Docker Swarm operator is broken: json: cannot unmarshal string into Go struct field Resources.MemoryBytes of type int64 #24821

jnunezgts · 2020-10-17T14:47:38Z

jnunezgts
Oct 17, 2020

Apache Airflow version:
1.10.12, using SQLLite as the backend

Kubernetes version (if you are using kubernetes) (use kubectl version):
N/A. Using Docker Swarm 19.03.8

Environment:

Cloud provider or hardware configuration:
No cloud, bare-metal server:

HP ProLiant DL560 Gen8, BIOS P77 12/20/2013, 64 cpus

OS (e.g. from /etc/os-release):

Fedora release 29 (Twenty Nine)

Kernel (e.g. uname -a):

Linux server.company.com 4.19.82-1300.fc29.x86_64 #1 SMP Fri Nov 8 10:49:58 EST 2019 x86_64 x86_64 x86_64 GNU/Linux

Install tools:

pip

Others:

Python 3.7.2 (default, Jan 16 2019, 19:49:22)
[GCC 8.2.1 20181215 (Red Hat 8.2.1-6)] on linux

Docker info:

Client:
 Debug Mode: false

Server:
 Containers: 21
  Running: 0
  Paused: 0
  Stopped: 21
 Images: 12
 Server Version: 19.03.8
 Storage Driver: overlay2
  Backing Filesystem: <unknown>
  Supports d_type: true
  Native Overlay Diff: true
 Logging Driver: json-file
 Cgroup Driver: systemd
 Plugins:
  Volume: local
  Network: bridge host ipvlan macvlan null overlay
  Log: awslogs fluentd gcplogs gelf journald json-file local logentries splunk syslog
 Swarm: active
  NodeID: j0pl320hoxuqcaaa14z2znvgo
  Is Manager: true
  ClusterID: kpgz783mpw8aapdxchtwdu2ff
  Managers: 1
  Nodes: 4
  Default Address Pool: 10.0.0.0/8
  SubnetSize: 24
  Data Path Port: 4789
  Orchestration:
   Task History Retention Limit: 5
  Raft:
   Snapshot Interval: 10000
   Number of Old Snapshots to Retain: 0
   Heartbeat Tick: 1
   Election Tick: 10
  Dispatcher:
   Heartbeat Period: 5 seconds
  CA Configuration:
   Expiry Duration: 3 months
   Force Rotate: 0
  Autolock Managers: false
  Root Rotation In Progress: false
  Node Address: 172.29.248.55
  Manager Addresses:
   172.29.248.55:2377
 Runtimes: runc
 Default Runtime: runc
 Init Binary: docker-init
 containerd version: 7ad184331fa3e55e52b890ea95e65ba581ae3429
 runc version: dc9208a3303feef5b3839f4323d9beb36df0a9dd
 init version: fec3683
 Security Options:
  seccomp
   Profile: default
 Kernel Version: 4.19.82-1300.fc29.x86_64
 Operating System: Fedora 29 (Twenty Nine)
 OSType: linux
 Architecture: x86_64
 CPUs: 64
 Total Memory: 125.9GiB
 Name: server.company.com
 ID: 7ESU:O253:JGNS:YJIY:XXX:CYTI:WFQC:6L5C:XXXX:62IO:VH23:XXXX
 Docker Root Dir: /opt/docker
 Debug Mode: false
 HTTP Proxy: http://proxy.company.com:8080/
 HTTPS Proxy: http://proxy.company:8080/
 No Proxy: localhost,127.0.0.1,server.company.com,.company.com
 Registry: https://index.docker.io/v1/
 Labels:
 Experimental: false
 Insecure Registries:
  privatereg.company.com:5000
  localhost:5000
  server.company.com:5000
  127.0.0.0/8
 Live Restore Enabled: false

What happened:

Created the following DAG to schedule a one time shot job:

from datetime import time
from datetime import datetime
from datetime import timedelta
from airflow import DAG
from airflow.contrib.operators.docker_swarm_operator import DockerSwarmOperator

