Fix manual task trigger failing for k8s. by snjypl · Pull Request #28394 · apache/airflow

snjypl · 2022-12-15T22:06:06Z

Fix #28391 manual task trigger from UI fails for k8s executor

Manual task trigger from UI fails for k8s executor. the executor.job_id
is currently set to "manual". the task instance queued_by_job_id field
is expected to be None|Integer. this causes the filter query in
clear_not_launched_queued_tasks method in kubernetes_executor to fail
with psycopg2.errors.InvalidTextRepresentation invalid input syntax for integer: "manual" error.

setting the job_id to None fixes the issue.

Fixes: #28391

^ Add meaningful description above

Read the Pull Request Guidelines for more information.
In case of fundamental code changes, an Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in a newsfragment file, named {pr_number}.significant.rst or {issue_number}.significant.rst, in newsfragments.

arjunanan6 · 2023-01-09T11:42:08Z

@snjypl To test if this works, I've patched our installation of airflow with the changes in this pr, and unfortunately there are still problems when executing a task manually:

[2023-01-09T11:34:50.901+0000] {app.py:1741} ERROR - Exception on /run [POST]
Traceback (most recent call last):
File "/home/airflow/.local/lib/python3.9/site-packages/flask/app.py", line 2525, in wsgi_app
response = self.full_dispatch_request()
File "/home/airflow/.local/lib/python3.9/site-packages/flask/app.py", line 1822, in full_dispatch_request
rv = self.handle_user_exception(e)
File "/home/airflow/.local/lib/python3.9/site-packages/flask/app.py", line 1820, in full_dispatch_request
rv = self.dispatch_request()
File "/home/airflow/.local/lib/python3.9/site-packages/flask/app.py", line 1796, in dispatch_request
return self.ensure_sync(self.view_functions[rule.endpoint])(**view_args)
File "/home/airflow/.local/lib/python3.9/site-packages/airflow/www/auth.py", line 47, in decorated
return func(*args, **kwargs)
File "/home/airflow/.local/lib/python3.9/site-packages/airflow/www/decorators.py", line 125, in wrapper
return f(*args, **kwargs)
File "/home/airflow/.local/lib/python3.9/site-packages/airflow/utils/session.py", line 75, in wrapper
return func(*args, session=session, **kwargs)
File "/home/airflow/.local/lib/python3.9/site-packages/airflow/www/views.py", line 1822, in run
executor.start()
File "/home/airflow/.local/lib/python3.9/site-packages/airflow/executors/kubernetes_executor.py", line 529, in start
raise AirflowException("Could not get scheduler_job_id")
airflow.exceptions.AirflowException: Could not get scheduler_job_id

snjypl · 2023-01-09T22:10:53Z

@arjunanan6 can you please try the the latest code? i removed the job_id check.

also, while viewing the log you might face this issue depending on your logging configuration: #28817

arjunanan6 · 2023-01-10T07:46:30Z

@snjypl I will test it later this evening. Logging-wise, we should be okay as we use remote logging (WASB).

arjunanan6 · 2023-01-14T09:15:19Z

@snjypl Just got time now... I can't run your patch though, it fails to build:

