All Products
Search

This Product

    MaxCompute:Flag parameters

    Document Center

    MaxCompute:Flag parameters

    Last Updated:Mar 28, 2026

    This topic lists the Flag parameters for MaxCompute. Search this page to find a parameter's scope, property name, and description.

    Category

    Scope

    Parameter

    Description

    Value

    Affected commands

    Ecosystem

    Session

    console.sql.result.instancetunnel

    Enables or disables InstanceTunnel. For more information about Tunnel commands, see Usage instructions.

    • True: Enabled.

    • False: Disabled.

    SELECT

    Scheduling

    Session

    odps.stage.mapper.mem

    Sets the memory size for each map worker.

    Range: 256 MiB to 12288 MiB. Default: 1024 MiB.

    ALL

    Scheduling

    Session

    odps.stage.reducer.mem

    Sets the memory size for each reduce worker.

    Range: 256 MiB to 12288 MiB. Default: 1024 MiB.

    ALL

    Scheduling

    Session

    odps.stage.joiner.mem

    Sets the memory size for each join worker.

    Range: 256 MiB to 12288 MiB. Default: 1024 MiB.

    ALL

    Scheduling

    Session

    odps.stage.mem

    Sets the memory size for all workers in a specified MaxCompute task. This property has a lower priority than odps.stage.mapper.mem, odps.stage.reducer.mem, and odps.stage.joiner.mem.

    Range: 256 MiB to 12288 MiB. No default value.

    ALL

    Scheduling

    Session

    odps.stage.mapper.split.size

    Sets the input data size for each map worker, which corresponds to the split size of the input file. This indirectly controls the number of workers in each map stage.

    Note

    • When you configure this item, avoid using Limit in SQL statements because Limit restricts a worker to a single concurrent operation.

    • You can use the Split size hint feature in MaxCompute to adjust the split size, control concurrency, and optimize computing performance.

    Unit: MiB. Default: 256 MiB.

    ALL

    Scheduling

    Session

    odps.sql.split.size

    Sets the size of a single data split for table-level parallel processing.

    Unit: MiB.

    Example: {"table1": 1024, "table2": 512}.

    ALL

    Scheduling

    Session

    odps.sql.split.row.count

    Sets the number of rows in a single split for table-level parallel processing.

    Note

    Applies only to internal, non-transactional, and non-clustered tables.

    An integer greater than 0.

    Example: {"table1": 100, "table2": 500}.

    ALL

    Scheduling

    Session

    odps.sql.split.dop

    Sets the table-level parallelism.

    Note

    Applies only to internal, non-transactional, and non-clustered tables.

    An integer greater than 0.

    Example: {"table1": 1, "table2": 5}.

    ALL

    Scheduling

    Session

    odps.stage.reducer.num

    Sets the number of workers for each reduce stage.

    Note

    When you configure this setting, avoid using Limit in SQL statements because Limit restricts a worker to a single concurrent operation.

    -

    ALL

    Scheduling

    Session

    odps.stage.joiner.num

    Sets the number of workers for each join stage.

    Note

    When configuring this item, avoid using Limit in SQL statements because Limit limits the concurrency of a worker to one.

    -

    ALL

    Scheduling

    Session

    odps.stage.num

    Sets the concurrency (number of workers) for a specified MaxCompute task. This property has a lower priority than odps.stage.mapper.split.size, odps.stage.reducer.num, and odps.stage.joiner.num.

    Note

    When you configure this setting, you should avoid using Limit in SQL statements, because Limit restricts a worker to single-threaded execution.

    -

    ALL

    Scheduling

    Project

    odps.instance.priority.enable

    Enables or disables job priority for projects that use subscription resources. For more information about the job priority feature, see Job priority.

    • True: Enabled.

    • False: Disabled.

    Note

    Only the project owner or users who have the Super_Administrator role can enable this feature.

    ALL

    SQL

    Session

    odps.sql.reshuffle.dynamicpt

    Enables or disables reshuffling for dynamic partitions to prevent the creation of too many small files.

    • True: Enabled.

    • False: Disabled.

    Note

    If only a few dynamic partitions are generated, set this property to False to prevent data skew.

    • INSERT INTO

    • INSERT OVERWRITE

    SQL

    Session

    odps.sql.udf.getjsonobj.new

    Specifies whether the GET_JSON_OBJECT function returns the original string without escaping special characters.

    Note

    For more information about the GET_JSON_OBJECT function, see String functions.

