[airflow] branch main updated: Fix reload gunicorn workers (#32102)

[ep...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

ephraimanierobi pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/airflow.git


The following commit(s) were added to refs/heads/main by this push:
     new f78a8363a6 Fix reload gunicorn workers (#32102)
f78a8363a6 is described below

commit f78a8363a6542a4b4c94642434424da7af7fcbbb
Author: Aleksandr Musorin <Sm...@mail.ru>
AuthorDate: Mon Aug 7 07:48:20 2023 +0200

    Fix reload gunicorn workers (#32102)
    
    * Toggle gunicorn --preload with reload_on_plugin_change
    
    Since gunicorn can't reload a new code if starts with ``--preload``
    setting, we need to check ``reload_on_plugin_change`` before set it up.
    
    Gunicorn can't reload a new code because the code is preloaded into the
    master process and worker are launched with ``fork``, they will still have
    the old code.
    
    * added warning message
    
    * Improve warning message
    
    Co-authored-by: Jed Cunningham <66...@users.noreply.github.com>
    
    * Improve warning message
    
    Co-authored-by: Jed Cunningham <66...@users.noreply.github.com>
    
    * Improve warning message
    
    Co-authored-by: Jed Cunningham <66...@users.noreply.github.com>
    
    * Minor tweak of warning message
    
    * Mention prod issue in config description
    
    ---------
    
    Co-authored-by: Jed Cunningham <66...@users.noreply.github.com>
    Co-authored-by: Tzu-ping Chung <ur...@gmail.com>
---
 airflow/cli/commands/webserver_command.py | 16 +++++++++++-----
 airflow/config_templates/config.yml       |  3 ++-
 2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/airflow/cli/commands/webserver_command.py b/airflow/cli/commands/webserver_command.py
index 5399f8ba64..17d48ad8da 100644
--- a/airflow/cli/commands/webserver_command.py
+++ b/airflow/cli/commands/webserver_command.py
@@ -422,11 +422,17 @@ def webserver(args):
 
         run_args += ["airflow.www.app:cached_app()"]
 
-        # To prevent different workers creating the web app and
-        # all writing to the database at the same time, we use the --preload option.
-        # With the preload option, the app is loaded before the workers are forked, and each worker will
-        # then have a copy of the app
-        run_args += ["--preload"]
+        if conf.getboolean("webserver", "reload_on_plugin_change", fallback=False):
+            log.warning(
+                "Setting reload_on_plugin_change = true prevents running Gunicorn with preloading. "
+                "This means the app cannot be loaded before workers are forked, and each worker has a "
+                "separate copy of the app. This may cause IntegrityError during webserver startup, and "
+                "should be avoided in production."
+            )
+        else:
+            # To prevent different workers creating the web app and
+            # all writing to the database at the same time, we use the --preload option.
+            run_args += ["--preload"]
 
         gunicorn_master_proc: psutil.Process | subprocess.Popen
 
diff --git a/airflow/config_templates/config.yml b/airflow/config_templates/config.yml
index f868a6c6ea..f61f6baca0 100644
--- a/airflow/config_templates/config.yml
+++ b/airflow/config_templates/config.yml
@@ -1400,7 +1400,8 @@ webserver:
     reload_on_plugin_change:
       description: |
         If set to True, Airflow will track files in plugins_folder directory. When it detects changes,
-        then reload the gunicorn.
+        then reload the gunicorn. If set to True, gunicorn starts without preloading, which is slower, uses
+        more memory, and may cause race conditions. Avoid setting this to True in production.
       version_added: 1.10.11
       type: boolean
       example: ~