  DEFAULT_ARGS = {
      'retry_delay': timedelta(minutes=5),
      'retries': 1,
      'email_on_failure': True,
      'email_on_retry': False,
      'email': ['[email protected]']
  }

with DAG('24_7_box', description='24 x 7. With retries', default_args=DEFAULT_ARGS, schedule_interval='0 * * * Mon-Sun', start_date=datetime(2019, 7, 23), max_active_runs=1, catchup=False) as twenty_four_by_seven_dag:
      # See:
      # https://airflow.apache.org/docs/stable/_api/airflow/contrib/operators/docker_swarm_operator/index.html
      # https://airflow.apache.org/docs/stable/_modules/airflow/contrib/operators/docker_swarm_operator.html
      SLEEP_TASK = DockerSwarmOperator(
          task_id="SLEEP_TASK",
          image="fedora:29",
          api_version="auto",
          command="/bin/sleep 60",
          docker_url="unix://var/run/docker-sysavtbuild.sock",
          force_pull=False,
          mem_limit="500m",
          auto_remove=True,
      )

      SLEEP_TASK

What you expected to happen:

I was expecting the container to be created and be alive for 60 seconds, exit with code=0 after that. No ouput.

Other have reported success in the past using Docker Swarm Operator.

Not sure. The Airflow log shows the following:

[2020-10-17 09:46:58,475] {taskinstance.py:1150} ERROR - 400 Client Error: Bad Request ("json: cannot unmarshal string into Go struct field Resources.MemoryBytes of type int64")

I can run this command from docker CLI as follows:

[user@server dags]$ docker run --rm --detach fedora:29 /bin/sleep 45
29912c34f43e2dfa20d417cb80113059a183518b99215609c0aa7b37874c27db
[user@server dags]$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED             STATUS              PORTS               NAMES
29912c34f43e        fedora:29           "/bin/sleep 45"     7 seconds ago       Up 6 seconds                            gifted_pare

How to reproduce it:

Copy the DAG provided into ~/airflow/dags
Turn ON the DAG
Trigger the DAG or let the scheduler run it. Error will show up eventually

Anything else we need to know:

Airflow.log

*** Reading local file: /home/user/airflow/logs/avt_24_7_box/SLEEP_TASK/2020-10-17T13:24:26.101897+00:00/2.log [2020-10-17 09:46:58,312] {taskinstance.py:670} INFO - Dependencies all met for [2020-10-17 09:46:58,321] {taskinstance.py:670} INFO - Dependencies all met for [2020-10-17 09:46:58,321] {taskinstance.py:880} INFO - -------------------------------------------------------------------------------- [2020-10-17 09:46:58,321] {taskinstance.py:881} INFO - Starting attempt 2 of 2 [2020-10-17 09:46:58,321] {taskinstance.py:882} INFO - -------------------------------------------------------------------------------- [2020-10-17 09:46:58,328] {taskinstance.py:901} INFO - Executing on 2020-10-17T13:24:26.101897+00:00 [2020-10-17 09:46:58,335] {standard_task_runner.py:54} INFO - Started process 35637 to run task [2020-10-17 09:46:58,371] {standard_task_runner.py:77} INFO - Running: ['airflow', 'run', '24_7_box', 'SLEEP_TASK', '2020-10-17T13:24:26.101897+00:00', '--job_id', '55', '--pool', 'default_pool', '--raw', '-sd', 'DAGS_FOLDER/avt_test3.py', '--cfg_path', '/tmp/tmpaivrdhuu'] [2020-10-17 09:46:58,372] {standard_task_runner.py:78} INFO - Job 55: Subtask SLEEP_TASK [2020-10-17 09:46:58,398] {logging_mixin.py:112} INFO - Running %s on host %s server.company.com [2020-10-17 09:46:58,467] {docker_swarm_operator.py:105} INFO - Starting docker service from image fedora:29 [2020-10-17 09:46:58,475] {taskinstance.py:1150} ERROR - 400 Client Error: Bad Request ("json: cannot unmarshal string into Go struct field Resources.MemoryBytes of type int64") Traceback (most recent call last): File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/docker/api/client.py", line 259, in _raise_for_status response.raise_for_status() File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/requests/models.py", line 941, in raise_for_status raise HTTPError(http_error_msg, response=self) requests.exceptions.HTTPError: 400 Client Error: Bad Request for url: http+docker://localhost/v1.40/services/create

