Python SDK example: Configuration operation examples for typical scenarios-MaxCompute(MaxCompute)-阿里云帮助中心

PyODPS exposes all SDK configuration options through odps.options. Set any option directly on this object before running operations, and PyODPS applies it globally for that session.

import odps

# Set an option
odps.options.connect_timeout = 30

# Read the current value
print(odps.options.connect_timeout)  # 30

Configuration options

Options are grouped into three categories: general, data upload and download, and DataFrame.

General

Option	Type	Description	Default
`end_point`	String	The MaxCompute endpoint.	None
`default_project`	String	The default project.	None
`log_view_host`	String	The hostname in the Logview URL.	None
`log_view_hours`	Integer	How long Logview retains logs, in hours.	24
`local_timezone`	Boolean or pytz timezone	The time zone used for datetime values. `True` uses the local system time zone, `False` uses UTC. A pytz timezone object is also accepted.	None
`lifecycle`	Integer	The lifecycle applied to all tables.	None
`temp_lifecycle`	Integer	The lifecycle applied to temporary tables.	1
`biz_id`	String	The user ID.	None
`verbose`	Boolean	Whether to print log output.	False
`verbose_log`	Object	The log receiver.	None
`chunk_size`	Integer	The write buffer size.	1496
`retry_times`	Integer	The number of times to retry a failed request.	4
`connect_timeout`	Integer	The connection timeout.	5
`read_timeout`	Integer	The read timeout.	120
`pool_connections`	Integer	The number of connections cached in the connection pool.	10
`pool_maxsize`	Integer	The maximum size of the connection pool.	10
`completion_size`	Integer	The maximum number of items returned in object completion listings.	10
`notebook_repr_widget`	Boolean	Whether to render interactive graphs.	True
`sql.settings`	—	Global hints applied to all MaxCompute SQL runs.	None
`sql.use_odps2_extension`	Boolean	Whether to enable the MaxCompute 2.0 language extension.	False

Network and retry settings

The following options work together to control connection behavior. Set them as a group when tuning for network conditions or high-throughput workloads:

import odps

# Increase timeouts and retry count for unstable networks
odps.options.connect_timeout = 30
odps.options.read_timeout = 300
odps.options.retry_times = 6
odps.options.pool_connections = 20
odps.options.pool_maxsize = 20

Data upload and download

Option	Type	Description	Default
`tunnel.endpoint`	String	The Tunnel endpoint.	None
`tunnel.use_instance_tunnel`	Boolean	Whether to use Instance Tunnel to retrieve upload and download results.	True
`tunnel.limited_instance_tunnel`	Boolean	Whether to limit the number of records retrieved through Instance Tunnel.	True
`tunnel.string_as_binary`	Boolean	Whether to return STRING data as `bytes` instead of `unicode`.	False

DataFrame

Option	Type	Description	Default
`interactive`	Boolean	Whether DataFrame is running in an interactive environment.	Based on the measured value
`df.analyze`	Boolean	Whether to enable non-MaxCompute built-in functions.	True
`df.optimize`	Boolean	Whether to enable all DataFrame optimizations.	True
`df.optimizes.pp`	Boolean	Whether to enable predicate pushdown optimization.	True
`df.optimizes.cp`	Boolean	Whether to enable column pruning optimization.	True
`df.optimizes.tunnel`	Boolean	Whether to enable Tunnel optimization for DataFrame.	True
`df.quote`	Boolean	Whether to wrap field and table names with backticks (`` ` ``) in the MaxCompute SQL backend.	True
`df.libraries`	—	The resource name of the third-party library used for DataFrame operations.	None