Don't get all values in DB by using python

1

Hi,

I’m new to InfluxDB and maybe don’t understand the special behaviour of time series data bases.

I do the following with python:

jsondata =
for i in range(1000):
jsondata.append(
{
“measurement”: “NKK”,
“tags”: {
“position”: “51”,
},
“time”: datetime.now().strftime(‘%Y-%m-%dT%H:%M:%S.%fZ’),
“fields”: {
“value”: math.sin(time.time()/10.)+random.uniform(-0.1,0.1)
}
}
)

client.write_points(jsondata,time_precision=‘u’,batch_size=100)

results = client.query(‘SELECT “value” FROM “nkktable”.“autogen”.“NKK” WHERE time > now() - 1d GROUP BY “position”’)

points1 = results.get_points(tags={‘position’:‘51’})
for i,point in enumerate(points1):
print(“%5i %s %f” % (i,point[‘time’], point[‘value’]))

what I get is not 1000 points but only about 50. If I select directly from database there are also only 50 values. So it seems that not all data have been stored.

What I’m doing wrong?

Best regards

Michael

2

I’m not an expert of python, but for what I understand you are using:
Static Measurement - That’s ok
Static Tags - This is also ok
Dynamic Time - this might be the problem
Random Fields - that’s ok

To keep it simple, if you write a point with the same Tags and Time in a Measurement you will update it’s previous value and not write a new point.
Have a look at your data before you insert them. if they share the same time (rounded to the defined precision) you are updating points over and over again.

3

Yes, this was the problem. Python was too fast at this point.

I now generate the time values by hand (it’s only a first test and nothing for practice) and add always a microsecond. Now all data are in. In real systems there will be a millisecond difference and if not the value can be neglected. So all is ok.

Thank you very much.