This topic describes the Frontend (FE), Backend (BE), broker, and system parameters and how to modify these parameters.

Background information

Parameters are classified into dynamic parameters and static parameters. Some FE parameters are dynamic parameters that can be modified online. Static parameters are parameters that take effect only after a service restart.

This topic describes the following parameters:

FE dynamic parameters

You can run the following command to modify a dynamic parameter: 
ADMIN SET FRONTEND CONFIG ("key" = "value");

Logging

Parameter Default value Description
qe_slow_log_ms 5000 The threshold that is used to determine whether a query is a slow query. If the response time of a query exceeds this threshold, the query is recorded as a slow query. Unit: milliseconds.

Metadata and cluster management

Parameter Default value Description
catalog_try_lock_timeout_ms 5000 The timeout period to obtain the catalog lock. Unit: milliseconds.
edit_log_roll_num 50000 The maximum number of image log entries that can be written before a log file is created for these log entries.
ignore_unknown_log_id FALSE Specifies whether to ignore unknown log IDs. Valid values:
  • TRUE: ignores unknown log IDs.
  • FALSE: The FE exits if unknown log IDs exist.

Some log IDs may be unknown to BEs of earlier versions if an FE is rolled back to an earlier version.

ignore_meta_check FALSE Specifies whether to ignore the metadata gap. Valid values:
  • FALSE: ignores the metadata gap.
  • TRUE: does not ignore the metadata gap.
max_backend_down_time_second 3600 The maximum amount of time for a BE to rejoin after the BE disconnects to the FE. Unit: seconds.
drop_backend_after_decommission TRUE Specifies whether to delete a BE after the BE is decommissioned. Valid values:
  • TRUE: deletes the BE.
  • FALSE: does not delete the BE.

Query engine

Parameter Default value Description
expr_children_limit 10000 The number of IN predicates that can be involved in a query.
expr_depth_limit 3000 The number of layers that can be nested in a query.
max_allowed_in_element_num_of_delete 10000 The maximum number of elements allowed in the IN predicate of a DELETE statement.
max_layout_length_per_row 2147483647 The maximum length of a single row.
disable_cluster_feature TRUE Specifies whether to disable the logical cluster feature. Valid values:
  • TRUE: disables this feature.
  • FALSE: enables this feature.
enable_materialized_view TRUE Specifies whether to enable materialized views. Valid values:
  • TRUE: enables materialized views.
  • FALSE: disables materialized views.
enable_decimal_v3 TRUE Specifies whether to support the DECIMAL V3 data type. Valid values:
  • TRUE: supports this data type.
  • FALSE: does not support this data type.
enable_sql_blacklist FALSE Specifies whether to enable the blacklist for SQL queries.
  • TRUE: enables the blacklist.
  • FALSE: disables the blacklist.
Note If you enable this feature, queries in the blacklist cannot be executed.
dynamic_partition_check_interval_seconds 600 The interval at which new data is checked. If new data is detected, StarRocks automatically creates partitions for the data. Unit: seconds.
dynamic_partition_enable TRUE Specifies whether to enable the dynamic partitioning feature. Valid values:
  • TRUE: enables this feature.
  • FALSE: disables this feature.
max_partitions_in_one_batch 4096 The maximum number of partitions that can be created at a time.
max_query_retry_time 2 The maximum number of query retries on the FE.
max_create_table_timeout_second 60 The maximum timeout period for creating a table. Unit: seconds.
max_running_rollup_job_num_per_table 1 The maximum number of concurrent rollup jobs for each table.
max_planner_scalar_rewrite_num 10_0000 The maximum number of times that the optimizer can rewrite a ScalarOperator.
statistics_manager_sleep_time_sec 60*10 The interval at which statistics tables are created. Unit: seconds.
statistic_collect_interval_sec 120*60 The interval at which statistics are collected. Unit: seconds.
statistic_update_interval_sec 24 *60* 60 The default interval at which statistics are updated. Unit: seconds.
statistic_sample_collect_rows 200000 The number of rows to collect for sampled collection.
enable_statistic_collect TRUE Specifies whether to enable the statistics collection feature. Valid values:
  • TRUE: enables this feature.
  • FALSE: disables this feature.
enable_local_replica_selection FALSE
Specifies whether the optimizer preferentially selects tablet replicas on BEs that have the same IP address as the FE.
  • TRUE: preferentially selects tablet replicas on BEs that have the same IP address as the FE.
  • FALSE: does not preferentially select tablet replicas on BEs that have the same IP address as the FE.
max_distribution_pruner_recursion_depth 100 The maximum recursion depth for partition pruning.