#10 0.625 patching file airflow/utils/log/file_task_handler.py
#10 0.625 Hunk #1 FAILED at 215.
#10 0.625 1 out of 1 hunk FAILED -- saving rejects to file /tmp/rejects
#10 0.632 patching file airflow/utils/log/file_task_handler.py
#10 0.632 Hunk #1 FAILED at 215.
#10 0.633 1 out of 1 hunk FAILED -- saving rejects to file /tmp/rejects
#10 0.634 --- airflow/utils/log/file_task_handler.py
#10 0.634 +++ airflow/utils/log/file_task_handler.py
#10 0.634 @@ -215,7 +215,7 @@ def _read(self, ti: TaskInstance, try_number: int, metadata: dict[str, Any] | No
#10 0.634 selector = PodGenerator.build_selector_for_k8s_executor_pod(
#10 0.634 dag_id=ti.dag_id,
#10 0.634 task_id=ti.task_id,
#10 0.634 - try_number=ti.try_number,
#10 0.634 + try_number=try_number,
#10 0.634 map_index=ti.map_index,
#10 0.634 run_id=ti.run_id,
#10 0.634 airflow_worker=ti.queued_by_job_id,
#10 0.634 --- airflow/utils/log/file_task_handler.py
#10 0.634 +++ airflow/utils/log/file_task_handler.py
#10 0.634 @@ -215,7 +215,7 @@ def _read(self, ti: TaskInstance, try_number: int, metadata: dict[str, Any] | No
#10 0.634 selector = PodGenerator.build_selector_for_k8s_executor_pod(
#10 0.634 dag_id=ti.dag_id,
#10 0.634 task_id=ti.task_id,
#10 0.634 - try_number=try_number,
#10 0.634 + try_number=ti.try_number,
#10 0.634 map_index=ti.map_index,
#10 0.634 run_id=ti.run_id,
#10 0.634 airflow_worker=ti.queued_by_job_id,

snjypl · 2023-01-17T20:59:44Z

@snjypl Just got time now... I can't run your patch though, it fails to build:

#10 0.625 patching file airflow/utils/log/file_task_handler.py
#10 0.625 Hunk #1 FAILED at 215.
#10 0.625 1 out of 1 hunk FAILED -- saving rejects to file /tmp/rejects
#10 0.632 patching file airflow/utils/log/file_task_handler.py
#10 0.632 Hunk #1 FAILED at 215.
#10 0.633 1 out of 1 hunk FAILED -- saving rejects to file /tmp/rejects
#10 0.634 --- airflow/utils/log/file_task_handler.py
#10 0.634 +++ airflow/utils/log/file_task_handler.py
#10 0.634 @@ -215,7 +215,7 @@ def _read(self, ti: TaskInstance, try_number: int, metadata: dict[str, Any] | No
#10 0.634 selector = PodGenerator.build_selector_for_k8s_executor_pod(
#10 0.634 dag_id=ti.dag_id,
#10 0.634 task_id=ti.task_id,
#10 0.634 - try_number=ti.try_number,
#10 0.634 + try_number=try_number,
#10 0.634 map_index=ti.map_index,
#10 0.634 run_id=ti.run_id,
#10 0.634 airflow_worker=ti.queued_by_job_id,
#10 0.634 --- airflow/utils/log/file_task_handler.py
#10 0.634 +++ airflow/utils/log/file_task_handler.py
#10 0.634 @@ -215,7 +215,7 @@ def _read(self, ti: TaskInstance, try_number: int, metadata: dict[str, Any] | No
#10 0.634 selector = PodGenerator.build_selector_for_k8s_executor_pod(
#10 0.634 dag_id=ti.dag_id,
#10 0.634 task_id=ti.task_id,
#10 0.634 - try_number=try_number,
#10 0.634 + try_number=ti.try_number,
#10 0.634 map_index=ti.map_index,
#10 0.634 run_id=ti.run_id,
#10 0.634 airflow_worker=ti.queued_by_job_id,

@arjunanan6 can you please try it now?

arjunanan6 · 2023-01-18T07:27:16Z

@snjypl That solved the previous issue, and the execution is attempted now. However, manual runs still fail because the service account gets rejected from creating a pod for the execution:

