Check and Enable First Observation
This document was translated by ChatGPT
#1. Introduction
This chapter will guide you on how to check the running status and enable the first observation after completing the deployment of DeepFlow Agent.
#2. Check
#2.1 Check Collector Running Status
Check the list of collectors on the DeepFlow web page to confirm that the DeepFlow Agent is running.
Access path: System
- Collectors
- List
, refer to the steps in the image below:

Check Collector Running Status
After checking, confirm that the collector is running and there are no abnormal messages before proceeding to the next step.
#2.2 Check Application Observability Data
Check the application call logs on the DeepFlow web page to confirm that the data collection function of the DeepFlow Agent has started running.
Access path: Tracing
- Call Logs
, refer to the steps in the image below:

Check Application Observability Data
After checking, confirm that the application call logs of the collector have been normally collected and displayed before proceeding to the next step.
#3. Start Your First Observation Journey
If you want to understand the status of your cloud resources, you can go to Resources
- Resource Pool
- Computing Resources
/Network Resources
, guide link.
If you want to understand the status of your container cluster resources and applications, you can go to Resources
- Resource Pool
- Container Resources
- Container Cluster
/Namespace
/Container Node
/Ingress
/Container Service
/Workload
/Container POD
, guide link.
If you want to observe the application topology of a specific cloud server or container service, you can go to Tracing
- Topology Analysis
, guide link.
If you want to observe the application performance of a specific cloud server or container service, you can go to Tracing
- Resource Analysis
, guide link.
If you want to perform distributed tracing for a specific business request, you can go to Tracing
- Distributed Tracing
, guide link.
If you want to continuously profile the performance of a specific application process, you can go to Profiling
- Continuous Profiling
, guide link.
If you want to observe the network traffic topology of a specific cloud server or container service, you can go to Network
- Topology Analysis
, guide link.
If you want to observe the network performance of a specific cloud server or container service, you can go to Network
- Resource Analysis
, guide link.
In addition, there are more observability usage methods waiting for you to explore...