Loading and unloading

Parameter Default value Description
load_straggler_wait_second 300 The maximum loading lag that a BE replica can tolerate. Unit: seconds.

If this value is exceeded, data is cloned from other replicas.

desired_max_waiting_jobs 100 The maximum number of pending jobs.

This limit applies to the total number of jobs of all types, such as table creation, loading, and schema change jobs.

max_running_txn_num_per_db 100 The maximum number of concurrent load jobs.
max_load_timeout_second 259200 The maximum timeout period for a load job. This limit applies to all types of load jobs. Unit: seconds.
min_load_timeout_second 1 The minimum timeout period for a load job. This limit applies to all types of load jobs. Unit: seconds.
load_parallel_instance_num 1 The number of concurrent instances on a single BE.
disable_hadoop_load FALSE Specifies whether to disable the feature that is used to load data from Hadoop. Valid values:
  • TRUE: disables this feature.
  • FALSE: enables this feature.
disable_load_job FALSE Specifies whether to accept a load job if the cluster encounters an error. Valid values:
  • TRUE: accepts the load job.
  • FALSE: does not accept the load job.
db_used_data_quota_update_interval_secs 300 The interval at which the quota for a database is updated. Unit: seconds.
history_job_keep_max_second 604800 The maximum retention period for historical jobs. Unit: seconds.
label_keep_max_num 1000 The maximum number of load jobs that can be retained within a period of time.

You can specify the retention period by using the label_keep_max_second parameter.

label_keep_max_second 259200 The retention period of labels. Unit: seconds.
max_routine_load_job_num 100 The maximum number of Routine Load jobs.
max_routine_load_task_concurrent_num 5 The maximum number of concurrent tasks supported in a Routine Load job.
max_routine_load_task_num_per_be 5 The maximum number of concurrent Routine Load tasks on a BE. The value must be less than or equal to the number specified by the routine_load_thread_pool_size parameter.
max_routine_load_batch_size 524288000 The maximum volume of data that can be loaded by a Routine Load task.
routine_load_task_consume_second 3 The maximum amount of time for which a Routine Load task consumes data. Unit: seconds.
routine_load_task_timeout_second 15 The timeout period for a Routine Load task. Unit: seconds.
max_tolerable_backend_down_num 0 The maximum number of faulty BEs allowed. If this value is exceeded, Routine Load jobs cannot be automatically recovered.
period_of_auto_resume_min 5 The interval at which Routine Load jobs are automatically recovered.
spark_load_default_timeout_second 86400 The timeout period for a Spark Load job. Unit: seconds.
spark_home_default_dir STARROCKS_HOME_DIR/lib/spark2x The root directory of a Spark client.
stream_load_default_timeout_second 600 The timeout period for a Stream Load job. Unit: seconds.
max_stream_load_timeout_second 259200 The maximum timeout period for a Stream Load job. Unit: seconds.
insert_load_default_timeout_second 3600 The timeout period for an INSERT INTO statement that is used to load data. Unit: seconds.
broker_load_default_timeout_second 14400 The timeout period for a Broker Load job. Unit: seconds.
min_bytes_per_broker_scanner 67108864 The minimum volume of data that can be processed by a Broker Load instance. The default value specifies 64 MB.
max_broker_concurrency 100 The maximum number of concurrent instances for a Broker Load job.
export_max_bytes_per_be_per_task 268435456 The maximum volume of data that can be exported from a single BE by a single data unload job. The default value specifies 256 MB.
export_running_job_num_limit 5 The maximum number of concurrent data unload jobs.
export_task_default_timeout_second 7200 The timeout period for a data unload job. Unit: seconds. The default value specifies 2 hours.

