Use the Migration Helper utility

Overview

The Migration Helper utility helps prepare an existing Airflow 2 project for migration to Airflow 3. It analyzes DAG source code, Airflow configuration, and project dependencies and checks whether DAGs can be imported with the target Airflow version.

Migration Helper runs against a project that is still using Airflow 2. The project itself is not upgraded during the analysis. Instead, the utility runs in a separate virtual environment and checks the project against Airflow 3 compatibility rules.

The utility uses several types of checks:

  • Ruff Airflow rules for detecting Airflow 3 compatibility issues in Python code;

  • Airflow configuration checks;

  • DAG import checks;

  • dependency compatibility checks;

  • metadata database migration checks.

Requirements

Migration Helper requires:

  • Python 3.10 or later;

  • Git available in PATH (required only for --modified-only; other checks can run without Git);

  • a dedicated Python virtual environment is recommended.

Installation

It is recommended to install Migration Helper and its dependencies into a dedicated virtual environment to keep the tool’s dependencies isolated from the system Python.

  1. Create a virtual environment:

    $ python3 -m venv .venv
  2. Activate the environment. Use the following command for Linux-like systems:

    $ source .venv/bin/activate

    On Windows, run:

    $ .venv\Scripts\activate
  3. Install Migration Helper. To install only the utility itself without Airflow (DAG import and DB migration checks will not be available, and the configuration check will use the static rule table), run:

    $ pip install airflow-migration-helper

    For the full set of checks, install Migration Helper with the airflow extra. The extra installs Airflow in the Migration Helper environment:

    $ pip install "airflow-migration-helper[airflow]"
  4. Ensure that the installation was successful. For example, by requesting help:

    $ migration-helper -h

    Command output:

    usage: migration-helper [-h] [--version] COMMAND ...
    
    Airflow 2 -> 3 migration utility.
    
    positional arguments:
      COMMAND
        analyze   Scan TARGET for Airflow 3 compatibility issues (read-only).
        fix       Scan TARGET and auto-apply fixes for Airflow 3 compatibility.
    
    options:
      -h, --help  show this help message and exit
      --version   show program's version number and exit

Migration Helper usage

Migration Helper provides two commands:

  • analyze — checks the target project in read-only mode;

  • fix — runs the checks and applies the automatic fixes currently supported by the utility.

Use the following sequence to prepare an existing DAG project for migration to Airflow 3:

  1. Install Migration Helper in an isolated environment.

  2. Configure a separate Airflow environment for DAG import checks. Do not point these checks at the production Airflow home or metadata database.

  3. Initialize a test metadata database with python -m airflow db migrate.

  4. Run the initial analysis using the analyze command.

  5. Review findings that require manual changes.

  6. Run supported automatic fixes, review the changes, and run the analysis again.

  7. Optionally check the metadata database of the Airflow 2 environment with --db-migrate.

After all detected issues have been reviewed, test the DAGs in an Airflow 3 environment. Migration Helper should be used as one part of the migration process rather than as a replacement for functional testing of the migrated DAGs.

Prepare an isolated environment for DAG import checks

The DAG import check runs Airflow from the Migration Helper environment. Airflow needs its own configuration and an initialized metadata database to run the CLI command. Configure a separate AIRFLOW_HOME and use a test database so that the check does not read or modify the production Airflow environment.

For example, from the project directory, create a dedicated Airflow home and configure a local SQLite metadata database:

$ export AIRFLOW_HOME="$PWD/.migration-helper-airflow"
$ export AIRFLOW_CONFIG="$AIRFLOW_HOME/airflow.cfg"
$ export AIRFLOW__DATABASE__SQL_ALCHEMY_CONN="sqlite:///$AIRFLOW_HOME/metadata.db"
$ python -m airflow db migrate

Run these commands in the same shell and environment in which you will run Migration Helper. The airflow db migrate command initializes or migrates the test metadata database for the installed Airflow version. Do not use a production database for this setup step.

After initializing the test environment, run the analysis against the DAG directory:

$ migration-helper analyze ./dags/

The isolated environment is intended for importing and checking DAGs with the Airflow version installed alongside Migration Helper. If you need to test imports against a different Airflow version or provider set, use an environment with those versions installed.