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/airflow/models/taskinstance.py", line 984, in run_raw_task
result = task_copy.execute(context=context)
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/airflow/operators/docker_operator.py", line 277, in execute
return self.run_image()
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/airflow/contrib/operators/docker_swarm_operator.py", line 119, in run_image
labels={'name': 'airflow%s_%s' % (self.dag_id, self.task_id)}
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/docker/utils/decorators.py", line 34, in wrapper
return f(self, *args, **kwargs)
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/docker/api/service.py", line 190, in create_service
self._post_json(url, data=data, headers=headers), True
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/docker/api/client.py", line 265, in _result
self._raise_for_status(response)
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/docker/api/client.py", line 261, in _raise_for_status
raise create_api_error_from_http_exception(e)
File "/home/user/virtualenv/airflow/lib64/python3.7/site-packages/docker/errors.py", line 31, in create_api_error_from_http_exception
raise cls(e, response=response, explanation=explanation)
docker.errors.APIError: 400 Client Error: Bad Request ("json: cannot unmarshal string into Go struct field Resources.MemoryBytes of type int64")
[2020-10-17 09:46:58,481] {taskinstance.py:1194} INFO - Marking task as FAILED. dag_id=24_7_box, task_id=SLEEP_TASK, execution_date=20201017T132426, start_date=20201017T134658, end_date=20201017T134658
[2020-10-17 09:46:58,509] {configuration.py:373} WARNING - section/key [smtp/smtp_user] not found in config
[2020-10-17 09:46:58,583] {email.py:132} INFO - Sent an alert email to ['[email protected]']
[2020-10-17 09:47:03,312] {local_task_job.py:102} INFO - Task exited with return code 1

2020-10-17T14:47:39Z

boring-cyborg[bot]
bot Oct 17, 2020

Thanks for opening your first issue here! Be sure to follow the issue template!

0 replies

jnunezgts · 2020-10-18T09:41:10Z

jnunezgts
Oct 18, 2020
Author

Hello,
I think I found the bug. If I remove completely the memory limit parameter from the DAG mem_limit="500m", the task runs without a problem. The documentation says that you can put a float value in bytes or a String that will get parsed correctly (like the Docker operator) but that's not the case here

    :param mem_limit: Maximum amount of memory the container can use.
        Either a float value, which represents the limit in bytes,
        or a string like ``128m`` or ``1g``.
    :type mem_limit: float or str

0 replies

mik-laj · 2020-10-19T12:52:14Z

mik-laj
Oct 19, 2020
Collaborator

@jnunezgts Can I assign you to this report? I am happy to help with the review.
CC: @nullhack

0 replies

jnunezgts · 2020-10-19T13:02:15Z

jnunezgts
Oct 19, 2020
Author

@jnunezgts Can I assign you to this report? I am happy to help with the review.
CC: @nullhack

Hello,
That works. You may need to do a little bit off hand holding as I haven't submitted a patch to an OpenSource project in a while. Also I need to debug this to see why the DockerOperator is happy with the flag but the SwamDockerOperator is not (I suspect is how they pass the values between them but it is a wild guess at this point).

0 replies

mik-laj · 2020-10-19T13:16:48Z

mik-laj
Oct 19, 2020
Collaborator

Don't worry. Be happy. We will definitely help you. We also have a special channel for new contributors - #airflow-how-to-pr (
)

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Docker Swarm operator is broken: json: cannot unmarshal string into Go struct field Resources.MemoryBytes of type int64 #24821

{{title}}

{{editor}}'s edit

{{editor}}'s edit

Replies: 5 comments

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

Select a reply

Docker Swarm operator is broken: json: cannot unmarshal string into Go struct field Resources.MemoryBytes of type int64 #24821

jnunezgts Oct 17, 2020

Replies: 5 comments

boring-cyborg[bot] bot Oct 17, 2020

jnunezgts Oct 18, 2020 Author

mik-laj Oct 19, 2020 Collaborator

jnunezgts Oct 19, 2020 Author

mik-laj Oct 19, 2020 Collaborator

jnunezgts
Oct 17, 2020

boring-cyborg[bot]
bot Oct 17, 2020

jnunezgts
Oct 18, 2020
Author

mik-laj
Oct 19, 2020
Collaborator

jnunezgts
Oct 19, 2020
Author

mik-laj
Oct 19, 2020
Collaborator