influxopmon - Influx DB plugin for Operational Monitoring

Influxopmon converts a JSON object into an influxdb insert statement, then inserts the statement into an influx database. It is part of the Operational Monitoring library (opmonlib) and used to colled the opmon data.

The latest tags are available at https://github.com/DUNE-DAQ/influxopmon/tags.

Building and running :

The library's constructor takes an URI argument. The URI's syntax is the database : Application name: influxdb://<host>:<port>/<endpoint>?db=<mydb>. Authentication is managed using a reverse proxy, meaning that if the database should be accessed using authentication, the endpoint and port should be setfor the reverse prox

The library should be used calling the library's "publish" function with as argument a nlohmann::json object. More details about the implementation are available in the "Technical information" chapter.

The library output is the return statement from the CPR message and, if successfull, the insertion of the JSON content to the TSDB.

URI example :

the influxopmon URI presents as such: influx://opmondb.cern.ch:31002/write?db=influxdb

Translating in the full, following URI eyample:

daq_application -c rest://localhost:12345 --name yourchoosenname -i influx://opmondb.cern.ch:31002/write?db=influxdb 

Step-by-step :

1. In a build environment clone the latest influxopmon tag from DUNE-DAQ
2. Verify your environment includes the opmonlib module
3. Compile your environment
4. In a runtime environment, run the call URI

Technical informations

Classes

influxOpmonService

the class inherits dunedaq::opmonlib::OpmonService

explicit influxOpmonService

The constructor takes as parameter the URI described in the "Building and running" chapter and initializes variables containing the address and database where the queries should be sent, the delimiter tag and 0..N influx db tags. The delimiter tag is the keyword in a json file delimiting and describing data entries, for a Time Series Database (TSDB), it is the time tag. A TSDB typically has 0..N data tags describing the specificities of a certain data entry, such as its position or a captor identifier.

public publish

The publish function is called on data collection by the opmonlib. It calls set_inserts_vector, get_inserts_vector, transforms the vector returned by get_inserts_vector into individual queries lines and calls the execution_command.

private execution_command

Takes as parameters the url to which the message should be sent and the command to be sent. Uses cpr::Post and awaits cpr::Response.

To debug the application, print the cpr::Response response.

JsonConverter

public set_inserts_vector/get_inserts_vector

Handles the conversion by calling json_to_influx_function and conversion errors has the following parameters:

Param 1 bool : if true, the tags are not added to the query. Param 2 vector : is a vector of key-words delimiting tags Param 3 string : is the keyword delimiting the timestamp Param 4 string : is a string formatted flattened json object

set_inserts_vector Void and sets sets insertsVector, insertsVector can be accessed using get_inserts_vector.

private json_to_influx_function

Converts a flattened json into a vector of inserts according to a delimiter passed in parameter. Calls check_data_type and convertTimeToNS

private convertTimeToNS

Influxopmon changes, if necessary, UNIX EPOCH time format to UNIX EPOCH time format in nanoseconds, to adapt to influx db the format put in parameter.

private check_data_type

Puts the correct influx parameter delimiter according to the data type of an entry, for example using ' " ' for string delimitation.

Notes :

The database is queried using CPR, which is a libcurl wrapper.

In DUNE DAQ build order, "influxopmon" should be built after "opmonlib".

Influxopmon finds the following packages: daq-cmake, opmonlib, CPR, libcurl

For any further information, contact Yann Donon (yann.donon@cern.ch).