System Variable | Meaning | Parameter Type | Default Value | Value Range | Scope | Supporting SET_VAR Hint |
libra_enable_runtime_filter | Whether to enable Runtime Filter. | BOOL | ON | ON: Enable Runtime Filter. OFF: Disable Runtime Filter. | Global & Session | Yes |
libra_runtime_filter_type | Set the Runtime Filter type that can be assigned. | VARCHAR | MIN_MAX, BLOOM_FILTER, IN_FILTER | BLOOM_FILTER: Builds a Bloom Filter for the JOIN KEY of the JOIN BUILD side to filter the Probe side data. MIN_MAX: Builds maximum and minimum values for the JOIN KEY of the JOIN BUILD side to filter the Probe side data. IN: Builds a value list for the JOIN KEY of the JOIN BUILD side to filter the Probe side data. Empty string: Indicates disabling the Runtime Filter feature. | Global & Session | Yes |
libra_enable_cost_based_runtime_filter | Enable/Disable cost-based Runtime Filter. | BOOL | ON | ON: Enable cost-based Runtime Filter assignment. OFF: Disable cost-based Runtime Filter assignment. | Global & Session | Yes |
libra_max_in_runtime_filter_ndv | The maximum value of NDV (Number of Distinct Values) on the BUILD side when an IN type Runtime Filter is generated in cost-based Runtime Filters. | INT | 1024 | 0-MaxValue | Global & Session | Yes |
runtime_filter_wait_time_ms | The maximum time to initially wait for RF Ready on the RF USE side. 0 indicates waiting indefinitely until timeout or interrupting query execution. | INT | 100 | [0, MaxValue) | Global & Session | Yes |
System Variable | Meaning | Parameter Type | Default Value | Value Range | Scope | Supporting SET_VAR Hint |
libra_enable_late_materialization | Whether to enable late materialization. | BOOL | OFF | ON: Enable late materialization. OFF: Disable late materialization. | Global & Session | Yes |
System Variable | Meaning | Parameter Type | Default Value | Value Range | Scope | Supporting SET_VAR Hint |
max_threads | Parallelism during query execution. | INT | Number of the CPU cores of the node | 0-MaxValue (recommended not to exceed twice the number of CPU cores) | Global & Session | Yes |
System Variable | Meaning | Parameter Type | Default Value | Value Range | Scope | Supporting SET_VAR Hint |
max_bytes_before_external_agg_uniq_exact | The limit of the hashset when the uniqExactDisk calculation function is used for CountDistinct. If the hashset exceeds this limit, switching to external memory will be started. The default is 0, indicating no external memory computing. | INT | 0 | [0, MaxValue) | Global & Session | Yes |
one_bucket_max_temp_file_size | The maximum size of each file written to the bucket when the uniqExactDisk calculation function is used for CountDistinct. If this value is exceeded, switching to a new file for writing is required. | INT | 0 | [0, MaxValue) | Global & Session | Yes |
count_distinct_implementation | The name of the calculation function used for CountDistinct. The default is uniqExactDisk. | VARCHAR | uniqExactDisk | uniqExactDisk: Indicates a precise CountDistinct calculation method with the writing feature. uniq: Indicates an approximate CountDistinct calculation method. uniqExact: Indicates a precise CountDistinct calculation method with the pure memory. | - | Yes |
join_build_concurrency | Specifies the parallelism used for Join Build. The default is 0, indicating that the Default or specified MaxThreads is used. | INT | 0 | [0, MaxValue) | Global & Session | Yes |
enable_local_tunnel | Supports communication optimization for local Sender and Receiver. | BOOL | ON | ON: Enable communication optimization for Local Channel. OFF: Disable communication optimization for Local Channel. | Global & Session | Yes |
mpp_max_packet_size | The maximum size of a packet sent by the Sender. 0 indicates no limit. | INT | 0 | [0, MaxValue) | Global & Session | Yes |
tunnel_buffer_size | The BUFFER quantity of the send buffer queue on the Sender side. The default is 1. | INT | 1 | [0, MaxValue) | Global & Session | Yes |
pipeline_executor_use_thread_manager | Use a dynamic thread pool model for the Pipeline model. | BOOL | OFF | ON: Enable using a dynamic thread pool for the Pipeline model. OFF: Disable using a dynamic thread pool for the Pipeline model. | Global & Session | Yes |
enable_order_by_push_down | Enable/Disable the ORDER BY push-down feature. | BOOL | ON | ON: Enable the ORDER BY push-down feature. OFF: Disable the ORDER BY push-down feature. | Global & Session | Yes |
max_block_size | Set the number of rows in the data block passed in the Pipeline execution engine. | INT | 65409 | [1, MaxValue) | Global & Session | Yes |
shuffle_after_join | Set whether to shuffle data to multiple Pipelines after join. | BOOL | ON | ON: Enable the shuffle feature. OFF: Disable the shuffle feature. | Global & Session | Yes |
disable_mysql_compatible | Whether to add Cast in expression calculation for MySQL compatibility. | BOOL | TRUE | TRUE: Not compatible with MySQL. FALSE: Compatible with MySQL. | Global & Session | Yes |
Was this page helpful?