Tornado based cron with web interface, JSON API and history

tornado, web, cron
pip install ereb==1.1.1


Hey, it's Ereb

PyPI version

Basically, it's easy installable cron with web interface. Just pip3 install ereb && ereb and open localhost:8888 Ereb is not based on system crond, it works with own scheduler written using Tornado. It has:

  • stderr, stdout of each task run
  • historical data about slowest and most failing tasks
  • no memory leaks (hope so, but uptime >months without memory leak is normal)
  • crond syntax
  • Slack notifications
  • Tyler The Creator as a logo


And the best thing about ereb: it really works and made already thousands task runs on our servers.

How to install

pip3 install ereb

Config flags

--datadog-agent-host             host where datadog agent is running (default
--datadog-agent-port             port where datadog agent is running (default
--datadog-metrics                send metrics to datadog (default False)
--datadog-tag                    value for special `ereb:%datadog_tag%` tag
                                   in datadog metrics, sent with every event
--history-dir                    directory for history storage (default ./var)
--notifier-config                notifier json config (default ./notifier.json)
--notifier-host                  host for links in notifications (default hostname)
--notify-to                      notifications channel (default logger)
--port                           port to listen on (default 8888)
--tasks-dir                      directory with tasks config files (default ./etc)

Generic task run

It is possible to run generic command without creating new task. POST requiest to generic_tasks/run with name, cmd and timeout params. Example:

curl -v -X POST 'http://localhost:8888/generic_tasks/run' -d '{"name": "foo", "cmd": "echo bar", "timeout": 60 }'

After that in Ereb new task with name __generic_foo will start

JSON api

get /status => info about next runs get /tasks => tasks list get /tasks/foo => extended info about foo task with recent task runs post /tasks/foo => update foo task config get /tasks/foo/task_runs/%id% => get task run %id% info for foo task get /tasks/foo/run => manual run of foo task get /tasks/foo/shutdown => kill currenly running foo task


Backend development

First you do pip3 install -r requirements.txt


python3 ereb/

UI development

For js app building you need webpack (and nodejs of course) You can use webpack-dev-server, so

cd ereb/ereb-wi
npm install
npm install -g webpack
npm install -g webpack-dev-server


And your live-reloadable app will be served on localhost:8080

After work done, you have to make production build with cd ereb/ereb-wi && webpack -p and commit changes to repo :)

Migration from crontab

Use utility

crontan -l | python3 --output_dir=./etc

to generate tasks from your crontab file. Important! Check new tasks after that!

Task config example

  "cmd": "while :; do echo 'Hit CTRL+C'; sleep 1; done",
  "cron_schedule": "* * * * *",
  "description": "",
  "enabled": true,
  "group": "",
  "timeout": 10,
  "name": "infinite_loop",
  "shell_scripts": [],
  "task_id": "infinite_loop"


To turn on notifications

cp notifier.json.example notifier.json

and make proper shell commands. Token %s in config will be replaced with your real error message.

Then use --notify-to parameter for running, for example to use slack do:

python3 --notify-to=slack

Clean history

rm -rf var/


Example monit config (for ereb installed via pip)

check process ereb pidfile /PATH/TO/EREB/tmp/
    start program = "ereb & echo $! > /PATH/TO/EREB/tmp/"
    stop program = "/bin/bash -c '/bin/kill `cat /PATH/TO/EREB/tmp/`'" with timeout 65 seconds
    group system


My web interface stopped responding

This may happen if you have a big number of tasks and/or their schedule is quite frequent. In this case history storage's task_runs table (we have sqlite3 database) gets pretty big after a while and causes task scheduler component to block the web interface's i/o loop.

There is an issue already opened up about this problem, but until it is resolved you can try cleaning up your task_runs table.

To clean up all history data for task runs older than a month, simply log into sqlite3 (assuming your database file has a standard var/ereb.db location):

sqlite3 var/ereb.db

and run a DELETE command:

DELETE FROM task_runs WHERE started_at < datetime('now', '-1 month');