Updated on 2024-03-14 GMT+08:00

GUC Parameters

force_read_from_rw

Parameter description: Forcibly reads data from another logical cluster (reads data from the logical cluster where the table resides).

Parameter type: user

Value range: Boolean

Default value: off

Configurable or not: Configuration is not recommended.

kv_sync_up_timeout

Parameter description: Specifies the waiting timeout period for KV synchronization.

Parameter type: user

Value range: an integer ranging from 0 to 2147483647

Default value: 10min

Configurable or not: configurable

enable_cudesc_streaming

Parameter description: Specifies whether to enable cross-logical cluster access through the cudesc streaming path (obtaining cudesc and delta table data from the logical cluster where the table resides).

Parameter type: superuser

Value range: enumerated values
  • off: disables cudesc streaming.
  • on: enables cudesc streaming, including read and write.
  • only_read_on: enables cudesc streaming read.

Default value: on

Configurable or not: configurable

enable_cu_align_8k

Parameter description: Specifies whether to forcibly align CUs to 8 KB.

Parameter type: user

Value range: Boolean

Default value: off

Configurable or not: configurable

enable_cu_batch_insert

Parameter description: Specifies whether to enable batch insert for column storage.

Parameter type: user

Value range: Boolean

Default value: on

Configurable or not: configurable

enable_disk_cache

Parameter description: Specifies whether to enable file cache.

Parameter type: user

Value range: Boolean

Default value: on

Configurable or not: configurable

enable_disk_cache_recovery

Parameter description: Specifies whether the file cache can be restored when the cluster is restarted.

Parameter type: user

Value range: Boolean

Default value: on

Configurable or not: Configuration is not recommended.

disk_cache_block_size

Parameter description: Specifies the size of a single block cached in the file system, in KB.

Parameter type: postmaster

Value range: an integer ranging from 8 KB to 1 TB

Default value: 1MB

Configurable or not: configurable

disk_cache_max_size

Parameter description: Specifies the total size limit of the file system cache, in KB.

Parameter type: sighup

Value range: an integer ranging from 1 MB to 1 PB

Default value: 5GB

Configurable or not: configurable

disk_cache_max_open_fd

Parameter description: Specifies the maximum number of files that can be concurrently opened in the file system cache.

Parameter type: postmaster

Value range: an integer ranging from 0 to INT_MAX

Default value: 1000

Configurable or not: configurable

dfs_max_memory

Parameter description: Specifies the upper limit (unit: KB) of the memory used for reading and writing a foreign table.

Parameter type: user

Value range: an integer ranging from 131072 to 10485760

Default value: 256MB

Configurable: Yes

enable_aio_scheduler

Parameter description: Enables the user-mode I/O control framework. After this function is enabled, all OBS I/O requests are taken over by the user-mode I/O control framework. Also, enables asynchronous reads/writes.

Parameter type: sighup

Value range: Boolean

Default value: on

Configurable: Yes

obs_worker_pool_size

Parameter description: Specifies the maximum number of threads for the agent to perform OBS read/write operations when the user-mode I/O management and control framework is enabled.

Parameter type: postmaster

Value range: an integer ranging from 4 to 2048

Default value: 128

Configurable or not: Configuration is not recommended.

async_io_acc_max_memory

Parameter description: Queries the maximum memory (unit: KB) that can be used by the asynchronous read/write acceleration feature in a single task thread.

Parameter type: user

Value range: an integer ranging from 4096 KB to INT_MAX/2 KB

Default value: 128MB

Configurable: Yes

enable_metaversion

Parameter description: Specifies whether to enable the DN global metadata cache. After metadata is enabled on DNs, extra memory space is occupied. The memory space is controlled by local_metacache_size and global_metacache_size.

Parameter type: superuser

Value range: Boolean

Default value: off

Configurable or not: Configuration is not recommended.

local_metacache_size

Parameter description: Specifies the size of metadata cached in a local session on a DN. In extreme scenarios, if the metadata memory used by an SQL statement exceeds the value of this parameter, the SQL statement does not report an error. After the SQL statement is executed, LRU elimination is performed until the memory usage is less than the value of this parameter.

Parameter type: superuser

Value range: an integer ranging from 1024 KB to INT_MAX KB

Default value: 128MB

Configurable or not: Configuration is not recommended.

global_metacache_size

Parameter description: Specifies the size of the DN global metadata cache.

Parameter type: superuser

Value range: an integer ranging from 1024 KB to INT_MAX KB

Default value: 256MB

Configurable or not: Configuration is not recommended.

enable_metadata_partprune

Parameter description: Specifies whether to enable the metadata partition pruning function. If this parameter is enabled, DNs do not cache pruned metadata.

Parameter type: superuser

Value range: Boolean

Default value: on

Configurable or not: Configuration is not recommended.

fast_tablesize

Parameter description: Enables fast table size calculation, which may cause errors.