Storage

Parameter Default value Description
enable_strict_storage_medium_check FALSE Specifies whether the FE checks the free storage space of BEs when a table is created. Valid values:
  • TRUE: checks the free storage space of BEs.
  • FALSE: does not check the free storage space of BEs.
capacity_used_percent_high_water 0.75 The upper limit of disk usage on a BE.

If the disk usage exceeds 75%, table creation or clone jobs will not be sent to this BE until the disk usage decreases to a value lower than the threshold.

storage_high_watermark_usage_percent 85 The maximum space usage in the storage directory of a BE.
storage_min_left_capacity_bytes 2 *1024* 1024*1024 The minimum remaining space in the storage directory of a BE. The default value specifies 2 GB.
storage_flood_stage_left_capacity_bytes 1 *1024* 1024*1024 The remaining space threshold in the storage directory of a BE.

If the remaining space is less than this value, load or restore jobs are rejected. The default value specifies 1 GB.

storage_flood_stage_usage_percent 95 The space usage threshold in the storage directory of a BE.

If the space usage is greater than this value, load or restore jobs are rejected.

catalog_trash_expire_second 86400 The retention period of metadata in the recycle bin after a table or database is deleted. Unit: seconds. The default value specifies one day.

If the retention period is longer than this value, data cannot be restored.

alter_table_timeout_second 86400 The timeout period for a schema change job. Unit: seconds. The default value specifies one day.
balance_load_disk_safe_threshold 0.5 This parameter is valid only when the disk_and_tablet policy is used.

If the disk usage of all BE disks is less than 50%, the disk usage is considered balanced.

balance_load_score_threshold 0.1
  • If the be_load_score policy is used, a BE whose load is 10% lower than the average load is in a low load state, and a BE whose load is 10% higher than the average load is in a high load state.
  • If the disk_and_tablet policy is used and the difference between the maximum and minimum BE disk usage is greater than 10%, the disk usage is considered unbalanced, and load rebalancing is implemented among tablets.
disable_balance FALSE Specifies whether to disable tablet balancing. Valid values:
  • FALSE: disables tablet balancing.
  • TRUE: enable tablet balancing.
max_scheduling_tablets 2000 The number of tablets being scheduled.

If the number of tablets being scheduled is greater than this value, the load balancing among tablets is skipped.

max_balancing_tablets 100 The number of tablets being balanced.

If the number of tablets being balanced is greater than this value, the load rebalancing among tablets is skipped.

disable_colocate_balance FALSE Specifies whether to disable replica balancing for colocate tables. Valid values:
  • TRUE: disables replica balancing for colocate tables.
  • FALSE: enables replica balancing for colocate tables.
recover_with_empty_tablet FALSE Specifies whether to use an empty tablet replica instead if a tablet replica is lost or corrupted. Valid values:
  • TRUE: uses an empty tablet replica instead.
  • TRUE: does not use an empty tablet replica instead.

You can use an empty tablet replica instead to ensure that a query can be executed even if a tablet replica is lost or corrupted. However, the result may be incorrect because of the missing data.

min_clone_task_timeout_sec 3*60 The minimum timeout period for cloning a tablet. Unit: seconds. The default value specifies 3 minutes.
max_clone_task_timeout_sec 2 *60* 60 The maximum timeout period for cloning a tablet. Unit: seconds. The default value specifies 2 hours.
tablet_create_timeout_second 1 The timeout period for creating a table. Unit: seconds.
tablet_delete_timeout_second 2 The timeout period for deleting a table. Unit: seconds.
tablet_repair_delay_factor_second 60 The interval at which tablet replicas are repaired. Unit: seconds.
consistency_check_start_time 23 The start time when the FE initiates a replica consistency check. The default value specifies 23:00.
consistency_check_end_time 4 The end time when the FE initiates a replica consistency check. The default value specifies 4:00.
check_consistency_default_timeout_second 600 The timeout period for a replica consistency check. Unit: seconds.

Others

Parameter Default value Description
plugin_enable TRUE Specifies whether to enable the plug-in feature. Valid values:
  • TRUE: enables the plug-in feature.
  • FALSE: disables the plug-in feature.
Note You can install or uninstall plug-ins only on master nodes.
max_small_file_number 100 The maximum number of small files that can be stored.
max_small_file_size_bytes 1024*1024 The maximum size of a small file. The default value specifies 1 MB.
backup_job_default_timeout_ms 86400*1000 The timeout period for a backup job. Unit: seconds. The default value specifies one day.
report_queue_size 100 The maximum number of jobs that can wait in a report queue. The report is about disk, task, and tablet information of BEs.

