SkyWalking 9.x showcase
This showcase would follow the latest changes of SkyWalking 9.x, even before the official release.
This showcase repository includes an example music application and other manifests to demonstrate the main features of SkyWalking. The music application is composed of several microservices that are written in different programming languages. Here is the architecture:
%% please read this doc in our official website, otherwise the graph is not correctly rendered. graph LR; loadgen[load generator] --> ui("UI (React)") --> app("app server (NodeJS)") --> gateway("gateway (Spring)"); gateway --> songs("songs (Spring)") & rcmd("recommendations (Python)"); rcmd --> songs; songs --> db("database (H2)");
Please run the showcase in a brand new test cluster, otherwise the undeploy process may delete some resources that you have installed before running this showcase (for example cert-manager). If you don’t do this in a new test cluster, it’s all on your own risks!
The showcase uses GNU Make and Docker containers to run commands, so please make
sure you have
make installed and Docker daemon running.
To deploy the full features of this showcase application, you may need up to 4 CPU cores and 4 GB memory, please increase the Docker daemon resources or Kubernetes cluster resources if you find containers / Pods failed to start up. Alternatively, you can also only deploy part of the features that interest you if you don’t want to increase the resources, via the guide in Customization.
Make sure you have a running Kubernetes cluster and
kubectl can access to that cluster.
git clone https://github.com/apache/skywalking-showcase.git cd skywalking-showcase make deploy.kubernetes
This will install SkyWalking components, including OAP in cluster mode with 2 nodes, SkyWalking UI, microservices with SkyWalking agent, microservices without SkyWalking agent but managed by Istio, 2 Pods to mimic virtual machines and export metrics to SkyWalking, and enable kubernetes cluster monitoring as well as SkyWalking self observability.
For more advanced deployments, check Customization documentation below.
Notice, when run this showcase locally such as KinD, the images are downloaded inside the KinD, which could take over 10 mins(depend on local network).
make deploy.kubernetes if some timeout errors break the process.
The variables defined in
Makefile.in can be overridden to customize the showcase, by specifying an
environment variable with the same name, e.g.:
export ES_VERSION=7.14.0 make <target>
or directly specifying in the
make command, e.g.:
make <target> ES_VERSION=7.14.0.
make help to get more information.
The showcase is composed of a set of scenarios with feature flags, you can deploy some of them that interest you by
FEATURE_FLAGS variable defined in
Makefile.in, as documented
in Customization, e.g.:
make deploy.kubernetes FEATURE_FLAGS=single-node,agent
Currently, the features supported are:
||Use the java agent injector to inject the Skywalking Java agent and deploy microservices with other SkyWalking agent enabled.||The microservices include agents for Java, NodeJS server, browser, Python.|
||Deploy microservices with SkyWalking agent pre-installed.||In Kubernetes sceanarios, please use
||Deploy SkyWalking OAP in cluster mode, with 2 nodes, and SkyWalking UI, ElasticSearch as storage.||Only one of
||Deploy only one single node of SkyWalking OAP, and SkyWalking UI, ElasticSearch as storage.||Only one of
||Enable SkyWalking self observability.||This is enabled by default for platform Docker Compose.|
||Start 2 virtual machines and export their metrics to SkyWalking.||The “virtual machines” are mimicked by Docker containers or Pods.|
||Start microservices WITHOUT SkyWalking agent enabled, and configure SkyWalking to analyze the topology and metrics from their access logs.||Command
||Deploy OpenTelemetry and export Kubernetes monitoring metrics to SkyWalking for analysis and display on UI.|
||Deploy OpenTelemetry and export Istio control plane metrics to SkyWalking for analysis and display on UI.|
||Deploy tools to trigger events, and SkyWalking Kubernetes event exporter to export events into SkyWalking.|
||Deploy SkyWalking Satellite to load balance the monitoring data.|
||Deploy OpenFunction and export trace to SkyWalking.||Command ofn is required to run this feature.|
||Deploy tools to submit trace profiling tasks.||Only support deployment with SkyWalking agents installed, currently Java agent and Python agent support trace profiling.|
||Deploy SkyWalking Rover and detect the processes in the Kubernetes environment.||Only support deployment in the Kubernetes environment, docker is not supported.|
To deploy the example application in Kubernetes, please make sure that you have
kubectl command available, and it can
connect to the Kubernetes cluster successfully.
kubectl get nodes to check the connectivity before going to next step. The typical error message that indicates
kubectl cannot connect to a cluster is:
The connection to the server localhost:8080 was refused - did you specify the right host or port?
# Deploy make deploy.kubernetes # Undeploy make undeploy.kubernetes # Redeploy make redeploy.kubernetes # equivalent to make undeploy.kubernetes deploy.kubernetes
# Deploy make deploy.docker # Undeploy make undeploy.docker # Redeploy make redeploy.docker # equivalent to make undeploy.docker deploy.docker
After deploy the showcase, the business system would send monitoring traffic to the OAP node, and one agent/sidecar connect to one OAP node directly.
If the business traffic is unbalanced, it would cause the OAP node receive unbalanced monitoring data. So, you could add the Satellite component. After deploy the showcase with the satellite component, the monitoring traffic would send to the Satellite service, and satellite load balances the traffic to the OAP nodes.
%% please read this doc in our official website, otherwise the graph is not correctly rendered. graph LR; agent["business app(agent)"] --> satellite("satellite") --> oap("oap"); envoy["sidecar(envoy)"] --> satellite;