Quiet checks

When running the utility in CI pipelines, you can run checks in quiet mode by using --quiet or -q:

$ migration-helper analyze ./dags/ --quiet

The --quiet parameter suppresses progress and summary output. Report files are still written when --report-dir is specified.

Reports

You can generate report files using the following options:

  • --report-dir — to save the reports to a specific directory. For example:

    $ migration-helper analyze ./dags/ --report-dir ./reports/
  • --output-format — report format: json, html, or all (default). For example:

    $ migration-helper analyze ./dags/ \
      --report-dir ./reports/ \
      --output-format json

    Report files are only written when --report-dir is specified. With the default format (all), both migration_report.json and migration_report.html are created in the report directory.

The JSON report contains the following top-level fields.

Field Description

generated_at

UTC timestamp when the report was generated

target

File or directory analyzed by Migration Helper

summary

Total number of findings, separated into errors and warnings, plus the list of checkers that failed (checkers_failed)

checkers

Per-checker status, including status, reason for non-success statuses, and issue counts. Failed checkers include diagnostic details

by_file

Findings grouped by affected file

by_type

Findings grouped by rule ID

Each finding can contain the following information.

Field Description

file

Affected file or logical target, for example airflow.cfg or requirements.txt

line

Source line where the problem was detected. The value can be null when no source location is available

col

Source column where the problem was detected. The value can be null

rule_id

Rule or error identifier, for example AIR301

message

Short description of the finding

severity

Finding severity: error, warning, or info

fix_safety

Indicates whether a finding has a supported automatic fix (safe, unsafe, or null). A finding without an automatic fix is displayed as manual in the console table

fix_description

Description of the recommended fix or manual action

source

Checker that generated the finding, for example ruff, airflow_config, dag_check, dep_check, or db_migrate

Example JSON report (abridged):

{
  "generated_at": "2026-09-29T10:53:50+00:00",
  "target": "dags",
  "summary": {
    "total": 11,
    "errors": 3,
    "warnings": 8
  },
  "by_file": {
    "/home/admin/dags/example_report.py": [
      {
        "file": "/home/admin/dags/example_report.py",
        "line": 21,
        "col": 2,
        "rule_id": "AIR311",
        "message": "`airflow.decorators.dag` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.",
        "severity": "warning",
        "fix_safety": "unsafe",
        "fix_description": "`dag` has been moved to `airflow.sdk` since Airflow 3.0 (with apache-airflow-task-sdk>=1.0.0).",
        "source": "ruff"
      },
      ...
    ]
  },
  "by_type": {
    "AIR301": [ ... ],
    "AIR311": [ ... ]
  }
}

Checks for a specific version

You can select specific target Airflow and Python versions for checks. The dependency check uses Airflow 3.2.1 by default. Specify the target Airflow and Python versions explicitly when necessary:

$ migration-helper analyze ./dags/ \
  --target-airflow-version 3.2.1 \
  --target-python-version 3.10

The Python version is the major and minor version, for example 3.10 or 3.11.

Check only modified files

If a DAG project is managed by Git, you can use --modified-only to check only modified and untracked files under the target path reported by git status --porcelain --untracked-files=all:

$ migration-helper analyze <path> --modified-only

If the Git state cannot be resolved, Migration Helper checks all files instead.

Check DAGs

For migration preparation, start with the analyze command. It only checks and does not modify the analyzed files.

Example command:

$ migration-helper analyze

Example output:

Analyzing . …

File                                Line  Rule    Sev      Message                                                                                                                                          Fix
----------------------------------  ----  ------  -------  -----------------------------------------------------------------------------------------------------------------------------------------------  ------
/home/admin/dags/example_report.py  21    AIR311  warning  `airflow.decorators.dag` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.             unsafe
/home/admin/dags/example_report.py  23    AIR311  warning  `airflow.datasets.Dataset` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.           safe
/home/admin/dags/example_report.py  44    AIR311  warning  `airflow.decorators.task_group` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.      unsafe
/home/admin/dags/example_report.py  47    AIR311  warning  `airflow.decorators.task` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.            unsafe
/home/admin/dags/example_report.py  51    AIR301  error    `airflow.operators.python.get_current_context` is removed in Airflow 3.0                                                                         unsafe
/home/admin/dags/example_report.py  58    AIR311  warning  `airflow.decorators.task` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.            unsafe
/home/admin/dags/example_report.py  64    AIR301  error    `airflow.operators.python.get_current_context` is removed in Airflow 3.0                                                                         unsafe
/home/admin/dags/example_report.py  78    AIR311  warning  `airflow.decorators.task` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.            unsafe
/home/admin/dags/example_report.py  85    AIR301  error    `airflow.operators.python.get_current_context` is removed in Airflow 3.0                                                                         unsafe
/home/admin/dags/example_report.py  108   AIR311  warning  `airflow.decorators.task` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.            unsafe
/home/admin/dags/example_report.py  130   AIR311  warning  `airflow.models.baseoperator.chain` is removed in Airflow 3.0; It still works in Airflow 3.0 but is expected to be removed in a future version.  unsafe
airflow/airflow.cfg                 —     CFG001  warning  [core] min_serialized_dag_update_interval moved to [scheduler].                                                                                  safe
airflow/airflow.cfg                 —     CFG002  error    [core] executor='SequentialExecutor' removed in Airflow 3. Use 'LocalExecutor'.                                                                  manual
airflow/airflow.cfg                 —     CFG001  warning  [scheduler] job_heartbeat_sec renamed to scheduler_heartbeat_sec.                                                                                safe
airflow/airflow.cfg                 —     CFG003  warning  [logging] base_log_folder: verify path is accessible from all workers.                                                                           manual

17 issue(s): 4 error(s), 13 warning(s)

Per-checker breakdown:
  ruff            11 issue(s)
  airflow_config  6 issue(s)
  dag_check       0 issue(s)
  dep_check       0 issue(s)
  db_migrate      0 issue(s)

If the run completes with no findings at all, including warnings, and every checker completes successfully, the command confirms that you are ready to migrate:

Analyzing . …
No issues found. Ready for Airflow 3.

Check only specific DAGs

To analyze a specific DAG or a directory with DAGs, run:

$ migration-helper analyze <path>

where <path> is the directory with DAGs or a path to the DAG you want to check.

Check DAGs for import errors

The DAG import check runs the following Airflow command from the Migration Helper checker environment:

$ airflow dags list-import-errors

This check is useful for finding DAGs that cannot be imported with the Airflow version installed in the checker environment.

Example finding:

bundle_name  | filepath                   | error
=============+============================+===========================================================================================================================================
example_dags | example_asset_partition.py | Traceback (most recent call last):
             |                            |   File "<attrs generated methods airflow.sdk.definitions.dag.DAG>", line 89, in __init__
             |                            |     __attr_validator_catchup(self, __attr_catchup, self.catchup)
             |                            |   File "/usr/lib/ado_3.2.1_arenadata1/airflow/lib/python3.10/site-packages/airflow/sdk/definitions/dag.py", line 606, in _validate_catchup
             |                            |     raise ValueError("start_date is required when catchup=True")
             |                            | ValueError: start_date is required when catchup=True

Import errors are diagnostic findings. Migration Helper does not automatically rewrite the affected DAG.

NOTE

The DAG import check executes Airflow from the checker environment. Therefore, the installed Airflow version and installed providers affect the result.

Check dependency files

The dependency check verifies the project’s declared dependencies against the target Airflow and Python versions.

It is separate from the DAG source-code checks. A DAG can contain no detected Airflow 3 API problems and still have dependencies that cannot be resolved for the target environment.

By default, the target versions are:

  • Airflow 3.2.1;

  • the Python version of the interpreter running Migration Helper.

By default, the utility checks requirements.txt.

To use another dependency file, add the --dependency-file option. Requirements files with the .txt extension and pyproject.toml are supported:

$ migration-helper analyze ./dags/ \
  --dependency-file requirements-prod.txt

The check resolves the declared dependencies against the Airflow constraints files for the selected target versions, which are bundled with the utility.

To use uv instead of pip for dependency resolution checks, use the --use-uv option:

$ migration-helper analyze ./dags/ --use-uv

Check Airflow configuration

The configuration check scans Airflow configuration files for settings that are incompatible with the target Airflow version.

The check uses a built-in table of Airflow configuration compatibility rules.

When Airflow is installed in the Migration Helper environment, the utility can additionally run:

$ airflow config lint

The Airflow CLI operates on the active configuration of the Migration Helper environment, determined by AIRFLOW_HOME and AIRFLOW_CONFIG.

