A common WSGI stack
Project description
Talisker is a runtime for your wsgi app that aims to provide a common platform for your python services.
tl;dr
Simply run your wsgi app with talisker as if it was gunicorn.:
talisker app:wsgi -c config.py …
Talisker will wrap your app in a some simple WSGI middleware, and configure logging to output structured logging like so:
logger = logging.getLogger('app')
logger.info('something happened', extra={'context': 'I haz it'})
will output:
2016-01-13 10:24:07.357Z INFO app "something happened" svc.context="I haz it" request_id=...
It also exposes some status endpoints you can use, go to the /_status/ url on your app to see them.
This all works out of the box by using the talisker runner instead of gunicorns, and there are many more features you can use too.
Elevator Pitch
Talisker is based on a number of standard python tools:
stdlib logging for logs
gunicorn for a wsgi runner
requests for http requests
statsd for metrics
raven for errors
werkzeug for thread locals and wsgi utilities
It is designed sepcifically to be used in both development and production, and aims to provide a default set of features out of the box:
standard log format, including ISO/UTC timestamps
structured logging with python stdlib
improved gunicorn access logs, with ms precision UTC timestamps
request id tracing
standard set of status endpoints for your app
easier statsd endpoint configuration
sentry/raven middleware (TODO)
All the above are available by just using the talisker entry point script, rather than gunicorn.
In addition, with a small amount of effort, your app can benefit from additional features:
simple deeper nagios checks - just implement a _status/check url in your app
per-thread requests connection pool managment (WIP)
automatic statsd metrics for outgoing HTTP requests (WIP)
more efficient statsd client management (WIP)
Additionally, talisker provides additional tools for integrating with your infrastructure:
grok filters for log parsing (WIP)
rsyslog templates and config for log shipping (TODO)
Talisker is opinionated, and derived directly from the authors’ needs and as such not currently very configurable. However, PR’s are very welcome!
For more information, see The Documentation, which should be found at:
0.6.1 (2016-09-12)
actually do remove old DEBUGLOG backups, as backupCount=0 does not remove any
0.6.0 (2016-09-09)
Propagate gunicorn.error log, and remove its default handler.
This allows consistant logging, making the choice in all cases that your gunicorn logs go to the same stream as your other application loglogging, making the choice in all cases that your gunicorn logs go to the same stream as your other application logs.
We issue a warning if the user tries to configure errorlog manually, as it won’t work as expected.
0.5.7 (2016-09-02)
Update publishing workflow
Add make changelog target
0.5.6 (2016-09-02)
more testing release process in prepartion for 0.6
0.5.5 (2016-09-02)
testing release process in prepartion for 0.6
0.5.4 (2016-08-10)
series of point release to fix various small bugs
0.5.0 (2016-08-10)
add grok filters for logstash
slight adjustment to logfmt serialisation: talisker now strips “ from tag values. This is due to a limitation in logstash.
0.4.1 (2016-08-05)
publish separate py2/py3 wheels, due to dependency differences
some doc changes
0.4.0 (2016-08-05)
First public release an PyPI.
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