FE static parameters

Logging

Parameter Default value Description
log_roll_size_mb 1024 The maximum size of a log file. If this value is exceeded, a new log file is created. The default value specifies 1 GB.
sys_log_dir StarRocksFe.STARROCKS_HOME_DIR/log The directory in which system log files are stored.
sys_log_level INFO The severity levels into which system logs are classified.

Valid values: INFO, WARNING, ERROR, and FATAL.

sys_log_verbose_modules Empty string The modules for which system logs are generated.

For example, if this parameter is set to org.apache.starrocks.catalog, system logs are generated only for the catalog module.

sys_log_roll_interval DAY The interval at which system log files are rotated.
sys_log_delete_age 7d The interval at which system log files are deleted.
sys_log_roll_num 10 The maximum number of system log files that can be retained within each retention period specified by the sys_log_roll_interval parameter.
audit_log_dir starrocksFe.STARROCKS_HOME_DIR/log The directory in which audit log files are stored.
audit_log_roll_num 90 The maximum number of audit log files that can be retained.
audit_log_modules "slow_query", "query" The modules for which audit logs are generated. By default, audit logs are generated for the slow_query and query modules.
audit_log_roll_interval DAY The interval at which audit log files are rotated. Valid values: DAY and HOUR.
audit_log_delete_age 30d The interval at which audit log files are deleted.
dump_log_dir STARROCKS_HOME_DIR/log The directory in which dump log files are stored.
dump_log_modules "query" The modules for which dump logs are generated. By default, dump logs are generated for the query module.
dump_log_roll_interval DAY The interval at which dump log files are rotated.

If this parameter is set to DAY, a suffix in the yyyyMMdd format is added to the names of dump log files. If this parameter is set to HOUR, a suffix in the yyyyMMddHH format is added to the names of dump log files.

dump_log_roll_num 90 The maximum number of dump log files that can be retained within each retention period specified by the dump_log_roll_interval parameter.
dump_log_delete_age 30d The interval at which dump log files are deleted.

Server

Parameter Default value Description
frontend_address 0.0.0.0 The IP address of the FE.
priority_networks Empty string The CIDR block of the BE, such as 10.10.**.**/24. This parameter can be used for servers that have multiple IP addresses. You can configure this parameter to specify the network that is preferentially used.
http_port 8030 The port of the HTTP server.
http_backlog_num 1024 The length of the backlog queue that is held by the HTTP server.
cluster_name StarRocks Cluster The name of the StarRocks cluster, which is displayed for Title on the web page.
rpc_port 9020 The port of the Thrift server on the FE.
thrift_backlog_num 1024 The length of the backlog queue that is held by the Thrift server.
thrift_server_type THREAD_POOL The service model that is used by the Thrift server on the FE. Valid values: SIMPLE, THREADED, and THREAD_POOL.
thrift_server_max_worker_threads 4096 The maximum number of worker threads that are supported by the Thrift server.
thrift_client_timeout_ms 0 The timeout period for requests from clients.

The default value 0 specifies that requests from clients never time out.

brpc_idle_wait_max_time 10000 The maximum amount of time for which bRPCs wait in the idle state. Unit: milliseconds. The default value specifies 10 seconds.
query_port 9030 The port of the MySQL server on the FE.
mysql_service_nio_enabled FALSE Specifies whether to enable asynchronous I/O for the FE. Valid values:
  • TRUE: enables asynchronous I/O for the FE.
  • FALSE: disables asynchronous I/O for the FE.
mysql_service_io_threads_num 4 The maximum number of threads that can be run by the MySQL server on the FE to process I/O events.
mysql_nio_backlog_num 1024 The length of the backlog queue that is held by the MySQL server.
max_mysql_service_task_threads_num 4096 The maximum number of threads that can be run by the MySQL server to process tasks.
max_connection_scheduler_threads_num 4096 The maximum number of threads that are supported by the connection scheduler.
qe_max_connection 1024 The maximum number of connections that can be established by all users to the FE.
check_java_version TRUE Specifies whether to check version compatibility between the executed and compiled Java programs. Value values:
  • TRUE: check version compatibility between the executed and compiled Java programs.
  • FALSE: does not check version compatibility between the executed and compiled Java programs.

