burzum
burzum

Reputation: 53

How to properly daemonize celerybeat with django?

There are a couple of questions similar to this one, but neither one has a proper solution, nor describes the exact same problem.

Periodic tasks work fine with my config if I start celery by myself from the command line, like so:

celery --app=proj.mycelery worker -B

The problem is when I try to daemonize celery. After following this tutorial, I start the service with:

sudo /etc/init.d/celerybeat start

and it seems to start fine, but the periodic task that is set to execute every 5 seconds, just doesn't happen.

These are my celery settings inside Django's settings.py:

BROKER_URL = 'amqp://guest:guest@localhost//'
CELERY_ACCEPT_CONTENT = ['json']
CELERY_TASK_SERIALIZER = 'json'
CELERY_RESULT_SERIALIZER = 'json'

This is my /etc/default/celerybeat configuration:

# Absolute or relative path to the 'celery' command:
CELERY_BIN="/home/burzum/.pyenv/versions/old_django/bin/celery"

# App instance to use
CELERY_APP="proj.mycelery"

# Where to chdir at start.
CELERYBEAT_CHDIR="/home/burzum/repos/proj/"

# Extra arguments to celerybeat
CELERYBEAT_OPTS="--schedule=/var/run/celery/celerybeat-schedule"

export DJANGO_SETTINGS_MODULE="proj.settings"

CELERYD_CHDIR="/home/burzum/repos/proj"

The /etc/init.d/celerybeat file is the same as taken from the tutorial (this one). I just added the following line at the beginning:

export PYTHONPATH='/home/burzum/repos'

The output of /var/log/celery/beat.log is:

[2015-11-23 09:15:18,304: INFO/MainProcess] beat: Starting...
[2015-11-23 09:15:23,307: INFO/MainProcess] Scheduler: Sending due task reports.tasks.test_periodic_task (reports.tasks.test_periodic_task)
[2015-11-23 09:15:28,310: INFO/MainProcess] Scheduler: Sending due task reports.tasks.test_periodic_task (reports.tasks.test_periodic_task)

So, it looks like the periodic task is being called, but nothing is happening.

Output of sudo /etc/init.d/celerybeat status is:

celery init v10.1.
Using configuration: , /etc/default/celerybeat
celerybeat (pid 11696) is up...

Output of starting the service with sudo sh -x /etc/init.d/celerybeat start is:

+ VERSION=10.1
+ export PYTHONPATH=/home/burzum/repos
+ echo celery init v10.1.
celery init v10.1.
+ id -u
+ [ 0 -ne 0 ]
+ [ -L /etc/init.d/celerybeat ]
+ SCRIPT_FILE=/etc/init.d/celerybeat
+ basename /etc/init.d/celerybeat
+ SCRIPT_NAME=celerybeat
+ scripts=
+ test -f /etc/default/celeryd
+ EXTRA_CONFIG=/etc/default/celerybeat
+ test -f /etc/default/celerybeat
+ scripts=, /etc/default/celerybeat
+ _config_sanity /etc/default/celerybeat
+ local path=/etc/default/celerybeat
+ ls -ld /etc/default/celerybeat
+ awk {print $3}
+ local owner=root
+ ls -ld /etc/default/celerybeat+ 
cut -b 6
+ local iwgrp=-
+ ls -ld+  /etc/default/celerybeat
cut -b 9
+ local iwoth=-
+ id -u root
+ [ 0 != 0 ]
+ [ - != - ]
+ [ - != - ]
+ . /etc/default/celerybeat
+ CELERY_BIN=/home/burzum/.pyenv/versions/old_django/bin/celery
+ CELERY_APP=proj.mycelery
+ CELERYBEAT_CHDIR=/home/burzum/repos/proj/
+ CELERYBEAT_OPTS=--schedule=/var/run/celery/celerybeat-schedule
+ export DJANGO_SETTINGS_MODULE=proj.settings
+ CELERYD_CHDIR=/home/burzum/repos/proj
+ echo Using configuration: , /etc/default/celerybeat
Using configuration: , /etc/default/celerybeat
+ CELERY_BIN=/home/burzum/.pyenv/versions/old_django/bin/celery
+ DEFAULT_USER=celery
+ DEFAULT_PID_FILE=/var/run/celery/beat.pid
+ DEFAULT_LOG_FILE=/var/log/celery/beat.log
+ DEFAULT_LOG_LEVEL=INFO
+ DEFAULT_CELERYBEAT=/home/burzum/.pyenv/versions/old_django/bin/celery beat
+ CELERYBEAT=/home/burzum/.pyenv/versions/old_django/bin/celery beat
+ CELERYBEAT_LOG_LEVEL=INFO
+ CELERY_APP_ARG=
+ [ ! -z proj.mycelery ]
+ CELERY_APP_ARG=--app=proj.mycelery
+ CELERYBEAT_USER=celery
+ CELERY_CREATE_DIRS=0
+ CELERY_CREATE_RUNDIR=0
+ CELERY_CREATE_LOGDIR=0
+ [ -z  ]
+ CELERYBEAT_PID_FILE=/var/run/celery/beat.pid
+ CELERY_CREATE_RUNDIR=1
+ [ -z  ]
+ CELERYBEAT_LOG_FILE=/var/log/celery/beat.log
+ CELERY_CREATE_LOGDIR=1
+ export CELERY_LOADER
+ CELERYBEAT_OPTS=--schedule=/var/run/celery/celerybeat-schedule -f /var/log/celery/beat.log -l INFO
+ [ -n  ]
+ dirname /var/log/celery/beat.log
+ CELERYBEAT_LOG_DIR=/var/log/celery
+ dirname /var/run/celery/beat.pid
+ CELERYBEAT_PID_DIR=/var/run/celery
+ CELERYBEAT_CHDIR=/home/burzum/repos/proj/
+ [ -n /home/burzum/repos/proj/ ]
+ DAEMON_OPTS= --workdir=/home/burzum/repos/proj/
+ export PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/sbin:/sbin
+ check_dev_null
+ [ ! -c /dev/null ]
+ check_paths
+ [ 1 -eq 1 ]
+ create_default_dir /var/log/celery
+ [ ! -d /var/log/celery ]
+ [ 1 -eq 1 ]
+ create_default_dir /var/run/celery
+ [ ! -d /var/run/celery ]
+ start_beat
+ echo Starting celerybeat...
Starting celerybeat...
+ _chuid --app=proj.mycelery --schedule=/var/run/celery/celerybeat-schedule -f /var/log/celery/beat.log -l INFO --workdir=/home/burzum/repos/proj/ --detach --pidfile=/var/run/celery/beat.pid
+ su celery -c /home/burzum/.pyenv/versions/old_django/bin/celery beat --app=proj.mycelery --schedule=/var/run/celery/celerybeat-schedule -f /var/log/celery/beat.log -l INFO --workdir=/home/burzum/repos/proj/ --detach --pidfile=/var/run/celery/beat.pid
+ exit 0

