search

hyperledger / fabric-samples

Samples for Hyperledger Fabric
https://wiki.hyperledger.org/display/fabric
Apache License 2.0
2.76k stars 3.36k forks source link

Kube test network : Start prom/grafana metrics services #622

Open jkneubuh opened 2 years ago

jkneubuh commented 2 years ago

@fraVlaca and @davidkel have done some great work wiring up Prometheus / Grafana services in the test-network and bare-metal networks derived from fabric-test-operator.

Tracking time-series metrics of system behavior is great. It's virtually impossible to do any useful diagnosis and/or benchmarking on a distributed system without the ability to correlate "what happened" and "when did it happen" at a system-wide level.

Carry forward the metrics work by setting up a yaml descriptor and ingress to launch prom/grafana on the kube test network. This will allow us to poke + prod Fabric networks with a consistent lens across the different runtime platforms.

Maybe: 

./network viz

or 

./network metrics

?

SamYuan1990 commented 2 years ago

Hi @jkneubuh , @davidkel , @fraVlaca maybe you can take a look at my labs project. https://github.com/hyperledger-labs/PerformanceSandBox

in PerformanceSandBox It use Prometheus and jaeger operator to set up monitoring env for test network on k8s, so far my target for this project is on blockchain observability. and it now has

  1. Prometheus and jeager support
  2. a sample to use erc 721 code as a sample for deploy a customer chaincode on test network k8s.
  3. podman or docker
  4. github actions as CI
  5. Tape instead of caliper is ready for now for performance test/traffic generation.
  6. caliper PR open as waiting for caliper 0.5 release.

Performance SandBox also default topic for https://wiki.hyperledger.org/display/PSWG/PSWG+April+12%2C+2022 Once up on a time I had discussed at https://github.com/hyperledger/fabric-samples/discussions/551 that's the reason why I created the labs project. To avoid code duplicate, as I implemented couchdb support in PerformanceSandBox then test-network-k8s....(https://github.com/hyperledger/fabric-samples/pull/651) I am open discussion and free for any talk. I am not sure if we should merge some code and maintain in a single project. How about we keep some basic features in test-network-k8s as a sample and we have a sandbox for performance research? and meanwhile we can use PSWG meeting as a place to sync up project status? CC: @denyeart , @mbwhite