Sentry in Docker

Maintenance

Since Sentry now officially supports installation via Docker - this repository will no longer be updated.

More info on Sentry's Docker installation https://docs.sentry.io/server/installation/docker/.

Warning about build tags

Latest changes introduced some new build tags:

• 8.0 - current stable version (8.0.X)
• 7.7 - old stable version (7.7.4) - no longer updated
• 7.6 - old stable version (7.6.2) - no longer updated
• 7.5 - old stable version (7.5.6) - no longer updated
• 7.4 - old stable version (7.4.3) - no longer updated
• 7.3 - old stable version (7.3.2) - no longer updated
• 7.2 - old stable version (7.2.0) - no longer updated
• 7.1 - old stable version (7.1.4) - no longer updated
• 7.0 - older stable version (7.0.2) - no longer updated
• 6.4 - even older stable version (6.4.4) - no longer updated
• dev - current master on github (infrequent builds)
• latest (the default one used earlier) - is now the same as 8.0

if you want to keep your builds same as before update your Dockerfiles and change FROM slafs/sentry to FROM slafs/sentry:6.4.

This is my approach for running Sentry inside Docker. Almost everything here is configurable via environment variables (including DATABASES and CACHES settings). It can be easily configured to run with redis (cache, buffers and celery broker), postgres database, LDAP and REMOTE_USER authentication backends.

Quickstart

to run a Sentry instance with default settings (with sqlite, locmem cache and no celery) run:

docker run -d --name=sentry --volume=/tmp/sentry:/data -p 80:9000 -e SECRET_KEY=randomvalue -e SENTRY_URL_PREFIX=http://sentry.mydomain.com slafs/sentry

You can visit now http://sentry.mydomain.com (assuming sentry.mydomain.com is mapped to your docker host) and login with default credentials (username admin and password admin) and create your first team and project.

Your sqlite database file and gunicorn logs are available in /tmp/sentry directory.

Contributing

Try not to fork this repo just to create your own Docker image with some minor tweak. Please open an issue on GitHub and maybe we can include your use case directly within this image :).

You can even write a test case for your feature ;). See CONTRIBUTING.md.

Also feel free to give feedback and comments about this image in general.

Advanced usage

Copy the file with environment variables environment.example e.g cp environment.example environment and after tweaking some values run sentry like this

docker run -d --name=sentry --volume=/tmp/sentry:/data -p 80:9000 --env-file=environment slafs/sentry

ENTRYPOINT for this image is a little wrapper script around default sentry executable. Default CMD is start which runs upgrade command (initializes and upgrades the database) and creates a default administrator (superuser) and then runs a http service (as in sentry --config=... start).

All commands and args are passed to the sentry executable. This means

docker run [docker options] slafs/sentry --help

refers to running:

sentry --config=/conf/sentry.conf.py --help

inside the container.

Admin user

You can specify a username, password and email address to create an initial sentry administrator.

Add those variables to your environment file

SENTRY_ADMIN_USERNAME=slafs
SENTRY_ADMIN_PASSWORD=mysecretpass
[email protected]

See django docs for details about createsuperuser command.

The default administrator username is admin with password admin (and [email protected] as email adress).

Postgres

It is recommended to run your sentry instance with PostgreSQL database. To link your sentry instance with a postgres container you can do it like this:

0. Pull an official PostgreSQL image: docker pull postgres (if you haven't already).
1. Run postgres container (from the official Docker image): docker run -d --name=postgres_container postgres.
2. Add DATABASE_URL=postgres://postgres:@postgresdb/postgres to environment file.
3. Run sentry with linked postgres container: docker run -d --name=sentry --volume=/tmp/sentry:/data -p 80:9000 --env-file=environment --link=postgres_container:postgresdb slafs/sentry

Notice that an alias for your linked postgres container (postgresdb) is the same as a postgres host in DATABASE_URL variable.

DATABASE_URL is a value that is parsed by an external app called dj-database-url.

Redis

Redis container can be used to improve sentry performance in number of ways. It can be used as a:

• cache backend,
• sentry buffers,
• celery broker.

You can link your sentry instance with a redis container like this:

0. Pull an official Redis image: docker pull redis (if you haven't already).
1. Run redis container (from the official Docker image): docker run -d --name=redis_container redis

docker run -d --name=sentry --volume=/tmp/sentry:/data -p 80:9000 --env-file=environment --link=postgres_container:postgresdb --link=redis_container:redis slafs/sentry```

If you want to use a different container alias for redis you should add SENTRY_REDIS_HOST=your_redis_alias to environment file.

Cache with redis

To use a redis cache backend add CACHE_URL=hiredis://redis:6379/2 (make sure you won't have a trailing / in this setting) to environment file (where redis is the alias of your linked redis container). See django-cache-url docs for available formats.

Sentry buffers with redis

To use sentry update buffers with redis you must add SENTRY_USE_REDIS_BUFFERS=True to environment file.

