Backends¶
Kaneda provides builtin backends to store metrics and events in a persistent storage. If you want to use your
custom backend you need to subclass BaseBackend
and implement your custom report
method which
is the responsible to store the metrics data.
Elasticsearch¶
Elasticsearch is a search based NoSQL database that works very well with metrics data. It provides powerful tools to analyze data and build real-time dashboards easily with Kibana.
Note
Before using Elasticesearch as backend you need to install Elasticsearch Python client:
pip install elasticsearch
-
class
kaneda.backends.
ElasticsearchBackend
(index_name, app_name, client=None, connection_url=None, host=None, port=None, user=None, password=None, timeout=0.3)[source]¶ Elasticsearch backend.
Parameters: - index_name – name of the Elasticsearch index used to store metrics data. Default name format will be index_name-YYYY.MM.DD.
- app_name – name of the app/project where metrics are used.
- client – client instance of Elasticsearch class.
- connection_url – Elasticsearch connection url (https://user:secret@localhost:9200). It can be used passing a single connection_url (a string) or passing multiple connection_urls (a list).
- host – server host. It can be used passing a single host (a string) or passing multiple hosts (a list).
- port – server port.
- user – HTTP auth username.
- password – HTTP auth password.
- timeout – Elasticsearch connection timeout (seconds).
MongoDB¶
MongoDB is a document oriented NoSQL database. Is a great tool to store metrics as it provides a powerful aggregation framework to perform data analysis.
Note
Before using MongoDB as backend you need to install MongoDB Python client:
pip install pymongo
-
class
kaneda.backends.
MongoBackend
(db_name, collection_name, client=None, connection_url=None, host=None, port=None, timeout=300)[source]¶ MongoDB backend.
Parameters: - db_name – name of the MongoDB database.
- collection_name – name of the MongoDB collection used to store metric data.
- client – client instance of MongoClient class.
- connection_url – Mongo connection url (mongodb://localhost:27017/).
- host – server host.
- port – server port.
- timeout – MongoDB connection timeout (milliseconds).
RethinkDB¶
RethinkDB is an open source scalable, distributed NoSQL database built for realtime applications.
Note
Before using RethinkDB as backend you need to install RethinkDB Python client:
pip install rethinkdb
-
class
kaneda.backends.
RethinkBackend
(db, table_name=None, connection=None, host=None, port=None, user=None, password=None, timeout=0.3)[source]¶ RethinkDB backend.
Parameters: - db – name of the RethinkDB database.
- table_name – name of the RethinkDB table. If this is not provided, it will be used the name of the metric.
- host – server host.
- port – server port.
- user – auth username.
- password – auth password.
- timeout – RethinkDB connection timeout (seconds).
Logger¶
You can use a logger instance of the logging library from the Python standard lib. Useful for debugging.
InfluxDB¶
InfluxDB is an open source time series database with no external dependencies. It’s useful for recording metrics, events, and performing analytics.
Note
Before using InfluxDB as backend you need to install InfluxDB Python client:
pip install influxdb
Warning
InfluxDB can store other type of data besides time series. However it has some restrictions:
Metrics tags field can’t be a
list
only adict
:# bad metrics.timing('user.profile_load_time', 230, tags=['login', 'edit_profile']) # good metrics.timing('user.profile_load_time', 230, tags={'from': 'login', 'to': 'edit_profile'})
Custom
metric value field can’t be alist
nor a nesteddict
:# bad metrics.custom('zone.search', metric='query_time', value={'times': [120, 230]}) metrics.custom('zone.search', metric='query_time', value={'times': {'start': 120}, {'end': 230}}) # good metrics.custom('zone.search', metric='query_time', value={'start_time': 120, 'end_time': 230})
-
class
kaneda.backends.
InfluxBackend
(database, client=None, connection_url=None, host=None, port=None, username=None, password=None, timeout=0.3)[source]¶ InfluxDB backend.
Parameters: - database – name of the InfluxDB database.
- client – client instance of InfluxDBClient class.
- connection_url – InfluxDB connection url (influxdb://username:password@localhost:8086/databasename).
- host – server host.
- port – server port.
- username – auth username.
- password – auth password.
- timeout – InfluxDB connection timeout (seconds).