    • For MaxCompute projects created on or after January 21, 2021, the GET_JSON_OBJECT function returns the original string by default.

    • For MaxCompute projects created before January 21, 2021, the GET_JSON_OBJECT function escapes reserved JSON characters by default to avoid affecting existing jobs.

    • True: Enabled.

    • False: Disabled.

    UDF

    SQL

    Session

    odps.sql.udf.jvm.memory

    An out-of-memory (OOM) error can occur if a UDF processes large amounts of data for in-memory computing or sorting. If this happens, you can increase this property's value. However, this is only a temporary solution, and optimizing the UDF code is the recommended long-term fix.

    Range: 256 MiB to 12288 MiB. Default: 1024 MiB.

    UDF

    SQL

    Session

    odps.function.timeout

    Sets the timeout duration for a UDF.

    Range: 0s to 3600s. Default: 600s.

    UDF

    SQL

    Session

    odps.sql.session.resources

    Specifies the resources referenced by a UDT. You can specify multiple resources, separated by commas (,).

    For more information about referenced resources, see UDT overview.

    Uploaded resources.

    UDT

    SQL

    Session

    odps.sql.udt.display.tostring

    Specifies whether to wrap all columns whose final output type is a UDT with java.util.Objects.toString(...).

    • True: Enabled.

    • False: Disabled.

    UDT

    SQL

    Session

    odps.sql.session.java.imports

    Specifies the Java packages referenced by a UDT. You can specify multiple packages, separated by commas (,). For more information about Java packages, see UDT overview.

    Uploaded Java packages.

    UDT

    SQL

    Session

    odps.sql.skewjoin

    Enables or disables the SKEWJOIN feature to mitigate data skew.

    • True: Enabled.

    • False: Disabled.

    • SELECT

    • JOIN

    SQL

    Session

    odps.sql.skewinfo

    Specifies the skewed keys and their corresponding values for a SKEWJOIN operation. For more information, see Optimize data skew.

    -

    • SELECT

    • JOIN

    SQL

    Session

    odps.sql.udf.ppr.deterministic

    Specifies whether to enable partition pruning in UDFs. For more information about partition pruning, see WHERE clause (WHERE_condition).

    • True: Enabled.

    • False: Disabled.

    UDF

    SQL

    Session

    odps.sql.udf.ppr.to.subquery

    Specifies whether to ignore errors when partition pruning information cannot be backfilled. For more information about partition pruning, see WHERE clause (WHERE_condition).

    • True: Enabled.

    • False: Disabled.

    UDF

    SQL

    Session

    odps.optimizer.enable.range.partial.repartitioning

    Specifies whether to enable the shuffle remove mechanism for range-clustered tables.

    • True: Enabled.

    • False: Disabled.

    • INSERT OVERWRITE

    • CREATE TABLE

    SQL

    Session

    odps.optimizer.merge.partitioned.table

    When a query repeatedly uses the same partitioned table, enable this property to merge partitions. This action minimizes I/O on the table and improves performance.

    • True: Enabled.

    • False: Disabled.

    SELECT

    SQL

    Session

    odps.optimizer.skew.join.topk.num

    Sets the number of hot keys that the optimizer collects when it runs an aggregate operation. For more information, see SKEWJOIN HINT.

    -

    SKEWJOIN HINT

    SQL

    Session

    odps.optimizer.stat.collect.auto

    Specifies whether to enable the Freeride feature, which automatically collects column statistics for tables. For more information about optimizer statistics collection, see Optimizer.

    • True: Enabled.

    • False: Disabled.

    • CREATE TABLE

    • INSERT INTO

    • INSERT OVERWRITE

    SQL

    Session

    odps.optimizer.stat.collect.plan

    Specifies a collection plan to collect statistics on specific columns. For more information about optimizer statistics collection, see Optimizer.

    -

    • CREATE TABLE

    • INSERT INTO

    • INSERT OVERWRITE

    SQL

    Session

    odps.sql.external.net.vpc

    Specifies whether to enable VPC support for external tables. For more information, see Hologres external tables.

    • True: Enabled.

    • False: Disabled.

    CREATE TABLE

    SQL

    Session

    odps.sql.groupby.position.alias

    Specifies whether to treat integer constants in a GROUP BY clause as column ordinals from the SELECT list.

    • True: Enabled.

    • False: Disabled.

    • SELECT

    • GROUP BY

    SQL

    Session

    odps.sql.groupby.skewindata

    Specifies whether to enable the anti-skew mechanism for GROUP BY operations.