It does not automatically run the Airflow CLI against the airflow.cfg file found in the analyzed project.

To check a particular configuration file through the Airflow CLI path, set AIRFLOW_CONFIG to that file.

Example:

$ AIRFLOW_CONFIG=/path/to/airflow.cfg \
  migration-helper analyze ./dags/

Example report:

airflow/airflow.cfg — CFG002  error    [core] executor='SequentialExecutor' removed in Airflow 3. Use 'LocalExecutor'.

Configuration findings can therefore come from either the static configuration scan or the Airflow configuration validation performed in the checker environment.

Apply configuration fixes

NOTE

The configuration fixes are optional for ADO users. When performing a standard upgrade via ADCM, the Airflow configuration update is executed automatically as part of the bundle upgrade process.

The fix command runs the Airflow configuration update path. The --fix-config option requests that Airflow apply its recommended configuration updates in-place. In the current version, this update path is enabled automatically when fix runs, so adding --fix-config does not change the command behavior.

The file that Airflow changes is the active configuration file selected by AIRFLOW_CONFIG. If AIRFLOW_CONFIG is not set, Airflow uses the configuration file under AIRFLOW_HOME (normally $AIRFLOW_HOME/airflow.cfg). The command does not automatically modify the airflow.cfg file merely because that file is inside the project directory being analyzed.

To update a particular file, set AIRFLOW_CONFIG to its path before running the command. For example:

$ AIRFLOW_CONFIG=/path/to/airflow.cfg \
  migration-helper fix ./dags/ --fix-config

Airflow may add newly recognized configuration parameters as commented-out entries. A commented entry is a reference to an available option or its default, not an active override. Airflow may leave such entries commented when the default value is sufficient. Uncomment and set a value only when you intentionally need to override the default.

Review the resulting file and the target Airflow configuration reference before applying the updated configuration to an environment.

Metadata database checks

NOTE

The Metadata DB checks are optional for ADO users. When performing a standard upgrade via ADCM, the metadata DB migration is executed automatically as part of the bundle upgrade process.

The metadata database check is optional.

You can run it with:

$ migration-helper analyze ./dags/ --db-migrate

The check runs:

$ airflow db check-migrations

The command is read-only and does not apply database migrations.

The check does not require running Airflow processes, but the checker environment must have access to a metadata database.

By default, the check uses the database configured for the checker environment. To check the metadata database of the existing Airflow 2 cluster, configure the checker environment with the appropriate database connection, for example:

$ export AIRFLOW__DATABASE__SQL_ALCHEMY_CONN='postgresql://user:password@host:5432/airflow'
$ migration-helper analyze ./dags/ --db-migrate

Migration Helper does not apply the migrations. After reviewing the result, apply migrations separately with the airflow db migrate command. Do not run this command against a production database unless you intend to apply its migrations.

Fix unsafe findings manually

For each problem detected, the report contains an error code that identifies the type of incompatibility. The code identifies the type of check that detected the issue and is included in the JSON and HTML reports.

The Ruff checker uses the AIR codes. Migration Helper also uses codes for configuration (CFG), DAG import, dependency, and database migration (DB) checks.

Code Severity Source Description

AIR001

Warning

Ruff

The variable name assigned to an Airflow Operator does not match the Operator’s task_id

AIR002

Warning

Ruff

A DAG does not explicitly specify a schedule argument, so its scheduling behavior changes when migrated from Airflow 2 to Airflow 3

AIR301

Error

Ruff

Airflow functionality is removed in Airflow 3.0

AIR302

Error

Ruff

Airflow functionality is moved from the Airflow core package to a provider in Airflow 3.0

AIR303

Error

Ruff

An Airflow function or method has a signature that is incompatible with Airflow 3.0

AIR304

Error

Ruff

A runtime-varying value is used as an argument to a DAG or task and can cause a new DAG version to be created on every parse

AIR311

Warning

Ruff

Airflow functionality is removed in Airflow 3.0 but remains available through a compatibility layer

AIR312

Warning

Ruff

Airflow functionality is moved to a provider in Airflow 3.0 but remains available through a compatibility layer

AIR321

Error

Ruff

Airflow functionality is moved in Airflow 3.1

CFG001

Warning or error

Airflow configuration

An Airflow configuration parameter is moved to another section or renamed in the target Airflow version

