The round-robin strategy is typically employed in the load-balancing configuration to distribute requests. The load balancer does not determine the specific node to which a request belongs. Instead, it forwards requests to each node based on the round-robin strategy. If a node receives a request that does not belong to it, it forwards the request to the corresponding node for processing and returns the result to the requester after processing.
However, if a node becomes abnormal and cannot respond to internally forwarded requests, the load balancer may misjudge the status of normal nodes. This can result in the client waiting a long time for a response, but no error is reported on the front end.
To prevent a single node failure from rendering the entire cluster unavailable, a health check function for inter-node request forwarding is added. This function applies limitations to abnormal nodes and provides frontend alerts.
Proper configuration of the following parameters can effectively enhance system stability and reliability, ensuring normal cluster operation in case of single-node failures.
Parameter Description: When a request forwarded between nodes does not receive a response within the set timeout period, the frontend will report an error and terminate the request.
Default Value: 90 (unit: second)
Modification Rule:
Enter an integer not less than 10.
If certain backend operations, such as report calculations or exports, require a longer processing time, the timeout should be extended appropriately.
The values of the proxy_read_timeout and proxy_send_timeout parameters in Nginx must be greater than the timeout configured on the platform.
Parameter Description: It refers to the maximum number of system attempts to forward a request within a single cycle.
Default Value: 8
Enter a positive integer.
The value of the max_fails parameter in Nginx is advised to be greater than or equal to that of Attempt Quantity configured on the platform.
Parameter Description: Within the set failure timeout period, if the number of failed requests exceeds the maximum attempt count, the node will be marked as failed, and new requests will no longer be forwarded to it.
The value of the fail_timeout parameter in Nginx is advised to be greater than or equal to that of Failure Duration configured on the platform.
If a node server rejects the connection, an error message will be displayed: "Error Code: 41300002. The server rejects the connection."
If the inter-node forwarding time exceeds the set timeout, the error message will be displayed: "Error Code: 41300001. The server response time is too long."
Modification via FineOps
1. Log in to FineOps as the admin and click the required O&M project. Choose Maintenance > Cluster Management.
2. Modify the parameter values under Advanced Cluster Configuration > Inter-Node Forwarding Configuration. For details about the default values and modification rules, see the "Supported Configuration Parameter" section of this document.
3. Click Save after modification. A pop-up appears, indicating that the parameter settings have been modified and you need to restart each node to apply the changes.
4. Restart each application node of the FanRuan project for the configuration to take effect. For details, see Closing or Restarting the FineBI Project.
Modification via the FanRuan Application
1. Log in to the FanRuan application as the admin, and choose System Management > Intelligent O&M > Cluster Configuration > Parameter Configuration.
2. Modify the inter-node forwarding configuration in Internal Forwarding Parameter. For details about the default values and modification rules, see the "Supported Configuration Parameter" section of this document.
If you enable a cluster, you are advised to dedicate a Redis service to a single FanRuan project as the state server.
If a single Redis service is connected to multiple FanRuan projects, the key prefixes of different projects may be identical, which may cause data conflicts.
Therefore, if a Redis service is shared by multiple FanRuan projects, configure a unique key prefix for each project to prevent such conflicts.
You can modify the key prefix using either of the following methods as needed.
2. Modify the key prefix in Advanced Cluster Configuration.
Default value: For projects not deployed via FineOps, the default key prefix of the state server is __fine_redis_key__.
Customization rule: Uppercase and lowercase English letters, Arabic numerals, and underscores are supported. Chinese and special characters are not supported.
2. Modify the key prefix in Redis Storage Key Prefix.
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