Metadata and cluster management

Parameter Default value Description
meta_dir StarRocksFe.STARROCKS_HOME_DIR/meta The directory in which metadata is retained.
heartbeat_mgr_threads_num 8 The number of threads that can be run by Heartbeat Manager to run heartbeat tasks.
heartbeat_mgr_blocking_queue_size 1024 The size of the blocking queue that stores heartbeat tasks run by Heartbeat Manager.
metadata_failure_recovery FALSE Specifies whether to forcibly reset the metadata of the FE.
Note Exercise caution when you configure this parameter.
edit_log_port 9010 The port that is used for communication among the master, follower, and observer FEs in the StarRocks cluster.
edit_log_type BDB The type of edit log that can be generated. Set the value to BDB.
bdbje_heartbeat_timeout_second 30 The timeout period for heartbeats in the BDB JE-based FE. Unit: seconds.
bdbje_lock_timeout_second 1 The timeout period for a lock on the BDB JE-based FE. Unit: seconds.
max_bdbje_clock_delta_ms 5000 The maximum clock offset that is allowed between the master FE and the follower or observer FEs in the StarRocks cluster. Unit: milliseconds.
txn_rollback_limit 100 The maximum number of transactions that can be rolled back.
bdbje_replica_ack_timeout_second 10 The maximum amount of time for which the master FE can wait for ACK messages from a specified number of follower FEs.
master_sync_policy SYNC The policy based on which the master FE flushes logs to disk.
replica_sync_policy SYNC The policy based on which the follower FE replica flushes logs to disk.
meta_delay_toleration_second 300 The maximum duration by which the metadata on the follower and observer FEs can lag behind that on the master FE. Unit: seconds.
cluster_id -1 The ID of the cluster to which the FE belongs. FEs or BEs that have the same cluster ID belong to the same cluster.

The default value -1 specifies that a random cluster ID is generated when the master FE of the cluster is started for the first time.

Query engine

Parameter Default value Description
disable_colocate_join FALSE Specifies whether to enable colocate joins. Valid values:
  • FALSE: disables colocate joins.
  • TRUE: enables colocate joins.
enable_udf FALSE Specifies whether to enable user defined functions (UDFs). Valid values:
  • FALSE: disable UDFs.
  • TRUE: enables UDFs.
publish_version_interval_ms 10 The interval at which release validation tasks are issued.
statistic_cache_columns 10_0000 The number of rows that can be cached for the statistics table.

Loading and unloading

Parameter Default value Description
async_load_task_pool_size 10 The thread pool size of load jobs.
load_checker_interval_second 5 The polling interval for load jobs. Unit: seconds.
transaction_clean_interval_second 30 The interval at which finished transactions are cleaned up. Unit: seconds.
label_clean_interval_second 14400 The interval at which labels are cleaned up. Unit: seconds.
spark_dpp_version 1.0.0 The version of Spark Dynamic Partition Pruning (DPP).
spark_resource_path Empty string The root directory of the Spark dependency package.
spark_launcher_log_dir sys_log_dir/spark_launcher_log The directory in which Spark log files are stored.
yarn_client_path STARROCKS_HOME_DIR/lib/yarn-client/hadoop/bin/yarn The root directory of the YARN client package.
yarn_config_dir STARROCKS_HOME_DIR/lib/yarn-config The directory in which the YARN configuration files are stored.
export_checker_interval_second 5 The polling interval for unload threads. Unit: seconds.
export_task_pool_size 5 The thread pool size of unload jobs.
export_checker_interval_second 5 The interval at which unload jobs are scheduled. Unit: seconds.

Storage

Parameter Default value Description
storage_cooldown_second 2592000 The interval for storage media migration. Unit: seconds. The default value specifies 30 days.
default_storage_medium HDD The default storage media. Valid values: HDD and SSD.