    • True: Enabled.

    • False: Disabled.

    GROUP BY

    SQL

    Session

    odps.sql.orderby.position.alias

    Specifies whether to treat integer constants in an ORDER BY clause as column ordinals from the SELECT list.

    • True: Enabled.

    • False: Disabled.

    • ORDER BY

    • SELECT

    SQL

    Session

    odps.sql.mapjoin.memory.max

    Sets the memory size for a small table loaded into memory during a MAPJOIN operation. Unit: MiB.

    Note

    This parameter is also subject to the global total memory limit at the Task level. The global upper limit on the total memory for multiple small tables within the same Task is std::min(8G, odps.sql.mapjoin.memory.max * number of small tables in the task).

    取值范围：0 MiB~8192 MiB

    JOIN

    SQL

    Session

    odps.sql.distributed.map.join.memory.max

    Sets the memory size for each HashTableBuilder instance during a MAPJOIN operation. The default value is 2048. Unit: MiB.

    Range: 0 MiB to 8192 MiB

    JOIN

    SQL

    Session

    odps.sql.python.version

    Sets the Python version for running SQL statements.

    • cp27

    • cp37

    • UDTF

    • UDAF

    SQL

    Session

    odps.sql.select.output.format

    Specifies whether to display the header row in results from the MaxCompute client. For more information about how to use the MaxCompute client, see Connect by using the client (odpscmd).

    • {""needHeader"":false,""fieldDelim"":""""}: Does not display the header information.

    • {""needHeader"":true,""fieldDelim"":""""}: Displays the header information.

    ALL

    SQL

    Session

    odps.sql.unstructured.data.oss.use.https

    Specifies whether to retrieve data over HTTPS when you access an external table. For more information about OSS external tables, see ORC external tables.

    • True: Enabled.

    • False: Disabled.

    ALL

    SQL

    Session

    odps.sql.decimal.tostring.trimzero

    Specifies whether to remove trailing zeros after the decimal point for the Decimal data type. The default value is True, which means that trailing zeros are removed by default.

    • True: Removes trailing zeros after the decimal point.

    • False: Keeps trailing zeros after the decimal point.

    CAST

    SQL

    Session

    odps.sql.unstructured.tablestore.put.row

    Specifies whether to use the PutRow write mode for Tablestore external tables.

    Note

    For more information about PutRow, see PutRow.

    • True: Enabled.

    • False: Disabled.

    ALL

    SQL

    Session

    odps.sql.unstructured.external.max.dop

    Sets the maximum concurrency for workers that access external tables.

    An integer greater than 0.

    External table queries and writes

    SQL

    Session/Project

    odps.sql.unstructured.file.pattern.black.list

    When MaxCompute SQL reads an external table from an OSS or HDFS location, it can automatically ignore the _SUCCESS file written by Spark to prevent read failures. For example, at the Session level:

    CREATE EXTERNAL TABLE IF NOT EXISTS test_oss_pt( 
  t_id INT,
  t_string STRING)
PARTITIONED BY (pt1 INT,pt2 INT)
STORED BY 'com.aliyun.odps.CsvStorageHandler' 
WITH serdeproperties (
 'odps.properties.rolearn'='acs:ram::xxxxxx:role/aliyunodpsdefaultrole'
) 
LOCATION 'oss://oss-cn-hangzhou-internal.aliyuncs.com/oss-mc-test/Demo3/';

SET odps.sql.unstructured.file.pattern.black.list=.*_SUCCESS$;
INSERT INTO test_oss_pt PARTITION (pt1=8,pt2=8) VALUES (1,'tree');
SELECT t_id, t_string FROM test_oss_pt WHERE pt1=8 AND pt2=8;

    Value: .*_SUCCESS$.

    External table queries

    SQL

    Session

    odps.meta.exttable.stats.onlinecollect

    Data from external tables resides in data lakes. Because these are open, MaxCompute cannot manage their metadata locally. Without statistics, the optimizer often generates a conservative execution plan, resulting in poor performance.

    When this feature is enabled, the optimizer can collect table statistics during query execution to identify small tables. This allows the optimizer to improve the query plan by using techniques such as hash joins, join order optimization, shuffle reduction, and shorter execution pipelines.

    True: Enabled.

    False (default): Disabled.

    External table queries

    SQL

    Project

    odps.sql.allow.fullscan

    Specifies whether to allow full table scans in a project. Full table scans are resource-intensive. Disabling this feature is recommended for better performance.