If you have many redis containers/hosts you can set a list of those hosts in SENTRY_REDIS_BUFFERS variable so they can be used by sentry. Like this: SENTRY_REDIS_BUFFERS=redis1:6380,redis2:6381.

See sentry docs for details about redis buffer.

Celery with redis (and postgres)

To use Celery in sentry you must add CELERY_ALWAYS_EAGER=False to your environment file and run a celery worker like this:

docker run -d --name=sentry_celery_worker --link=redis_container:redis --link=postgres_container:postgresdb --volume=/tmp/sentry:/data --env-file=environment slafs/sentry celery worker -B

You can also set a different BROKER_URL via environment file by adding this: SENTRY_BROKER_URL=redis://otherredishost:6379/1

You can run as many celery worker containers as you want but remember that only one of them should be run with -B option.

If you're using default values for CELERY_RESULT_SERIALIZER, CELERY_TASK_SERIALIZER and CELERY_ACCEPT_CONTENT (default values support pickle) you have to set C_FORCE_ROOT env var (to say 1) to be able to run celery as a root user. You can avoid this by setting those three to json. Refer to Celery's serializer docs for more info about security implications of using pickle as a task serialization format.

Time-series storage with redis

To have time-series data you must add SENTRY_USE_REDIS_TSDB=True to the environment file.

By default Redis time-series storage will use the Redis connection defined by SENTRY_REDIS_HOST and SENTRY_REDIS_PORT. Similarly to configuring buffers, you can set SENTRY_REDIS_TSDBS to a list of Redis servers: SENTRY_REDIS_TSDBS=redis1:6379,redis2:6380

Email

You can configure all email settings by environment variables with SENTRY_ prefix. You have to also change an email backend and set it to SENTRY_EMAIL_BACKEND=django.core.mail.backends.smtp.EmailBackend or something similar.

LDAP

With this image You should be able to easily configure LDAP authentication for your Sentry instance. To enable it add SENTRY_USE_LDAP=True to your environment file. Then set the needed options by adding env variables with LDAP_ prefix (see the table below). LDAP authentication backend is provided by django-auth-ldap.

REMOTE_USER

To enable authentication via REMOTE_USER, add SENTRY_USE_REMOTE_USER=True to your environment file. See the AUTH_REMOTE_USER_* env variables below for further configuration.

Available environment variables

Refer to sentry documentation, django documentation, celery documentation and django-auth-ldap documentation for the meaning of each setting.

Extending the image

If your use case is out the scope of this generic image, you can extend this image to include your custom packages, configuration etc. and still be able to take advantage of all features here. All you need to do is create your own image based on this one.

For example to install and use a plugin like sentry-github and/or change a setting that isn't managed by this image you can have a following config script my_custom_settings.py:

# get all the configuration from the original image
from sentry_docker_conf import *  # noqa

# change some settings - for example the port
# (of course you can do this also by setting environment variable ``SENTRY_WEB_PORT``)
SENTRY_WEB_PORT = 5000

# configure sentry-github
GITHUB_APP_ID = 'GitHub Application Client ID'
GITHUB_API_SECRET = 'GitHub Application Client Secret'
GITHUB_EXTENDED_PERMISSIONS = ['repo']
...

and a Dockerfile similar to this:

# inherit from this image
FROM slafs/sentry

# who's the boss? :)
MAINTAINER me

# install your extra dependencies e.g. sentry-github
# (there are some dependency issues with redis)
RUN pip install -U sentry-github "redis<2.9.0"

# add your custom settings file
ADD my_custom_settings.py /conf/

# make sentry_docker_conf module importable from your custom script
ENV PYTHONPATH /conf

# use your custom settings file as a default config file for sentry to use
ENV SENTRY_CONF_FILE /conf/my_custom_settings.py

# expose the new port
EXPOSE 5000

Build your image:

docker build -t my_custom_sentry_image .

Now you can run your sentry instance just like before but with different image. So for example this:

docker run -d --name=... --volume=... -p 80:9000 -e SECRET_KEY=... -e SENTRY_URL_PREFIX=http://... slafs/sentry

becomes this:

docker run -d --name=... --volume=... -p 80:5000 -e SECRET_KEY=... -e SENTRY_URL_PREFIX=http://... my_custom_sentry_image

Image for development

If you'd like to use a preconfigured image just to be able to use it in your development process (for example to check if your application is properly logging to sentry) you can also create your own image for this.

Following Dockerfile should be OK::

# inherit from this image
FROM slafs/sentry

# who's the boss? :)
MAINTAINER me

ENV SECRET_KEY somethingsecret
ENV SENTRY_URL_PREFIX http://sentry.domain.com

ENV SENTRY_INITIAL_TEAM testteam
ENV SENTRY_INITIAL_PROJECT testproject
ENV SENTRY_INITIAL_KEY public:secret

RUN /usr/local/bin/sentry_run prepare

Now after building it (like in the above example) my_custom_sentry_image can be started without any env variables and you can use http://public:[email protected]:9000/2 as your SENTRY_DSN setting for raven (or other sentry client of your choice).