The default storage media is used if no storage media is specified when you create a table or partition.

schedule_slot_num_per_path 2 The number of tablet-related jobs that can concurrently run in the storage directory of a BE.
tablet_balancer_strategy disk_and_tablet The policy based on which load balancing is implemented among tablets. Valid values: disk_and_tablet and be_load_score.
tablet_stat_update_interval_second 300 The interval at which the FE retrieves tablet statistics from each BE. Unit: seconds. The default value specifies 5 minutes.

Others

Parameter Default value Description
plugin_dir STARROCKS_HOME_DIR/plugins The directory in which the plug-in installation packages are stored.
small_file_dir STARROCKS_HOME_DIR/small_files The root directory of small files.
max_agent_task_threads_num 4096 The maximum number of threads in the thread pool of agent jobs.
authentication_ldap_simple_bind_base_dn "" The base DN that is used to search for the authentication information about users.
authentication_ldap_simple_bind_root_dn "" The administrator DN that is used to search for the authentication information about users.
authentication_ldap_simple_bind_root_pwd "" The password of the administrator that is used to search for the authentication information about users.
authentication_ldap_simple_server_host "" The host on which the LDAP server runs.
authentication_ldap_simple_server_port 389 The port of the LDAP server.
authentication_ldap_simple_user_search_attr uid The name of the attribute that identifies users in LDAP objects.
tmp_dir starrocksFe.STARROCKS_HOME_DIR/temp_ddir The directory in which temporary files are stored, such as files generated during backup and restore procedures.
locale zh_CN.UTF-8 The character set.
hive_meta_load_concurrency 4 The maximum number of concurrent threads that are supported for Hive metadata.
hive_meta_cache_refresh_interval_s 4096 The interval at which the cached metadata of Hive external tables is updated. Unit: seconds.
hive_meta_cache_ttl_s 3600 *2 The amount of time after which the cached metadata of Hive external tables expires. Unit: seconds. The default value specifies 2 hours.
hive_meta_store_timeout_s 3600 *24 The timeout period for a connection to a Hive metastore. Unit: seconds. The default value specifies 24 hours.
es_state_sync_interval_second 10 The interval at which the FE obtains Elasticsearch indexes. Unit: seconds.
enable_auth_check TRUE Specifies whether to enable the authentication check feature. Valid values:
  • TRUE: enables this feature.
  • FALSE: disables this feature.
enable_metric_calculator TRUE Specifies whether to enable the feature that is used to periodically collect metrics. Valid values:
  • TRUE: enables this feature.
  • FALSE: disables this feature.

BE parameters

All BE parameters are static parameters.

Parameter Default value Description
be_port 9060 The port of the Thrift server on the BE. This port is used to receive requests from FEs.
brpc_port 8060 The bRPC port that is used to view the network statistics of bRPCs.
brpc_num_threads -1 The number of bthreads of a bRPC. The default value -1 specifies the same number as CPU cores.
priority_networks Empty string The CIDR block of the BE, such as 10.10.**.**/24. This parameter can be used for servers that have multiple IP addresses. You can configure this parameter to specify the network that is preferentially used.
heartbeat_service_port 9050 The port of the heartbeat service on the BE. This port is used to receive heartbeats from FEs.
heartbeat_service_thread_count 1 The number of threads for the heartbeat service.
create_tablet_worker_count 3 The number of threads that are used to create a tablet.
drop_tablet_worker_count 3 The number of threads that are used to delete a tablet.
push_worker_count_normal_priority 3 The number of threads that are used to process a load job with the NORMAL priority.
push_worker_count_high_priority 3 The number of threads that are used to process a load job with the HIGH priority.
publish_version_worker_count 2 The number of threads that are used to publish a version.
clear_transaction_task_worker_count 1 The number of threads that are used to clear transactions.
alter_tablet_worker_count 3 The number of threads that are used to change a schema.
clone_worker_count 3 The number of threads that are used for cloning.
storage_medium_migrate_count 1 The number of threads that are used for storage media migration. For example, hot data can be migrated from an SSD to an SATA disk by using these threads.
check_consistency_worker_count 1 The number of threads that are used to check the consistency of tablets.
report_task_interval_seconds 10 The interval at which a job is reported. Unit: seconds.

