Use LogView V2.0 to view job execution details - MaxCompute

LogView records and displays the status of MaxCompute jobs. Open the LogView URL after submitting a job to monitor progress, inspect execution details, and diagnose failures. The interactive directed acyclic graph (DAG), execution playback, and Fuxi Sensor resource charts give you the visibility to identify slow tasks, locate failed instances, and analyze CPU or memory bottlenecks.

Access LogView

When you submit a MaxCompute job, the system automatically generates a URL starting with https://logview.aliyun.com/logview.

MaxCompute client: Copy the URL and paste it into a browser.
DataWorks: Click the URL directly to open the LogView page.

The LogView page has three sections:

No.	Section
1	Title and functionality section
2	Basic Info section
3	Job details section

Title and functionality section

This section shows the job ID and job name. The job ID uniquely identifies a MaxCompute job and is generated when the job is submitted. The job name appears only when the job is submitted via an SDK.

Use the icons on the right to perform the following actions:

Icon	Action
	Open the Logview_detail.txt file containing job details. The file is saved to your computer.
	Return to the LogView V1.0 page.
	Save job details as a file to your computer.

Basic Info section

The Basic Info section shows metadata about the job.

Parameter	Description
MaxCompute Service	The endpoint of MaxCompute on which the job runs. For more information, see Endpoints.
Project	The MaxCompute project the job belongs to.
Cloud account	The Alibaba Cloud account used to submit the job.
Type	The job type. Valid values: SQL, SQLRT, LOT, XLib, CUPID, AlgoTask, and Graph.
Status	The job status. See Job status values.
Start Time	The time at which the job was submitted.
End Time	The time at which the job completed.
Latency	The total run duration.
Progress	The current completion percentage.
Priority	The job priority.
Queue	The job's position in the resource quota group queue.

Job status values

Status	Description
Waiting	The job is queued in MaxCompute but has not yet been submitted to Job Scheduler.
Running	The job is being processed by Job Scheduler.
Success	The job completed successfully.
Failed	The job failed.
Canceled	The job was canceled.
Terminated	The job is complete.

Job details section

The job details section contains multiple tabs. The table below maps each tab to its diagnostic purpose so you can go directly to the right view.

Tab	When to use
Job Details	Understand execution structure; identify slow or stuck tasks by inspecting the DAG, heat charts, and instance-level status.
Fuxi Sensor	Analyze resource consumption — for example, investigate an out-of-memory (OOM) error or explain why a task runs slower than expected.
Result	View the job output on success, or the failure reason on failure.
SubStatusHistory	Understand what the job was doing at each stage and why it waited or failed.
SourceXML	Inspect raw job configuration: XML, settings, and flag values.
SQL Script	View the SQL script for the current task.
History	View the execution history of the current task.

Job Details tab

The Job Details tab is your primary view for understanding execution structure and identifying where a job is slow or stuck.

Progress chart

The progress chart in the upper part of the tab shows dependencies between Fuxi jobs, Fuxi tasks, and operators as an interactive DAG.

No.	Element
1	Breadcrumb navigation for switching between Fuxi jobs. For example, `JOB:_SQL_0_0_0_job_0` is a Fuxi job name.
2	Troubleshooting tools: Progress Chart, Input Heat Chart, Output Heat Chart, TaskTime Heart Chart, and InstanceTime Heart Chart.
3	Toolbar icons: refresh job status, zoom in or out, open MaxCompute Studio documentation, and switch to the parent level.
4	Zoom tool.
5	Fuxi tasks. See Fuxi task structure.
6	Playback controls. Click the play/stop icon to start or stop. Drag the progress bar to scrub. Start and end times appear on each side; the current playback time appears in the middle.
7	Thumbnail of the full DAG.

If the job has a single Fuxi job, the chart shows dependencies among Fuxi tasks.
If the job has multiple Fuxi jobs, the chart shows dependencies among those Fuxi jobs.

The playback feature is not available for Fuxi tasks in the Running state. AlgoTask jobs (such as Platform for AI (PAI) jobs) contain only one Fuxi task, so no progress chart is displayed. For non-SQL jobs, only Fuxi jobs and Fuxi tasks are shown.

Fuxi task structure

A MaxCompute job consists of one or more Fuxi jobs. Each Fuxi job consists of one or more Fuxi tasks. Each Fuxi task runs across one or more Fuxi instances — more instances are added as input data volume grows.

Fuxi task prefixes indicate the task type:

Prefix	Task type
M	Data scanning
R	Reduce
J	JOIN
C	Virtual node (used for branch selection only; no computation)

Task names encode their dependencies. For example:

R4_3_9 — task R4 can run only after tasks M3 and C9_3 complete.
M2_4_9_10_16 — task M2 can run only after tasks R4_3_9, C9_3, R10_1_16, and C16_1 complete.
R/W on each task node shows the number of rows read and written.

Click or right-click a task node to view operator dependencies and operator graphs for that task.

Job status panel

The lower part of the Job Details tab lists Fuxi task and instance details.

No.	Element
1	The Fuxi Jobs tab for switching between Fuxi jobs.
2	Fuxi task details. Click a Fuxi task to display information about its instances. By default, the instances of the first Fuxi task of the first Fuxi job are shown. For AlgoTask jobs and jobs running in the Cupid console, a Sensor column appears — click the sensor icon to view CPU and memory information for that instance.
3	Instance groups by status. Click the count next to Failed to inspect failed instances.
4	Individual Fuxi instances and their details.

Fuxi instance IDs follow the format M1#0_0:

The first number is the auto-increment instance ID.
The second number is the rerun count (0 means the Fuxi task fails to rerun).

