Fault isolation provides fine-grained, instance-level traffic isolation when a service instance fails. This reduces the impact of the failure and improves service availability.
Add a fault isolation rule
Log on to the Service Mesh console.
In the navigation pane on the left, click Service Administration, and then click the Fault Isolation tab.
Click Add Isolation Group Rule and configure the following parameters:
Parameter
Description
Rule Name
The name of the fault isolation rule.
The name can be up to 255 characters long.
Application Name
Enter or select the client application.
Isolation Direction
The direction of fault isolation. Valid values:
Server-side Isolation: Detects and temporarily isolates faulty servers from the server-side.
Client-side Isolation: Detects and temporarily isolates faulty servers from the client-side.
Run Mode
The mode in which the fault isolation rule runs. Valid values:
Block Mode: The fault isolation rule takes effect. Faulty servers are temporarily isolated.
Monitoring Mode: Faulty servers are not isolated. Only logs are printed.
Time Window Size
The time window for fault detection. This parameter works with Minimum Calls in Time Window. Requests are collected only if the number of requests reaches the specified value within the specified time window.
Value range: (1, 60]
Unit: seconds
Minimum Calls in Time Window
The minimum number of requests within the specified time window.
Value ≥ 0.
Abnormal Ratio Threshold (%)
The threshold for the ratio of abnormal requests. This parameter works with Abnormal Ratio Multiplier to determine whether to isolate the target server.
If the ratio of abnormal requests > Abnormal Ratio Threshold × Abnormal Ratio Multiplier, the system considers the target server abnormal and isolates it.
Value range: (0, 100]
Abnormal Ratio Multiplier
The multiplier for the abnormal ratio. This parameter works with Abnormal Ratio Threshold to determine whether to isolate the target server.
Value ≥ 0.
Maximum Isolated Servers
The maximum number of servers to isolate when multiple servers fail.
Value ≥ 0.
Exact Traffic Match (Optional)
Only traffic that meets the exact match conditions is isolated.
You can configure multiple match rules. The rules are joined by a logical AND. The parameters are as follows:
Field: Select System Field.
Field Name: Select Service Provider Application Name.
Logic: Includes Equals, Not Equals, In, Not In, and Regex.
Field Value: Enter the value for the selected field.
Click Submit, and then click OK.
In the list of fault isolation rules, change the status of the rule you just created to On.
Configuration example
An application has three servers: A, B, and C. Their statuses are as follows:
A: Normal
B: Abnormal, with a 20% error rate
C: Abnormal, with a 40% error rate
The fault isolation rule is configured as follows:
Time Window Size: 10
Minimum Calls in Time Window: 20
Abnormal Ratio Threshold: 20
Abnormal Ratio Multiplier: 1
Maximum Isolated Servers: 1
If 20 requests are received within 10 seconds and the abnormal rate is 20% or greater (abnormal ratio multiplier × abnormal ratio threshold), the fault isolation rule is triggered. Because B has a higher failure rate, the system prioritizes fencing B. In this scenario, if the maximum number of fenced instances is set to 2, both A and B are fenced.
Edit a fault isolation rule
You can edit a fault isolation rule at any time. The changes take effect immediately after you submit them.
On the Fault Isolation tab, click Edit in the row of the target fault isolation rule.
Edit the rule as needed, and then click Submit.
Delete a fault isolation rule
You can delete a fault isolation rule. The deletion takes effect immediately. This operation cannot be undone, so proceed with caution.
On the Fault Isolation tab, click Delete in the row of the target fault isolation rule.
Click OK.