A job can be initiated to create a table, delete a table, load data, or change the schema of a table.

report_disk_state_interval_seconds 60 The interval at which disk status is reported. Unit: seconds.

The disk status includes the status and data volume of each disk.

report_tablet_interval_seconds 60 The interval at which tablet data are reported. Unit: seconds.

The data includes the latest versions of all tablets.

alter_tablet_timeout_seconds 86400 The timeout period for a schema change job. Unit: seconds.
sys_log_dir ${DORIS_HOME}/log The directory in which system logs are stored.

Levels of systems logs include INFO, WARNING, ERROR, and FATAL.

user_function_dir ${DORIS_HOME}/lib/udf The directory in which UDFs are stored.
sys_log_level INFO The severity levels into which system log entries are classified.

Valid values: INFO, WARNING, ERROR, and FATAL.

sys_log_roll_mode SIZE-MB-1024 The maximum size of a log file. If this value is exceeded, a new log file is created. The default value specifies 1 GB.
sys_log_roll_num 10 The maximum number of system log files that can be retained.
sys_log_verbose_modules Empty string The modules for which system logs are generated. For example, if this parameter is set to OLAP, system logs are generated only for the OLAP module.
sys_log_verbose_level 10 The level of the logs to be displayed. This parameter is used to control the output of logs that start with VLOG in codes.
log_buffer_level Empty string The policy based on which logs are flushed. By default, logs are buffered in memory.
num_threads_per_core 3 The number of threads that are started in each CPU core.
compress_rowbatches TRUE Specifies whether to compress the row batches in RPCs between BEs. This parameter is used for the data transmission between query layers.
serialize_batch FALSE Specifies whether to serialize the row batches in RPCs between BEs. This parameter item is used for the data transmission between query layers.
status_report_interval 5 The interval at which a profile is reported. Unit: seconds. This parameter is used to collect query statistics for FEs.
doris_scanner_thread_pool_thread_num 48 The number of threads in the thread pool that is used by the storage engine for concurrent scan tasks.
doris_scanner_thread_pool_queue_size 102400 The maximum number of tasks that the storage engine supports.
doris_scan_range_row_count 524288 The granularity in which the storage engine splits the range of a query task.
doris_scanner_queue_size 1024 The maximum number of scan tasks that the storage engine supports.
doris_scanner_row_num 16384 The maximum number of data rows that can be returned by each scan thread at a time.
doris_max_scan_key_num 1024 The maximum number of scan keys that can be split.
column_dictionary_key_ratio_threshold 0 The value ratio for the STRING type. If the ratio is less than this value, the dictionary compression algorithm is used.
column_dictionary_key_size_threshold 0 The size of the dictionary compression column. If the size is smaller than this value, the dictionary compression algorithm is used.
memory_limitation_per_thread_for_schema_change 2 The maximum memory for a schema change job.
file_descriptor_cache_clean_interval 3600 The interval at which the cached file descriptors are cleared. Unit: seconds. This parameter is used to clean up file handles that have not been used for a long time.
disk_stat_monitor_interval 5 The interval at which the disk status is checked. Unit: seconds.
unused_rowset_monitor_interval 30 The interval at which expired rowsets are cleared. Unit: seconds.
storage_root_path Empty string The directory in which data is stored.
max_tablet_num_per_shard 1024 The maximum number of tablets in each shard. This parameter is used to restrict the number of tablet child directories within each storage directory.
pending_data_expire_time_sec 1800 The maximum retention period of the pending data in the storage engine. Unit: seconds.
inc_rowset_expired_sec 1800 The retention period of loaded data in the storage engine in the incremental cloning scenario. Unit: seconds.
max_garbage_sweep_interval 3600 The maximum interval for garbage collection on disks. Unit: seconds.
min_garbage_sweep_interval 180 The minimum interval for garbage collection on disks. Unit: seconds.
snapshot_expire_time_sec 172800 The interval at which snapshots are deleted. Unit: seconds. The default value specifies 48 hours.
trash_file_expire_time_sec 259200 The interval at which the files in the recycle bin are deleted. Unit: seconds. The default value specifies 72 hours.
file_descriptor_cache_capacity 16384 The number of file descriptors that can be cached.
min_file_descriptor_number 60000 The minimum number of file descriptors in the BE process.
index_stream_cache_capacity 10737418240 The cache capacity for the statistical information of BloomFilter, Min, and Max.
storage_page_cache_limit 0 The capacity of the page cache.
disable_storage_page_cache TRUE Specifies whether to disable the page cache.
  • TRUE: disables the page cache.
  • FALSE: enables the page cache.
base_compaction_check_interval_seconds 60 The interval of thread polling for a base compaction. Unit: seconds.
base_compaction_num_threads_per_disk 1 The number of threads used for base compaction on each disk.
base_cumulative_delta_ratio 0.3 The ratio of cumulative file size to base file size. This is one of the conditions that trigger a base compaction.
base_compaction_interval_seconds_since_last_operation 86400 The interval since the last base compaction. This is one of the conditions that trigger a base compaction.
cumulative_compaction_check_interval_seconds 1 The interval of thread polling for a cumulative compaction. Unit: seconds.
min_cumulative_compaction_num_singleton_deltas 5 The minimum number of singleton files. This is one of the conditions that trigger a cumulative compaction.
max_cumulative_compaction_num_singleton_deltas 1000 The maximum number of singleton files. This is one of the conditions that trigger a cumulative compaction.
cumulative_compaction_num_threads_per_disk 1 The number of threads for cumulative compaction on each disk.
min_compaction_failure_interval_sec 120 The minimum interval that a tablet compaction can be scheduled since the last compaction failure. Unit: seconds.
max_compaction_concurrency -1 The maximum concurrency of base compactions and cumulative compactions.