Parameter type: user

Value range: Boolean

Default value: off

Configurable: Yes

analyze_sample_multiplier

Parameter description: Specifies the multiplier of the ratio of the foreign table stripe sampled using ANALYZE. The value 0 indicates that the stripe ratio is 100%.

Parameter type: superuser

Value range: an integer ranging from 0 to 100

Default value: 3

Configurable or not: Configuration is not recommended.

enable_parallel_analyze

Parameter description: Specifies whether to use the parallel sampling mode when sampling internal and foreign tables using ANALYZE.

Parameter type: user

Value range: Boolean

Default value: on

Configurable: Yes

parallel_analyze_workers

Parameter description: Specifies the number of concurrent threads when the parallel ANALYZE sampling mode is used.

Parameter type: user

Value range: an integer ranging from 0 to 64

Default value: 10

Configurable: Yes

pgxc_node_readonly

Parameter description: Specifies whether a CN or DN is read-only.

Parameter type: superuser

Value range: Boolean

Default value: off

Configurable: No

enable_foreign_meta_shipping

Parameter description: Specifies whether to enable the delivery of foreign table metadata. If this parameter is enabled, the read cluster can read and write foreign tables.

Parameter type: user

Value range: Boolean

Default value: on

Configurable: Yes

enable_batchsort_heapsort_opt

Parameter description: Specifies whether to enable heap sorting optimization, which optimizes the Order By...Limit... queries.

Parameter type: user

Value range: Boolean

Default value: on

Configurable: Not recommended.

enable_batchsort_ips4o

Parameter description: Specifies whether to enable the IPS4O sorting algorithm for Batchsortstate.

Parameter type: user

Value range: Boolean

Default value: off

Configurable: not recommended.

enable_batchsort_new_sorting

Parameter description: Specifies whether to enable new sort optimization for Batchsortstate.

Parameter type: user

Value range: Boolean

Default value: on

Configurable: Not recommended.

enable_batchsort_specializations

Parameter description: Specifies whether to enable new professional optimization for Batchsortstate. If enable_batchsort_new_sorting is disabled, this parameter is invalid.

Parameter type: user

Value range: Boolean

Default value: on

Configurable: Not recommended,

force_disable_text_abbrev

Parameter description: Specifies whether to forcibly disable the Prefix Key sorting optimization.

Parameter type: user

Value range: Boolean

Default value: off

Configurable: Not recommended.

enable_insert_dop

Parameter description: Specifies whether to enable DOP during data import. If DOP is enabled, data import performance is high, but more CPU and memory resources are consumed.

Parameter type: user

Value range: Boolean

Default value: off

Configurable or not: configurable

enable_insert_ft_dop

Parameter description: Specifies whether DOP is enabled when data is exported to an OBS foreign table. If DOP is enabled, data export performance is high, but more CPU and memory resources are consumed.

Parameter type: user

Value range: Boolean

Default value: off

Configurable or not: configurable

enable_insert_ft_dop_performance

Parameter description: This parameter takes effect only when enable_insert_ft_dop is set to ON. Specifies whether to enable the high performance mode when exporting data to an OBS partitioned foreign table. If this parameter is enabled, the data export performance is high, but the memory usage increases significantly. If you can evaluate that the number of partitions in the partitioned foreign table is small and the memory resources are sufficient, you can enable this parameter. Otherwise, you are advised to disable this parameter.

Parameter type: user

Value range: Boolean

Default value: off

Configurable or not: configurable

parquet_timestamp_skip_conversion

Parameter description: Specifies whether to convert the time to the local time zone when a foreign table reads data in Parquet format and the timestamp in int96 format.

  • When this parameter is set to off:

    When an int96 timestamp is read in the parquet file, it is converted from the UTC time zone to the local time zone.

  • When this parameter is set to on:

    When an int96 timestamp is read in the parquet file, it is not converted from the UTC time zone to the local time zone.

Parameter type: user

Value range: Boolean

Default value: off

Configurable or not: configurable

parquet_enable_integer_decimal

Parameter description: Specifies the conversion rule of the decimal/numeric types when data is written to a parquet foreign table. In the decimal/numeric type definition, if the range of precision is specified, the parameter semantics are as follows:

  • When this parameter is set to off:
    • 1 ≤ precision < 39: Data is written to the fixed-length array FIXED_LEN_BYTE_ARRAY. The format is the same as decimal in Apache Hive and Apache Impala.
    • precision ≥ 39: Data is written to the variable-length array BYTE_ARRAY.
  • When this parameter is set to on:
    • 1 ≤ precision < 39: Data is written to the Int64 type.
    • 19 ≤ precision < 39: Data is written to the fixed-length array FIXED_LEN_BYTE_ARRAY.
    • precision ≥ 39: Data is written to the variable-length array BYTE_ARRAY.

Parameter type: user

Value range: Boolean

Default value: on

Configurable: Yes