Library to convert OpenTSDB data to pandas datastructures for analysis. It also allows to name series based on metadata returned in the response.
Testing has primarily been done with the following configuration:
- RHEL/6.5, Python 2.6.6
- Ubuntu/trusty64 Python 2.7.6 ???
-
Atleast 1GB of memory
-
Preferrably more than 1 CPU/Core
atlas-devel blas-devel c++ compiler gcc gcc-gfortran libffi-devel make
pip install git+https://github.com/metrilyx/opentsdb-pandas.git
import requests
from opentsdb_pandas.response import OpenTSDBResponse
tsdb_url = "http://my.opentsdb/api/query"
resp = requests.get(tsdb_url+"?m=sum:nginx.stubstatus.request{host=*}&start=3m-ago")
#
# This can be a string, list or tuple
#
oResp = OpenTSDBResponse(resp.text)
#
# Get a DataFrame with epoch converted to pandas datetime.
#
df = oResp.DataFrame(convertTime=True)
#
# Get a DataFrame with custom column names. In this case set it to
# the short hostname.
#
df = oResp.DataFrame("!lambda x: x['tags.host'].split('.')[0]")
print df