The default value -1 specifies that no limit is imposed on the concurrency.

webserver_port 8040 The port of the HTTP server.
webserver_num_workers 48 The number of threads that are used by the HTTP server.
periodic_counter_update_period_ms 500 The interval at which the Counter statistics are collected. Unit: milliseconds.
load_data_reserve_hours 4 The retention period of files generated by small-scale loadings. Unit: hours.
load_error_log_reserve_hours 48 The retention period of data loading logs. Unit: hours.
number_tablet_writer_threads 16 The number of threads used for Stream Load.
streaming_load_max_mb 10240 The maximum size of a file that can be loaded into StarRocks in a stream.
streaming_load_rpc_max_alive_time_sec 1200 The RPC timeout for Stream Load.
fragment_pool_thread_num 64 The number of threads that are used for queries. By default, 64 threads are started. Threads are dynamically created for subsequent queries.
fragment_pool_queue_size 2048 The maximum number of queries that can be processed on a node.
enable_partitioned_aggregation TRUE Specifies whether to enable partition aggregation. Valid values:
  • TRUE: enables partition aggregation.
  • FALSE: disables partition aggregation.
enable_token_check TRUE Specifies whether to enable the token check.
  • TRUE: enable the token check.
  • FALSE: disables the token check.
load_process_max_memory_limit_bytes 107374182400 The maximum size of memory that can be used by all load threads on a node. The default value specifies 100 GB.
load_process_max_memory_limit_percent 30 The maximum percentage of memory that can be used by all load threads on a node.
sync_tablet_meta FALSE Specifies whether to enable the synchronization of tablet metadata.
thrift_rpc_timeout_ms 5000 The timeout period for a thrift RPC. Unit: milliseconds.
txn_commit_rpc_timeout_ms 10000 The timeout period for a transaction commit RPC. Unit: milliseconds.
routine_load_thread_pool_size 10 The thread pool size of Routine Load.
tablet_meta_checkpoint_min_new_rowsets_num 10 The minimum number of rowsets to create since the last tablet meta checkpoint.
tablet_meta_checkpoint_min_interval_secs 600 The interval of thread polling for generating a tablet meta checkpoint. Unit: seconds.
brpc_max_body_size 209715200 The maximum body size of a BRPC. The default value specifies 200 MB.
max_runnings_transactions 2000 The maximum number of transactions that the storage engine supports.
tablet_map_shard_size 32 The size of a tablet map shard.
enable_bitmap_union_disk_format_with_set FALSE Specifies whether to enable the new storage format of the BITMAP type. This type can improve the performance of bitmap_union.