[2023-01-18T07:20:26.236+0000] {kubernetes_executor.py:527} INFO - Start Kubernetes executor
[2023-01-18T07:20:26.261+0000] {kubernetes_executor.py:130} INFO - Event: and now my watch begins starting at resource_version: 0
[2023-01-18T07:20:26.395+0000] {kubernetes_executor.py:476} INFO - Found 0 queued task instances
[2023-01-18T07:20:26.438+0000] {base_executor.py:95} INFO - Adding to queue: ['airflow', 'tasks', 'run', 'x-y-z', 'X-Y', 'scheduled__2023-01-18T07:00:00+00:00', '--ignore
-all-dependencies', '--ignore-dependencies', '--local', '--pool', 'default_pool', '--subdir', 'DAGS_FOLDER/airflow-dags-sap/X-Y/x-y-z.py']
[2023-01-18T07:20:26.438+0000] {base_executor.py:215} INFO - task TaskInstanceKey(dag_id='x-y-z', task_id='x-y-z', run_id='scheduled__2023-01-18T07:00:00+00:00', try_number
=3, map_index=-1) is still running
[2023-01-18T07:20:26.508+0000] {kubernetes_executor.py:339} INFO - Creating kubernetes pod for job is TaskInstanceKey(dag_id='x-y-z', task_id='X-Y', run_id='scheduled__20
23-01-18T07:00:00+00:00', try_number=3, map_index=-1), with pod name x-y-z-78e3092210f94420bb0e98a744969f29
[2023-01-18T07:20:26.538+0000] {kubernetes_executor.py:274} ERROR - Exception when attempting to create Namespaced Pod: {
.
.
.
kubernetes.client.exceptions.ApiException: (403)
Reason: Forbidden
HTTP response headers: HTTPHeaderDict({'Audit-Id': 'ddc99dcf-9d70-4f88-8c7c-77f543879844', 'Cache-Control': 'no-cache, private', 'Content-Type': 'application/json', 'X-Content-Type-
Options': 'nosniff', 'X-Kubernetes-Pf-Flowschema-Uid': 'e7834783-2050-421a-b99e-0615f85f6e92', 'X-Kubernetes-Pf-Prioritylevel-Uid': 'e9e2e589-5d4c-442b-8568-f7bfbdbfaafd', 'Date': '
Wed, 18 Jan 2023 07:20:26 GMT', 'Content-Length': '315'})
HTTP response body: {"kind":"Status","apiVersion":"v1","metadata":{},"status":"Failure","message":"pods is forbidden: User \"system:serviceaccount:airflow-test-ns:airflow-test-webse
rver\" cannot create resource \"pods\" in API group \"\" in the namespace \"airflow-test-ns\"","reason":"Forbidden","details":{"kind":"pods"},"code":403}

Which is strange, because this SA is allowed to create pods, and other tasks are being executed by the same service account. I trimmed out the pod definition, but nothing looks particularly out of place there. Any idea why this fails only on a manual run attempt?

snjypl · 2023-01-18T11:18:10Z

@snjypl That solved the previous issue, and the execution is attempted now. However, manual runs still fail because the service account gets rejected from creating a pod for the execution:

[2023-01-18T07:20:26.236+0000] {kubernetes_executor.py:527} INFO - Start Kubernetes executor
[2023-01-18T07:20:26.261+0000] {kubernetes_executor.py:130} INFO - Event: and now my watch begins starting at resource_version: 0
[2023-01-18T07:20:26.395+0000] {kubernetes_executor.py:476} INFO - Found 0 queued task instances
[2023-01-18T07:20:26.438+0000] {base_executor.py:95} INFO - Adding to queue: ['airflow', 'tasks', 'run', 'x-y-z', 'X-Y', 'scheduled__2023-01-18T07:00:00+00:00', '--ignore
-all-dependencies', '--ignore-dependencies', '--local', '--pool', 'default_pool', '--subdir', 'DAGS_FOLDER/airflow-dags-sap/X-Y/x-y-z.py']
[2023-01-18T07:20:26.438+0000] {base_executor.py:215} INFO - task TaskInstanceKey(dag_id='x-y-z', task_id='x-y-z', run_id='scheduled__2023-01-18T07:00:00+00:00', try_number
=3, map_index=-1) is still running
[2023-01-18T07:20:26.508+0000] {kubernetes_executor.py:339} INFO - Creating kubernetes pod for job is TaskInstanceKey(dag_id='x-y-z', task_id='X-Y', run_id='scheduled__20
23-01-18T07:00:00+00:00', try_number=3, map_index=-1), with pod name x-y-z-78e3092210f94420bb0e98a744969f29
[2023-01-18T07:20:26.538+0000] {kubernetes_executor.py:274} ERROR - Exception when attempting to create Namespaced Pod: {
.
.
.
kubernetes.client.exceptions.ApiException: (403)
Reason: Forbidden
HTTP response headers: HTTPHeaderDict({'Audit-Id': 'ddc99dcf-9d70-4f88-8c7c-77f543879844', 'Cache-Control': 'no-cache, private', 'Content-Type': 'application/json', 'X-Content-Type-
Options': 'nosniff', 'X-Kubernetes-Pf-Flowschema-Uid': 'e7834783-2050-421a-b99e-0615f85f6e92', 'X-Kubernetes-Pf-Prioritylevel-Uid': 'e9e2e589-5d4c-442b-8568-f7bfbdbfaafd', 'Date': '
Wed, 18 Jan 2023 07:20:26 GMT', 'Content-Length': '315'})
HTTP response body: {"kind":"Status","apiVersion":"v1","metadata":{},"status":"Failure","message":"pods is forbidden: User \"system:serviceaccount:airflow-test-ns:airflow-test-webse
rver\" cannot create resource \"pods\" in API group \"\" in the namespace \"airflow-test-ns\"","reason":"Forbidden","details":{"kind":"pods"},"code":403}