CFG002

Warning or error

Airflow configuration

An Airflow configuration parameter is removed in the target Airflow version

CFG003

Warning

Airflow configuration

An Airflow configuration parameter has a changed default value or the configuration lint message does not match a more specific configuration category

CFG004

Info

Airflow configuration

A configuration change has been applied by airflow config update --all-recommendations --fix

CFG005

Warning

Airflow configuration

A configuration change is available and was reported by airflow config update --all-recommendations --dry-run

DEP_RESOLUTION_FAILED

Error

Dependencies

The declared project dependencies cannot be resolved for the selected Airflow and Python versions

DB001

Warning

Metadata database

The Airflow metadata database has pending migrations

Migration Helper reports whether a finding has a supported automatic fix. The fix_safety field in the report can have the following values:

  • safe — the checker has a known mechanical fix that can be applied safely;

  • unsafe — the checker has a possible fix, but the change requires human review;

  • null — no automatic fix is available; the issue must be handled manually.

Migration Helper supports a limited set of automatic fixes. Before applying the fixes, it is recommended to do a dry run and review the suggested changes:

$ migration-helper fix ./dags/ --safe-only --dry-run

To apply only safe fixes, run:

$ migration-helper fix ./dags/ --safe-only

To apply unsafe fixes as well, use --unsafe:

$ migration-helper fix ./dags/ --unsafe

If neither --safe-only nor --unsafe is specified, the command prints a notice and defaults to applying safe fixes only. Review all changes before deploying the DAGs: a mechanically valid code change may still alter the intended DAG behavior.

The following sections describe common unsafe findings that require manual review. The exact fix depends on the API or configuration item reported by Migration Helper. After making changes, run analyze again to verify the result.

AIR301: Removed functionality

AIR301 indicates that a function, class, attribute, import, or other Airflow functionality used by the DAG was removed in Airflow 3.0.

The DAG can no longer rely on this functionality after migration to Airflow 3. The affected API must be replaced with the Airflow 3 equivalent.

For example, an Airflow 2 DAG can use the days_ago function:

from airflow import DAG
from airflow.utils.dates import days_ago

dag = DAG(
    dag_id="example",
    start_date=days_ago(1),
)

Replace the removed API with an Airflow 3-compatible implementation. For example, use an explicit date instead of calculating the start date with days_ago:

from datetime import datetime

from airflow import DAG

dag = DAG(
    dag_id="example",
    start_date=datetime(2024, 1, 1),
)

The required change depends on the removed functionality. Check the Airflow 3 migration documentation for the corresponding replacement before changing the DAG.

AIR302: Functionality moved to a provider

AIR302 indicates that Airflow functionality was moved from the Airflow core package to a provider package.

The functionality is still available, but it must be imported from the provider that now owns it. The corresponding provider must also be installed in the Airflow environment.

For example, an Airflow 2 DAG can import FabAuthManager from the Airflow core package:

from airflow.auth.managers.fab.fab_auth_manager import FabAuthManager

auth_manager = FabAuthManager()

In Airflow 3, this functionality is provided by the FAB provider. Change the import to the provider package and make sure that the provider is included in the project dependencies:

from airflow.providers.fab.auth_manager.fab_auth_manager import FabAuthManager

auth_manager = FabAuthManager()

The required provider and import path depend on the functionality being migrated. Make sure that the provider version is compatible with the Airflow version used by the environment.

AIR303: Incompatible function signature

AIR303 indicates that an Airflow function or method used by the DAG has a different function signature in Airflow 3. The function itself may still exist, but one or more arguments used by the DAG are no longer accepted or have changed meaning.

For example, an Airflow 2 DAG can pass schedule_interval to the DAG constructor:

from airflow import DAG

dag = DAG(
    dag_id="example",
    schedule_interval="0 0 * * *",
)

In Airflow 3, the DAG API uses the schedule argument instead. A call that uses the Airflow 2 argument is therefore incompatible with the Airflow 3 signature.

Change the call to use the Airflow 3 signature:

from airflow import DAG

dag = DAG(
    dag_id="example",
    schedule="0 0 * * *",
)

The required change depends on the function and arguments used in the DAG. Do not change arguments mechanically based only on the error code. Check the Airflow 3 API for the affected function and update the call according to its new signature.