Upvotes: 2

Views: 2078

Answers (3)

guettli
guettli

Reputation: 27806

I guess you have a linux server with systemd. Use systemd, not "generic init-scripts".

There are docs for systemd, but they are below the old init.d way:

http://docs.celeryproject.org/en/master/userguide/daemonizing.html#usage-systemd

Quoting the docs:

This is an example systemd file:

/etc/systemd/system/celery.service:

[Unit]
Description=Celery Service
After=network.target

[Service]
Type=forking
User=celery
Group=celery
EnvironmentFile=/etc/conf.d/celery
WorkingDirectory=/opt/celery
ExecStart=/bin/sh -c '${CELERY_BIN} multi start ${CELERYD_NODES} \
  -A ${CELERY_APP} --pidfile=${CELERYD_PID_FILE} \
  --logfile=${CELERYD_LOG_FILE} --loglevel=${CELERYD_LOG_LEVEL} ${CELERYD_OPTS}'
ExecStop=/bin/sh -c '${CELERY_BIN} multi stopwait ${CELERYD_NODES} \
  --pidfile=${CELERYD_PID_FILE}'
ExecReload=/bin/sh -c '${CELERY_BIN} multi restart ${CELERYD_NODES} \
  -A ${CELERY_APP} --pidfile=${CELERYD_PID_FILE} \
  --logfile=${CELERYD_LOG_FILE} --loglevel=${CELERYD_LOG_LEVEL} ${CELERYD_OPTS}'

[Install]
WantedBy=multi-user.target

Upvotes: 0

trianglesis
trianglesis

Reputation: 83

This is a working example, finally.

  1. Make celery_service.conf
  2. Make celery.service

Run service with Centos systemctl, for example. And that's it.

I've added useful scripts:

Upvotes: 1

trianglesis
trianglesis

Reputation: 83

I have no proper answer on this too, and read tonns of tutorials which copy&paste the same :

Note -B is meant to be used for development purposes. For production environment, you need to start celery beat separately.

And nobody really show any live example of how to achieve that. I thought option

CELERYD_NODES="beat"

Will make the trick, but it doesn't work. So only adding option -B or --beat to your "celeryd" config will bring this separate worker to real beat state.

CELERYD_OPTS="--beat --scheduler=django_celery_beat.schedulers:DatabaseScheduler"

UPD: I found, this is a documentation issue, described here: https://github.com/celery/celery/issues/4304

Here you can see an example of proper implementation of beat via systemd: https://specialistoff.net/question/238

I've checked it now - beat looks up and working, need some time for testing, anyway.

Upvotes: 0

Related Questions