    • True: Allows full table scans.

    • False: Prohibits full table scans.

    SELECT

    SQL

    Project

    odps.table.lifecycle

    Specifies whether tables in a project must have a lifecycle configured.

    • Optional: The Lifecycle clause is optional when you create a table. If you do not configure a lifecycle for the table, the table never expires.

    • Mandatory: The Lifecycle clause is required. You must configure a lifecycle for the table.

    • Inherit: If a table is created without a specified lifecycle, its lifecycle is the value of odps.table.lifecycle.value.

    CREATE TABLE

    SQL

    Project

    odps.table.lifecycle.value

    Sets the lifecycle of a table in days.

    Range: 1 to 37231. Default: 37231.

    CREATE TABLE

    SQL

    Project

    READ_TABLE_MAX_ROW

    Sets the number of rows that a SELECT statement returns.

    Range: 1 to 10000. Default: 10000.

    SELECT

    SQL

    Project

    odps.output.field.formatter

    Sets dynamic data masking rules for SQL query results. For more information about dynamic data masking in MaxCompute, see Dynamic data masking.

    Custom dynamic data masking rules.

    SELECT

    SQL

    Project

    odps.sql.acid.table.enable

    Specifies whether to enable the ACID mechanism. For more information about ACID, see ACID semantics.

    • True: Enabled.

    • False: Disabled.

    ALL

    SQL

    Session/Project

    odps.io.oss.use.vipserver

    Specifies whether to use VIPServer to access OSS for external tables. If this property is not enabled, SLB is used. VIPServer provides higher traffic limits and supports traffic control, but it introduces stability risks, such as query failures. Assess these risks before enabling this property.

    • True: Enabled

    • False (default): Disabled

    External table queries

    SQL

    Session/Project

    odps.sql.executionengine.enable.string.to.date.full.format

    To convert a date string that includes hours, minutes, and seconds, set this property to True. The default value is False.

    • True: Enabled.

    • False: Disabled.

    CAST

    SQL

    Session/Project

    odps.sql.executionengine.enable.rand.time.seed

    If this property is set to False, the Rand function uses the current instance ID as the seed to initialize random numbers. This ensures function idempotence.

    If this property is set to True, the Rand function uses the current system time as the seed. In this case, the Rand function is not idempotent and cannot be used as a shuffle key. Rerunning the job will produce different results.

    The default value is False.

    • True: Enabled.

    • False: Disabled.

    RAND

    SQL

    Session/Project

    odps.sql.type.system.odps2

    Enables or disables data type 2.0. For more information about data type 2.0, see Data type version 2.0.

    • True: Enabled.

    • False: Disabled.

    MaxCompute 2.0 extended functions

    SQL

    Session/Project

    odps.sql.type.json.enable

    Enables or disables the JSON data type. For more information about the JSON data type, see JSON data type.

    • True: Enabled.

    • False: Disabled.

    MaxCompute JSON data type

    SQL

    Session/Project

    odps.sql.hive.compatible

    Enables or disables Hive-compatible mode. When enabled, MaxCompute supports various Hive-specific syntax, such as inputRecordReader, outputRecordReader, and Serde. For more information about compatible data types, see Hive-compatible data type version.

    • True: Enabled.

    • False: Disabled.

    ALL

    SQL

    Session/Project

    odps.sql.metering.value.max

    Sets a consumption limit for a single SQL query. For more information about consumption monitoring, see Consumption monitoring, alerting, and control.

    -

    ALL

    SQL

    Session/Project

    odps.sql.timezone

    Sets the time zone for a MaxCompute project. For more information about time zones, see Time zone configuration operations.

    -

    ALL

    SQL

    Session/Project

    odps.sql.unstructured.oss.commit.mode

    Specifies whether to enable multipart upload to write data to an OSS external table. For more information, see Write data to OSS.

    • True: Enabled.

    • False: Disabled.

    INSERT OVERWRITE

    SQL

    Session/Project

    odps.sql.groupby.orderby.position.alias

    Specifies whether to treat integer constants in GROUP BY and ORDER BY clauses as column ordinals from the SELECT list.

    Note

    For existing projects, enabling this property at the project level might affect existing jobs. Before modifying this property, verify that existing jobs still run correctly. Otherwise, set this property at the session level.

    • True: Enabled.

    • False: Disabled.

