar months on the graph...

I'm trying to use Telegraf with Jira's REST api, but I can't figure out what "endpoint" I should use. I have a specific project in my Jira instance that I'm working with, and I want to be able to grab multiple types of data to create timeseries charts.

Is there an example of how to use Jira with Telegraf, or at least another similar RESTful app that might help to point me the right way?
Wait while more posts are being loaded