Which is strange, because this SA is allowed to create pods, and other tasks are being executed by the same service account. I trimmed out the pod definition, but nothing looks particularly out of place there. Any idea why this fails only on a manual run attempt?

this issue is caused by airflow-webserver not haveing pod-launcher-role, you can try the fix in this PR: #29012

arjunanan6 · 2023-01-18T14:35:18Z

@snjypl Tried out your patch in 29012 as well, but still no luck. I'm still getting the same forbidden error. If you want, I can move pod creation specific comments to the discussion on #29012.

snjypl · 2023-01-18T16:40:00Z

@snjypl Tried out your patch in 29012 as well, but still no luck. I'm still getting the same forbidden error. If you want, I can move pod creation specific comments to the discussion on #29012.

yes @arjunanan6 i think, it would be better if you could open a new issue with details about your environment, deployment method etc. it will help in debugging and resolving the issue. we can link that issue to the other PR #29012 and address it there.

arjunanan6 · 2023-01-20T11:58:41Z

@snjypl A little update, I've tried out this fix in combination with #29012 locally, and still get the same forbidden error. I'll add more details on that PR since it's more relevant there.

potiuk

I thikn we need a ot more context in commit - what hapens here.

snjypl · 2023-01-23T08:08:28Z

I thikn we need a ot more context in commit - what hapens here.

i have added a more descriptive commit message.

Manual task trigger from UI fails for k8s executor. the executor.job_id is currently set to "manual". the task instance queued_by_job_id field is expected to be None|Integer. this causes the filter query in clear_not_launched_queued_tasks method in kubernetes_executor to fail with psycopg2.errors.InvalidTextRepresentation invalid input syntax for integer: "manual" error. setting the job_id to None fixes the issue.

Manual task trigger from UI fails for k8s executor. the executor.job_id is currently set to "manual". the task instance queued_by_job_id field is expected to be None|Integer. this causes the filter query in clear_not_launched_queued_tasks method in kubernetes_executor to fail with psycopg2.errors.InvalidTextRepresentation invalid input syntax for integer: "manual" error. setting the job_id to None fixes the issue. (cherry picked from commit 9510043)

boring-cyborgBot added area:CLI area:webserver Webserver related Issues labels Dec 15, 2022

snjypl changed the title ~~Fix manual trigger failing for k8s.~~Fix manual task trigger failing for k8s.Dec 15, 2022

snjypl mentioned this pull request Dec 15, 2022
Manual task trigger fails for kubernetes executor with psycopg2 InvalidTextRepresentation error #28391
Closed
2 tasks

