You can use Network Analysis to quickly locate network issues during project deployment, scaling, and troubleshooting, and download the evidence package for detailed analysis.
This function is intended for the following users:
1. Implementation and delivery engineers who need to confirm whether ports are open and links are reachable after deployment or scaling.
2. O&M admins who need to perform daily inspections and periodically monitor changes in network connectivity.
3. Technical support engineers who need to perform preliminary diagnosis of on-site issues and collect standardized troubleshooting evidence.
1. The platform cannot be accessed after deployment, or the project fails to start.
2. Access requests return 502 or 504 errors, and you need to check whether inter-node connectivity is unavailable or the container port is blocked.
3. External dependencies are unreachable, such as cloud domains, the template market, the plugin market, and OAuth domains.
4. You need to troubleshoot cross-layer issues, such as DNS resolution, port occupation, firewall and iptables rules, and container network exceptions.
5. Evidence collection is required for complex issues, such as conflicts between physical server network segments and container network segments, and incorrect container network modes.
Before executing network analysis, confirm that all conditions in the following table have been met. Otherwise, the results may be incomplete.
The OPS Agent component has been installed on all project nodes and is in the running status.
Information collection depends on this component. Nodes with an offline component only generate SYS-01 check results, and subsequent IP address and port probing is not performed.
1. Log in to FineOps and click Project Management on the left.
2. Select the target project to go to the project details page.
3. Choose Health Inspection > Network Analysis to go to the Network Analysis page. The top of the page displays the collection and check scope, and Analysis Setting is displayed in the upper-right corner.
The following table describes the collection and check scope.
This section describes how to execute a full check manually and obtain conclusions and an evidence package.
1. Click the Network Analysis button. FineOps starts the check.
2. Wait until the check ends. The check duration is related to the number of nodes and ports. A check for a single project usually takes 2 to 15 s.
3. View the results in Check Result, including the number of severe items, warning items, passed items, and the check duration.
4. Check abnormal items one by one in the result details table. The following table describes the columns in the details table.
5. To troubleshoot further, click Download Detailed Data Package in the lower-right corner of the result details table to export the diagnostic evidence package.
The package contains the connectivity matrix, port listening and mapping information, DNS resolution results, system and container events in the last seven days, and summaries of iptables and firewalld.
This section describes how to enable periodic checks, which are suitable for continuous observation of intermittent network issues.
1. Click Analysis Setting in the upper-right corner of the page.
2. Enable Automatic Analysis, which is disabled by default.
3. Enter a value in Execution Interval and select Minute(s) or Hour(s) as the unit. The default value is 24 hours. The current setting is displayed at the bottom of the panel, for example, "Current Setting: Run once every 24 Hour(s)."
4. Click Save for the setting to take effect.
This section describes how to trace historical check results and compare network status changes.
1. View historical check records in History for the Last 7 Days. The fields include the report type (manual or automatic), completion time, number of warning checks and passed checks, duration, and status.
2. Click View Result to review details.
3. Click Download Detailed Data Package to download the evidence package of the analysis.
The following table describes the check content and common causes of abnormal results.
When check results contain multiple alerts, interpret them in the following order to narrow down the issues quickly.
1. Check SYS-01 first to confirm whether all agent instances are running. If offline nodes exist, restore the agent first, and then run the analysis again.
2. Check APP-01 to confirm whether the application entry can be accessed and determine whether the failure occurs at the TCP, TLS, or HTTP layer.
3. When a port is unreachable, check NET-03, NET-02, and NET-04 in sequence. Check the host-published port first, then the internal container port, and finally the inter-node dependency port. Use the result to confirm whether the issue is in port mapping, inside the container, or in inter-node connectivity.
4. When external dependencies, such as online upgrades, the template market, and the plugin market, are abnormal, check the domain connectivity conclusion of NET-05.
5. If all check items pass but the issue persists, download the detailed data package. Collect system logs such as messages and journalctl if necessary.
Does a warning mean the environment is faulty?
No. A warning means that the result is abnormal or evidence is incomplete. For example, in an internal-only network, a warning for NET-05 is expected when domain names are unreachable. A successful auto-repair event is also shown as a warning for SYS-03. Check the Suggestion column and the actual deployment mode before taking action.
Why does the number of check results change?
Results are expanded by node, URL, port, kernel parameter, and repair event. The record count changes when the number of nodes, mapped ports, or repair events changes.
Does the diagnostic data package keep occupying disk space?
No. Exported packages are retained for seven days, with a maximum of 200 packages. Older packages are deleted automatically.
Should Automatic Analysis stay enabled?
No. Keep Automatic Analysis disabled by default. Enable it only for monitoring intermittent network issues, adjust the execution interval as needed, and disable it after the issue is located.
What if all check items pass but users still cannot access the project?
Download the detailed data package and confirm that all server-side layers are normal. Then check the client, office network, external load balancer, WAF, VPN, and cloud bandwidth. These issues require network administrator involvement because FineOps cannot collect client-side or external-network evidence.
滑鼠選中內容，快速回饋問題
滑鼠選中存在疑惑的內容，即可快速回饋問題，我們將會跟進處理。
不再提示
10s後關閉
Submitted successfully
Network busy