AIR304: Dynamic DAG or task value

AIR304 indicates that a value that can change when the DAG is parsed is used as an argument to a DAG or task.

Airflow parses DAG files repeatedly. Values such as the current time can therefore produce a different value on each parse. When such a value is used where Airflow expects a stable DAG or task definition, the DAG can continuously produce new versions.

For example:

from datetime import datetime

from airflow import DAG

dag = DAG(
    dag_id="example",
    start_date=datetime.now(),
)

where datetime.now() returns a different value each time the DAG file is evaluated. Use a stable value instead:

from datetime import datetime

from airflow import DAG

dag = DAG(
    dag_id="example",
    start_date=datetime(2024, 1, 1),
)

If a value must be calculated dynamically, calculate it at task execution time rather than when the DAG is parsed. Do not use a changing value to define a DAG or task property when that property is expected to remain stable between parses.

AIR311 and AIR312: Compatibility-layer functionality

AIR311 indicates that Airflow functionality was removed from its previous location but remains available through a compatibility layer. AIR312 indicates that functionality was moved to a provider but remains available through a compatibility layer.

These findings are warnings rather than immediate import failures, but a finding marked unsafe requires review because the utility cannot safely decide how to update the code in every context. Prefer importing the API from its supported Airflow 3 location and verify that the required provider or task SDK package is installed.

For example, a DAG may import the dag decorator from its compatibility location:

from airflow.decorators import dag

@dag(schedule=None)
def example():
    ...

In Airflow 3, import it from airflow.sdk:

from airflow.sdk import dag

@dag(schedule=None)
def example():
    ...

Similarly, if an operator has moved to a provider, update the import to the provider package and add the required provider to the project dependencies. Use the replacement path specified in the finding’s fix_description. Different APIs may have different destinations.

AIR321: Functionality moved in Airflow 3.1

AIR321 indicates that Airflow functionality was moved to a different location in Airflow 3.1.

The affected functionality may still be available, but the DAG must use its new location to be compatible with Airflow 3.1.

For example, an Airflow 3.0 DAG can use convert_to_utc from its previous location:

from airflow.utils.timezone import convert_to_utc

execution_time = convert_to_utc(execution_time)

In Airflow 3.1, the function was moved to the Airflow SDK. Change the import to the new location:

from airflow.sdk.timezone import convert_to_utc

execution_time = convert_to_utc(execution_time)

The required change depends on the functionality reported by the checker. Update both the import and any affected API usage according to the Airflow 3.1 API.

Configuration errors that require manual changes

Some configuration migration issues cannot be safely fixed automatically because the correct replacement depends on the deployment configuration or the intended behavior of the Airflow environment.

For configuration findings that have no automatic fix (fix_safety is null and the console shows manual), update the affected setting according to the migration message. The message identifies the old configuration option and, where available, the corresponding replacement.

For example, if the report states that SequentialExecutor was removed and recommends LocalExecutor, change the setting in the active configuration file:

[core]
executor = LocalExecutor

Do not rename configuration options based only on their names. Some Airflow 3 configuration changes also affect the meaning or default value of an option. Verify the migration rule and the Airflow configuration reference before applying the change.

When using migration-helper fix, note that Airflow updates the file selected by AIRFLOW_CONFIG, or the configuration file under AIRFLOW_HOME if AIRFLOW_CONFIG is unset. It does not necessarily update the configuration file in the analyzed project. Airflow may add new parameters as commented-out entries, which are not active overrides. Uncomment and assign a value only when you need to override the target Airflow default.

After changing the configuration, run the Migration Helper again and review the resulting report. If the utility reports another configuration issue, apply the corresponding migration rule before proceeding with the Airflow upgrade.

Verify the changes

After fixing all unsafe issues manually, run the Migration Helper again against the updated DAGs and configuration.

The migration is ready to proceed only after the report no longer contains errors that require manual intervention. Warnings may remain when the reported functionality is still supported through a compatibility layer or when the utility cannot determine whether a change is required for the specific environment.

If an error remains after the corresponding code or configuration has been changed, verify that the Migration Helper is analyzing the updated files and that the Airflow version and provider packages in the migration environment match the target Airflow environment.

Found a mistake? Seleсt text and press Ctrl+Enter to report it