Each instance exposes:

StdOut / StdErr: output and error messages, downloadable.
Debug: tools for debugging and troubleshooting.

If a Fuxi task shows the Interrupted state, the task may be rerun. In this case, the Progress column reflects only the rerun progress, so a task may show less than 100% even though the overall job succeeds. This is expected behavior.

Limits on Fuxi instances

The number of Fuxi instances is limited to 1 when any of the following conditions apply to the SQL statement:

The LIMIT clause is used.
A window function is used without the PARTITION BY KEY clause.
An aggregate function is used without the GROUP BY KEY clause.
A JOIN operation has no equi-join key.
The ORDER BY clause is used.

Fuxi Sensor tab

Use Fuxi Sensor to analyze resource consumption — for example, to investigate an out-of-memory (OOM) error or understand why a task is running slower than expected.

Fuxi Sensor shows CPU utilization and memory usage for individual Fuxi instances.

Fuxi Sensor is available in the China (Chengdu), China (Shenzhen), China (Shanghai), China (Hangzhou), China (Zhangjiakou), and China (Beijing) regions.

CPU utilization

The cpu_usage chart has two lines:

cpu_plan: the number of CPUs requested.
cpu_usage: the number of CPUs actually used.

On the y-axis, a value of 400 represents four processors. Adjust requested CPUs by changing the number of processors; the usable CPU count cannot be configured separately.

Memory usage

The mem_usage chart has two lines:

mem_plan: the amount of memory requested.
mem_usage: the amount of memory actually used.

mem_usage is the sum of two components:

Component	Description	Reclaimable
RSS (Resident Set Size)	Memory allocated after kernel page faults, used when `malloc` is called for non-file mappings.	No
PageCache	Memory the kernel uses to cache files for read and write operations (for example, log files).	Yes

The tab provides three sub-charts for memory analysis:

Memory details
RSS usage
PageCache usage

Result tab

The Result tab displays different content depending on the job outcome:

Success: the job result is shown.
Failed: the failure reason is shown.

To disable automatic result display, run:

setproject odps.forbid.fetch.result.by.bearertoken=true;

To change the result display format, set odps.sql.select.output.format:

-- Display results in CSV format
set odps.sql.select.output.format=csv;

-- Display results in plain text format
set odps.sql.select.output.format=HumanReadable;

SubStatusHistory tab

The SubStatusHistory tab shows the status progression of the current task. Use this tab to understand what the job was doing at each stage and why it may have waited or failed.

Status code	Description
Waiting for scheduling	The job was submitted and is waiting for the MaxCompute framework to schedule it. The wait is typically short.
Waiting for cluster resource	The MaxCompute framework detected insufficient resources in the Fuxi computing cluster.
Waiting for concurrent task slot	Project-level throttling was triggered. The project has a limit on the number of SQL tasks that can run simultaneously.
Waiting for data replication	The job is waiting for data replication to complete.
Waiting for execution slot	System-level throttling was triggered.
Waiting for cleaning up of previous task attempt	The job is waiting for cleanup of a previous task attempt.
Waiting for execution	The job is waiting to be distributed from the parent process queue to a child process. The wait is typically short.
Preparing for execution	The job is being distributed to a child process. If the child process is abnormal, this phase may take longer.
Task is executing	The job is running in the MaxCompute framework.
SQLTask is initializing	The SQL task is initializing.
SQLTask is compiling query	The SQL task is being compiled.
SQLTask is optimizing query	The SQL task is optimizing the query plan. For complex execution plans, this phase may take longer. If optimization takes an unusually long time, an error may occur.
SQLTask is splitting data sources	The SQL task is splitting data sources for optimization.
SQLTask is generating execution plan	The SQL task is generating the execution plan. A long duration here may indicate that data is being read from an excessive number of partitions or small files.
SQLTask is submitting execution plan	The SQL task is submitting the execution plan.
Job has been submitted	The job has been submitted to the computing cluster.
Offline Job Waiting for running	The job was submitted to the Fuxi cluster when resources appeared available, but resources were unavailable after submission. The job is waiting for resources. This status appears only once.
Offline Job is running	A running Fuxi job has insufficient resources — for example, because higher-priority jobs preempted them. Some Fuxi instances are in the `ready` state and waiting to run.
Offline Job is failed	The Fuxi job failed.
Offline Job is succeed	The Fuxi job completed successfully.
SQLTask is updating meta information	The SQL task is updating metadata and generating dynamic partitions. This process can be time-consuming.
SQLTask is finishing	The SQL task is finishing.
Online Job is cancelled by fuxi	The job in `service mode` was canceled.
Task rerun	The job is rerunning. A job in `service mode` may fail and switch to offline mode, or data may be replicated across clusters.
Online Job Waiting for running	The job in `service mode` is waiting to run.
Online Job is running	The job in `service mode` is running.
Online Job is failed	The job in `service mode` failed.
Online Job is succeed	The job in `service mode` completed successfully.
Online Job is cancelled by fuxi	The job in `service mode` was canceled.
Task key-path executing finished	The key execution path of the job is complete, but data such as DetailStatus has not yet been generated.
Task key-path is finished	The key execution path of the job is complete.
Instance key-path is finished	The key execution path of the instance is complete.
Task execution is finished	The job is complete and DetailStatus has been generated.
Instance execution is finished	The job is complete.
Execution failed	The job failed.

SourceXML tab

The SourceXML tab has three sub-tabs:

Sub-tab	Content
XML	Job levels and task information in XML format.
Settings	The current task's configuration settings.
Command	The current task's flag settings.

SQL Script tab

The SQL Script tab shows the SQL script for the current task.

History tab

The History tab shows the execution history of the current task.