2. Python- Deploy HTTP Server and Client
Each step should be performed in a separate terminal window.
Make sure your Ubuntu environment was prepared properly as described in the Preparation section.
Configure Environment Variables for Otel and Run Python Flask Server¶
Open the first terminal window in your Linux instance and set up environment and run Python Flask server using auto-instrumentation:
Important
If you are doing this workshop as part of a group, before the next step, add your initials do the APM environment:
edit the run-server.sh script below and add your initials to the environment i.e. change:
export OTEL_RESOURCE_ATTRIBUTES=deployment.environment=apm-workshop
to
export OTEL_RESOURCE_ATTRIBUTES=deployment.environment=sjl-apm-workshop
cd ~/otelworkshop/host/python
source run-server.sh
You will see the server startup text when this is run.
Run Python Client Application¶
Open a new terminal window in your Linux instance and run the Python client to sent POST requests to the Flask server:
Run the client Python app via the splunk-py-trace command to send requests to the Flask server:
Important
If you are doing this workshop as part of a group, before the next step, add your initials do the APM environment:
edit the run-client.sh script below and add your initials to the environment i.e. change:
export OTEL_RESOURCE_ATTRIBUTES=deployment.environment=apm-workshop
to
export OTEL_RESOURCE_ATTRIBUTES=deployment.environment=sjl-apm-workshop
cd ~/otelworkshop/host/python
source run-client.sh
The python-requests.py client will make calls to the flask server with a random short sleep time.
You can stop the requests with Ctrl+C
Validate span are being sent¶
Open a new terminal window in your Linux instance to check OpenTelemetry Collector Statistics to see that spans are being sent.
lynx localhost:55679/debug/tracez
Lynx is a text browser that was installed during with the setup-tools. Enabling a web browser to access your environment will allow for a full web GUI.
APM Dashboard¶
Traces / services will now be viewable in the APM dashboard. A new service takes about 90 seconds to register for the first time, and then all data will be available in real time.
The Environment pulldown will let you see the APM map associated with your individual environment that you set with your initials if this was done earlier.
Additionally span IDs will print in the terminal where flask-server.py is running. You can use Ctrl+C to stop the requests and server any time.
The Python server application will be called: py-otel-flask-server and the client will be called py-otel-client.
Navigate to Splunk Overvability -> APM
Service map of this python demo
Click on one of the peaks in the grey graph within "Services By Latency (P90)" on the right hand side and then click the trace to see spans. Also try out Tag Spotlight to see how application operations are broken down in a granular way. You can also try the Tags menu on top to search for a single trace or group of traces by key:value.
To learn more about traces and spans see the Splunk APM documentation
Where is the OpenTelemetry Instrumentation?¶
The run-server.sh and run-client.sh scripts set up the environment variables for OpenTelemetry and invoke the Python auto instrumentation:
spluk-py-trace is the auto instrumenting function that runs Python3 with the instrumentation that automatically emits spans from the Python app. No code changes are necessary. Splunk Observability Cloud has a Data Setup Wizard to guide through instrumentation setup.
OpenTelemetry repo for Python is here.
Important
Leave the Flask server running you'll need need this process for the next client examples in the workshop.