    • GROUP BY

    • ORDER BY

    • SELECT

    SQL

    Session/Project

    odps.ext.oss.orc.native

    Specifies whether to use the native ORC reader to read tables.

    • True: Enabled.

    • False: Disabled.

    SELECT

    SQL

    Session/Project

    odps.sql.job.max.time.hours

    Sets the maximum execution duration for a single job.

    Note

    To set this property at the project level, contact MaxCompute technical support by joining the DingTalk developer community (Group ID: 11782920 or use the application link).

    Range: 1 to 72. Default: 24.

    SQL jobs

    Billing

    Session

    odps.task.quota.preference.tag

    In projects that use subscription resources, this property assigns a job to a specific secondary quota group. If a job's quota tag matches a quota group's tag, the job is scheduled to that group with priority. Otherwise, the job is scheduled to the project's default quota group.

    Run the following statement to set the property.

    set odps.task.quota.preference.tag = payasyougo
    Note

    This property applies only to SQL jobs.

    tag_name is the tag of a quota group in MaxCompute Housekeeper. The quota group must be in the same region as the owner of the project to which the job belongs. tag_name can contain only letters, numbers, and underscores (_).

    ALL

    Billing

    Session

    odps.task.wlm.quota

    Specifies the quota group (by name) to use for running tasks in the current session.

    -

    ALL

    Security

    Project

    odps.forbid.fetch.result.by.bearertoken

    Specifies whether to prevent the Result tab in Logview from displaying job results. Use this property to protect data.

    • True: Prevents the display of job results.

    • False: Allows the display of job results.

    ALL

    Security

    Project

    LabelSecurity

    Specifies whether to enable the LabelSecurity mechanism. For more information about the LabelSecurity mechanism, see Label-based access control.

    • True: Enabled.

    • False: Disabled.

    ALL

    Security

    Project

    CheckPermissionUsingACL

    Specifies whether to enable the access control list (ACL) authorization mechanism. For more information about ACL-based authorization, see MaxCompute permissions.

    • True: Enabled.

    • False: Disabled.

    ALL

    Security

    Project

    CheckPermissionUsingPolicy

    Specifies whether to enable the policy-based authorization mechanism. For more information about policy-based authorization, see Policy-based access control.

    • True: Enabled.

    • False: Disabled.

    ALL

    Security

    Project

    ObjectCreatorHasAccessPermission

    Specifies whether an object creator has access permissions on the object by default.

    • True: The object creator has access permissions by default.

    • False: The object creator does not have access permissions by default.

    ALL

    Security

    Project

    ObjectCreatorHasGrantPermission

    Specifies whether an object creator has grant permissions on the object by default.

    • True: The object creator has grant permissions by default.

    • False: The object creator does not have grant permissions by default.

    ALL

    Security

    Project

    ProjectProtection

    Specifies whether to enable the data protection mechanism. For more information about the data protection mechanism, see Data protection mechanism.

    • True: Enabled.

    • False: Disabled.

    ALL

    Security

    Project

    odps.output.field.formatter

    Sets dynamic data masking rules for SQL query results. For more information about dynamic data masking, see Dynamic data masking.

    Custom dynamic data masking rules.

    SELECT

    Security

    Project

    odps.security.ip.whitelist

    Sets the IP whitelist for clients that access the project over the public internet, which is typically used for cloud service integrations. For more information about IP whitelists, see Manage IP whitelists.

    A list of IP addresses, separated by commas (,).

    ALL

    Security

    Project

    odps.security.vpc.whitelist

    Specifies the IP whitelist for accessing a project from a VPC. For more information about IP whitelists, see Manage IP whitelists.

    Format: RegionID_VPCID[IP Address]

    ALL

    Data types

    Project

    odps.sql.decimal.odps2

    Enables or disables the DECIMAL(precision, scale) data type of data type 2.0. For more information about data types, see Data type version 2.0.

    • True: Enabled.

    • False: Disabled.

    ALL

    Data storage

    Project

    odps.timemachine.retention.days

    Sets the retention period for backup data. For more information about data backup, see Local backup.

    Range in days: 0 to 30. Default: 1.

    ALL

    Metadata

    Project

    odps.schema.evolution.enable

    Enables schema evolution.

    • True: Enabled.

    • False (default): Disabled.

    Schema evolution.

    Querying flag parameters

    You can query flag parameters at the project or session level.

    • Project level: In the MaxCompute client (odpscmd), run setproject; to list all flag parameters for the project.

    • Session level: View session-level parameters in Logview.