Add observability to your django application - PyCon FR 2019
0 likes568 views
Discover how to add very simply observabilty to your Django application with django-prometheus, Prometheus and Glouton.
Ad
More Related Content
Similar to Add observability to your django application - PyCon FR 2019 (20)
Recently uploaded (20)
Ad
Add observability to your django application - PyCon FR 2019
- 1. Add Observability to your Django application Pierre Fersing & Lionel Porcheron
- 2. Who we are? Pierre Fersing, CTO & co-founder Bleemeo •Python Dev for +15 years •DevOps for +10 years •Toulouse DevOps Meetup Organizer Lionel Porcheron, CEO & co-founder Bleemeo •DevOps for +15 years (started my monitoring journey with nagios netsaint) •Toulouse DevOps Meetup Leader, Capitole du Libre Leader, PyconFR 2017 Organizer
- 3. Bleemeo? Observability& Monitoringas a service solution Prometheus, Graphite, Statsd, compatible 2 Open Source projects: ◦ Glouton, universalmonitoring agent written in Go with Prometheus, Statsd, Graphite,Nagios, Zabbixcompatibility ◦ SquirrelDB, a scalablePrometheus compatiblestorage backend based on Cassandra
- 4. Old Monitoring Days •Monitoryourserver as a blackbox •Only monitor server & services (web server, database) •Only availability,not metrics •Nagios& derivates
- 5. Microservices and modern era •Increase architecture complexity •Increase number of technical components to monitor •Moderns infrastructure base on containers are dynamic •Some components may come from third parties
- 6. •Graphite change the way we were doing monitoring: metrics became central •Graphite appeared in 2008 •Prometheus became de-facto standard for monitoring •Prometheus was "invented" in 2012 at Soundcloud and is now a (graduated) CNCF project •Ecosystem based on Prometheus: exporters, Grafana, software themselves (Kubernetes, Traefik & many others) Graphite... and Prometheus
- 7. s/monitoring/observability/ •No more monitoring as blackbox: we now know what is inside •Exports tons of metrics for future usage •Code need to be instrumented to provide business metrics •New Buzzzword 😎
- 8. Three pillars of observability
- 9. Observability Key Metrics The RED Method •(Request) Rate - the number of requests, per second, you services are serving. •(Request) Errors - the number of failed requests per second. •(Request) Duration - distributions of the amount of time each request takes. The USE Method •(Ressource) Utilization: as a percent over a time interval. eg, "one disk is running at 90% utilization". •(Ressource) Saturation: as a queue length. eg, "the CPUs have an average run queue length of four". •(Ressource) Errors: scalar counts. eg, "this network interface has had fifty late collisions".
- 10. Prometheus exporters •Prometheus exporters export on web page metrics (basic plain text page with a metric per line) •Prometheus poll regularly those endpoints •Prometheus exporters exist for almost everything: https://prometheus.io/docs/instrumenting/exporters/
- 11. Instrument your code •For Django application: django-prometheus •Django Middleware for metrics •Health checks and custom checks with django-health-checks
- 12. Logs management •ELK or Loki (log managementproject from Grafana Labs)
- 13. Tracing •Jaeger or Zipkin •Useful with microservices
- 14. Example Small Django application showing quote of the day Code is available on Bleemeo Labs github: https://github.com/bleemeolabs/quote
- 15. Questions? 👉 WANT TO TRY BLEEMEO? GET 30€ OF CREDITS WITH PYCONFR2019 VOUCHER