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch from bf09d07 to cacec62Compare January 6, 2023 22:09

snjypl marked this pull request as ready for review January 7, 2023 13:46

snjypl requested review from ashb, bbovenzi and ryanahamilton as code owners January 7, 2023 13:46

uranusjr approved these changes Jan 9, 2023
View reviewed changes

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch from cacec62 to d2da7c4Compare January 9, 2023 12:29

snjypl requested review from dstandish and jedcunningham as code owners January 9, 2023 21:40

eladkal added this to the Airflow 2.5.1 milestone Jan 11, 2023

ephraimbuddy modified the milestones: Airflow 2.5.1, Airflow 2.5.2 Jan 12, 2023

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch 2 times, most recently from ddf4336 to 44640d0Compare January 13, 2023 19:56

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch 3 times, most recently from bbd5c69 to 653083aCompare January 17, 2023 20:57

arjunanan6 mentioned this pull request Jan 20, 2023
Fix airflow-webserver cannot create resource pods for k8s exceutor deployed using helm chart #29012
Closed

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch from 653083a to 92f3ce0Compare January 20, 2023 20:43

potiuk approved these changes Jan 21, 2023
View reviewed changes

potiuk requested changes Jan 21, 2023
View reviewed changes

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch from 92f3ce0 to 4de0b0aCompare January 23, 2023 08:06

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch from 4de0b0a to aa0fedfCompare January 23, 2023 13:51

snjypl force-pushed the bugfix/28391-manual-trigger-fails-for-k8s branch from aa0fedf to dfccccdCompare January 24, 2023 11:51

potiuk approved these changes Jan 24, 2023
View reviewed changes

potiuk merged commit 9510043 into apache:mainJan 24, 2023

snjypl deleted the bugfix/28391-manual-trigger-fails-for-k8s branch January 24, 2023 15:35

pierrejeambrun added the type:bug-fix Changelog: Bug Fixes label Feb 27, 2023

pierrejeambrun mentioned this pull request Mar 10, 2023
Status of testing of Apache Airflow 2.5.2rc2 #30028
Closed
64 tasks

arjunanan6 mentioned this pull request Mar 19, 2023
Add webserver serviceaccount #30186
Closed

Navigation Menu

Uh oh!

Fix manual task trigger failing for k8s. - #28394

Fix manual task trigger failing for k8s.#28394
potiuk merged 1 commit into
apache:mainfrom
snjypl:bugfix/28391-manual-trigger-fails-for-k8s

snjypl commented Dec 15, 2022•
edited by potiuk
Loading

Uh oh!

arjunanan6 commented Jan 9, 2023•
edited
Loading

Uh oh!

snjypl commented Jan 9, 2023

Uh oh!

arjunanan6 commented Jan 10, 2023

Uh oh!

arjunanan6 commented Jan 14, 2023

Uh oh!

snjypl commented Jan 17, 2023

Uh oh!

arjunanan6 commented Jan 18, 2023

Uh oh!

snjypl commented Jan 18, 2023

Uh oh!

arjunanan6 commented Jan 18, 2023•
edited
Loading

Uh oh!

snjypl commented Jan 18, 2023•
edited
Loading

Uh oh!

arjunanan6 commented Jan 20, 2023

Uh oh!

potiuk left a comment

Uh oh!

snjypl commented Jan 23, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

Uh oh!

Conversation

snjypl commented Dec 15, 2022• edited by potiuk Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

arjunanan6 commented Jan 9, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

snjypl commented Jan 9, 2023

Uh oh!

arjunanan6 commented Jan 10, 2023

Uh oh!

arjunanan6 commented Jan 14, 2023

Uh oh!

snjypl commented Jan 17, 2023

Uh oh!

arjunanan6 commented Jan 18, 2023

Uh oh!

snjypl commented Jan 18, 2023

Uh oh!

arjunanan6 commented Jan 18, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

snjypl commented Jan 18, 2023• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

arjunanan6 commented Jan 20, 2023

Uh oh!

potiuk left a comment

Choose a reason for hiding this comment

Uh oh!

snjypl commented Jan 23, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

snjypl commented Dec 15, 2022•
edited by potiuk
Loading

arjunanan6 commented Jan 9, 2023•
edited
Loading

arjunanan6 commented Jan 18, 2023•
edited
Loading

snjypl commented Jan 18, 2023•
edited
Loading