Observability (monitoring, log collection, tracing) is a key aspect when operating highly distributed applications deployed to Kubernetes. There are quite a few frameworks/options provided by the CNCF landscape to help you that - Prometheus, FluentBit, Fluentd, Jaeger, OpenTracing, etc.
This Meetup introduces the audience to the available options and drives through battle tested use cases:
Application monitoring with Prometheus
This talk does a deep dive into Prometheus, and the layers we need to consider when monitoring an application deployed to Kubernetes. Gives a deep dive into Prometheus internals as relabels, federation, etc and showcases how we can rely on existing building blocks to automagically monitor a Kubernetes deployment.
It also does a sneak peak into Thanos - a highly available metric system with unlimited storage capacity built on Prometheus.
Sandor Guba, Banzai Cloud
The reason I like statefulsets, from Haproxy to Prometheus
- What is a statefulset?
- How can we use it with HAProxy with autoscaling backend
- How can we use it with Prometheus with auto discovery
Natko Marton, IBM Budapest Lab
Centralized log collection on Kubernetes
Collecting logs for debugging and further analysis of application behavior is a key requirement for successfully operating a distributed application. While this not straightforward, Kubernetes gives us good enough building blocks to start with. This talks introduced the audience into FluentBit, Fluentd and an Elastic stack to securely extract, move and collect logs into a centralized location.
Nandor Kracser, Banzai Cloud
Bonus track - End to end tracing
In case there is still some time left we'll do and end-to-end tracing demo using OpenCensus. OpenCensus is a library that automatically collects traces and metrics from applications, displays them locally, and sends them to an analysis/visualization tools as Jaeger, Stackdriver, Zipkin, Prometheus and more ...
If you are interested in the topics above and would like to be a bit more prepared before coming to the Meetup (and ask us hard questions) I suggest going through these posts about how we at Banzai Cloud monitor and collect logs on Kubernetes.