Cluster Parameter Configuration
The modification of cluster parameter configuration must be restarted before it takes effect. You can choose to “Restart Now” or “Restart Later” according to the actual situation.
”Restart Now”: The cluster restart operation is performed immediately, all nodes will be restarted one after another, and the service will not be interrupted during the restart. It is recommended to operate during off-peak business hours. After restarting, the new parameters will take effect immediately. “Restart Later”: The new parameters will take effect after the node service is restarted next time; if you need some parameters to take effect immediately without restarting the service, you can modify the related configuration for a single Topic on the Topic Management page.
- On the UKafka cluster list or detail page, click on “Cluster Configuration” to enter the cluster parameter configuration information page.
- Modify the cluster parameter configuration as needed
The parameters that can be configured for the cluster are shown in the following table:
| Parameter Name | Parameter Type | Range | Description |
|---|---|---|---|
| auto.create.topics.enable | bool | true/false | Whether to allow automatic creation of Topics |
| auto.leader.rebalance.enable | bool | true/false | Whether to enable data balance between nodes |
| default.replication.factor | int | 1-10 | Number of default replication factors (number of Topic partition replicas) |
| delete.topic.enable | bool | true/false | Whether to allow deletion of topic information |
| log.cleanup.policy | string | delete/compact | Log cleanup policy, delete/compact |
| log.retention.hours | int | 1-10000 | Minimum log retention time |
| message.max.bytes | int | 1-2147483646 | Maximum size of the message body |
| num.io.threads | int | 1-32 | Number of threads handling disk IO |
| num.network.threads | int | 1-32 | Number of threads handling network requests |
| num.partitions | int | 1-1000 | Number of partitions |
| replica.fetch.max.bytes | int | 1-2147483646 | Maximum size of messages during replica synchronization |
Points to note:
auto.create.topics.enable Production business is recommended to be turned off to avoid affecting the stable operation of the online environment.
default.replication.factor Must be at least 2, otherwise, it cannot guarantee service high availability.
replica.fetch.max.bytes Should be greater than or equal to message.max.bytes.
num.io.threads and num.network.threads Can control the node load and cluster performance balance.
- After the parameters have been modified, click on “Confirm” to enter the restart strategy page. Choose “Restart Now” or “Restart Later” as needed.