Help Center/ CloudTable Service/ User Guide/ Using Doris/ Doris Cluster O&M/ Modifying Doris Parameters to Optimize Cluster Performance
Updated on 2025-10-22 GMT+08:00

Modifying Doris Parameters to Optimize Cluster Performance

To ensure the optimal performance of Doris clusters, you can adjust the cluster configuration parameters multiple times on the console as required.

Constraints

  • Modifying the static parameters of a cluster requires a cluster restart, which may interrupt services. To avoid impacting services, perform this operation during off-peak hours.
  • You can modify parameters only when no task is running in the cluster.
  • Do not modify cluster parameters when the cluster is being restarted.

Procedure

  1. Log in to the CloudTable console.
  2. Select a region in the upper left corner.
  3. In the navigation pane on the left, choose Cluster Management.
  4. In the cluster list, find the target cluster and click the cluster name. The cluster details page is displayed.
  5. In the Parameter Configuration area, click the Parameter Configuration tab and modify Doris parameters.
  6. Select the target FE or BE node, select the parameter to be modified, and click in the Value column.
  7. Enter a new value in the text box and click . The system displays a message "The parameter changed to xx successfully. Save the modified value." The new value is followed by a red asterisk (*).

    To cancel the modification, click .

    Figure 1 Modifying a parameter

  8. Click Save Changes in the upper left corner above the parameter list. A dialog box is displayed.
  9. Check the parameter settings and click OK.

    Figure 2 Saving the modification
    • Click in the upper right corner. If Status of the parameter is Applied, the cluster does not need to be restarted. If Status is Unapplied, you need to manually restart the cluster for the modification to take effect.
    • Click Restart in the upper left corner of the Parameter Configuration page. Alternatively, return to the cluster list to restart the cluster.
      Manually restart the cluster after the static parameters are modified. Dynamic parameters take effect after being modified. You do not need to restart the cluster.
      Figure 3 Restarting the cluster
    • If the new parameter value is incorrect, click Quick Restore to cancel the modification.

  10. After modifying parameters, you can click the Change History tab to view the change history. The following table lists the parameters displayed on the Change History page.

    Table 1 Parameters

    Parameter

    Description

    Name

    Name of the modified parameter

    Old Value

    Parameter value before modification

    New Value

    Parameter value after modification

    Modified

    Time when the parameter was modified

Doris Parameter Modification Status

Table 2 Parameter statuses

Status

Description

Unmodified

The parameter has not been modified.

Unapplied

The parameter has been modified but has not been applied.

Applying

This is a static parameter and you need to restart the cluster after saving the modification.

Applied

The modification of the parameter takes effect.

  • After the modification of a static parameter is saved, the cluster needs to be restarted for the modification to take effect.
  • The modification of a dynamic parameter takes effect after being saved.

FE Node Parameters

Table 3 FE node parameters

Parameter

Static

Description

max_routine_load_job_num

No

  • Description: The total number of routine import jobs, including the jobs in the NEED_SCHEDULED, RUNNING, and PAUSE states. After this parameter value is exceeded, no new jobs can be submitted.
  • Type: int
  • Value range: (0,300)
  • Default value: 100

use_new_tablet_scheduler

Yes

  • Description: Whether to enable the new replica scheduling mode.
  • Type: boolean
  • Value range: [true,false]
  • Default value: true

max_broker_concurrency

No

  • Description: Maximal concurrency of broker scanners.
  • Type: int
  • Value range: (5,30)
  • Default value: 10

tablet_repair_delay_factor_second

No

  • Description: For different scheduling priorities, we will delay different time to start repairing in order to prevent a large number of unnecessary replica repair tasks from occurring in the process of routine restart and upgrade. This parameter is a reference coefficient. For HIGH priority, the delay is the reference coefficient multiplied by 1; for NORMAL priority, the delay is the reference coefficient multiplied by 2; for LOW priority, the delay is the reference coefficient multiplied by 3. That is, the lower the priority, the longer the delay waiting time. If the user wants to repair the replica as soon as possible, this parameter can be reduced appropriately.
  • Type: long
  • Value range: (0,9223372036854775807)
  • Default value: 60

balance_load_score_threshold

No

  • Description: Threshold of the cluster balance percentage. The default value is 0.1, that is, 10%. When the load core of a BE node is not higher than or less than 10% of the average load core, we think that the node is balanced. If you want to make the cluster load more even, you can adjust this parameter appropriately.
  • Type: double
  • Value range: (0,1)
  • Default value: 0.1

dynamic_partition_check_interval_seconds

No

  • Description: The execution frequency of dynamic partition threads. The default value is 600 (10 minutes), that is, scheduling is performed every 10 minutes.

    Unit: second

  • Type: long
  • Value range: (0,9223372036854775807)
  • Default value: 600

max_routine_load_task_concurrent_num

No

  • Description: Limit on the maximum number of concurrent subtasks for a routine import job. It is advised to keep the default value. Setting it too large may result in too many concurrent tasks and consume cluster resources.
  • Type: int
  • Value range: (0,10)
  • Default value: 5

dynamic_partition_enable

No

  • Description: Whether to enable Doris' dynamic partition feature. The default value is false, indicating that the dynamic partition feature is disabled. This parameter only affects the partitioning operation of dynamic partition tables, not normal tables.
  • Type: boolean
  • Value range: [true,false]
  • Default value: true

max_bytes_per_broker_scanner

No

  • Description:

    Maximum bytes a broker scanner can process in one broker load job. Generally, each BE has a broker scanner. Maximum amount of data processed by a single BE. The default value is 500 GB. If the amount of data imported by a single BE is greater than the max_bytes_per_broker_scanner value, an error is reported. In this case, you need to increase the value of max_bytes per_broker_scanner or adjust the number of concurrent import tasks. The calculation logic is as follows: Number of concurrent import tasks = Math.min (Source file size/Minimum processing volume min_bytes_per_broker scanner, Maximum number of concurrent tasks max broker concurrency, Number of current BE nodes). Processing volume of a single BE = Source file size/Number of concurrent import tasks. The default value of min_bytes_per_broker scanner is 64 MB, and the default value of max broker concurrency is 10.

    Unit: byte

  • Type: long
  • Value range: (0,9223372036854775807)
  • Default value: 3221225472

BE Node Parameters

Table 4 BE node parameters

Parameter

Static

Description

max_consumer_num_per_grou

Yes

  • Description: Maximum number of consumers in a data consumer group, used for routine load.
  • Type: int
  • Value range: (0,10)
  • Default value: 3

clone_worker_count

Yes

  • Description: Number of threads used to perform cloning tasks. The value affects the speed of replica equalization. In the case of low disk pressure, you can speed up replica balancing by adjusting this parameter.
  • Type: int
  • Value range: (0,10)
  • Default value: 3