The DCS Redis management console supports monitoring and alarm functions, and also supports the configuration of event alarms or threshold alarms.
Procedure
1. Log in to the Redis management console.
2. Select the region of the instance in the upper left corner of the management console.
3. On the instance list page, click the target instance name to go to the instance details management page.
4. Click Alarm Management to go to the alarm configuration management page.
5. To set up a notification group, first create a New Contact Group, then create a New Contact, and enter the contact information for alarm notification.
6. Click Create Notification Policy to set the notification information template.
7. Create Alarm Rule
8. View the alarm rules that have been set. You can add and delete alarm rules on this page;
Suggestions for Alarm Metrics and Alarm Policy Configuration
Metric Name | Range | Description | Suggestions for Setting Alarm Thresholds | Suggestions for Handling Alarms |
Metric Name | Range | Description | Suggestions for Setting Alarm Thresholds | Suggestions for Handling Alarms |
Alarm for too many Redis client connections | 0-10000 | This metric indicates the number of connected Redis clients. | Recommended alarm threshold: 8000 | The number of connections exceeding the maximum limit should be avoided, and the threshold can be adjusted according to your actual business conditions. |
Low Redis cache hit ratio | 0-100% | This metric indicates the cache hit ratio of Redis. The hit ratio formula is: keyspace_hits/(keyspace_hits+keyspace_misses) %. | Recommended alarm threshold: 80% | Suggestion: A low cache hit ratio may indicate that the cache data is not complete or the cache policy is not effective enough, which results in a large number of requests unable to obtain data from the Redis cache and increases the access load on back-end resources. It is recommended to set the cache expiration time properly, use appropriate cache expiration policies, and prefetch hotspot data to improve the cache hit ratio. |
Alarm for the number of blocked clients | 0-10000 | This metric indicates the number of clients that have been suspended by block operations. Block operations include BLPOP, BRPOP, BRPOPLPUSH. | Depending on the specific business. The generally recommended threshold is: 50. | Processing method: Check whether there are slow requests; suggestion: Optimize and adjust Redis configuration parameters, such as the maximum number of connections, maximum clients, etc., to ensure that their values are appropriate. |
Redis memory utilization alarm | 0-100% | This metric indicates the memory usage of the current Redis instance node. | Recommended alarm threshold: 70% | Processing method: Check whether unnecessary Redis memory can be cleared to reduce memory usage. Suggestion: If the existing cached data cannot be cleared, it is recommended to expand the capacity; |
Redis CPU utilization alarm | 0-100% | This metric indicates the CPU usage of the current Redis instance node. | Recommended alarm threshold: 70% | Processing method: Check whether there are requests to query big keys and hot keys in the current business. Suggestion: If there are big keys and hot keys, please see suggestions in the FAQ to reduce the size of big keys and hot keys. |
Alarm for the number of access client connections | 0-10000 | This metric indicates the number of clients connected to the proxy node. | Alarm threshold: 8000 | Main processing method: Check whether the business request volume is normal. If the business request is normal, you can continue to observe for a while. Suggestion: Avoid the number of connections exceeding the maximum limit, which can be adjusted according to your actual business conditions. Check whether the current number of business connections is normal; |
Alarm for too much access QPS | 0-100000 | This metric indicates the QPS of the proxy node. | Alarm threshold: 10000 | Processing method: Check whether the business request is normal. If the business request is normal, you can continue to observe for a while; |