A new global dashboard now exists. The dashboard contains an event graph for all projects which you have access to, as well as a list of the top events, and new events.
- API endpoints now properly send no-cache headers.
- ➕ Added a countdown to update buffers.
👍 Sentry now has support for buffering a majority of its counters via Redis. For more information, check 📚 the buffer section of the configuration documentation.
- Nearly all race conditions are now handled correctly using distributed cache locks.
👍 Sentry now maintains full support for CORS requests from external hosts. This means that clients like raven-js will now able to securely send messages to Sentry, without compromising their secret key.
- 🚚 Trusted domains were removed. These were not fully implemented.
- A new datepicker for selecting ranges of events.
- ➕ Added
Projects are now assigned to a single team, and that team may consist of many members. Additionally each team may own multiple projects. This makes it much easier to handle permissions across a single organization that has many projects.
⚡️ This update also includes initial changes to how filters work. ServerName Filter and SiteFilter are no longer specified in the
FILTERSsetting, but instead are included automatically if you load the respective
➕ Additional changes:
- 👍 Legacy data format is no longer supported (pre-Raven 1.x)
- 👍 API will now validate data before sending it into the queue, enabling better debugging for clients.
🔌 Plugins are now configurable per-project. This makes it easy to install a plugin globally and allow it to run only on selective projects.
🛠 Several minor fixes, as well as a backwards incompatible change with filters.
All filters must now accept the project argument in the initializer. The signature is now init(request, project).
This version focuses on improving membership control. It includes the ability to revoke project members access temporarily (by suspending them), as well as revoking users (outside of Sentry) via the user.is_active flag.
➕ Additionally, it brings the beginnings of an "invite user" flow, allowing you to invite users who may or may not already have accounts in the system. We plan to improve 🚀 this flow in an upcoming release to allow invitees easy registration within the system if they don't already have an account.
Several improvements and additions have been made around the administration, including 🐎 performance improvements.
If you're utilizing the queue, the installation guide now recommends you switch off the 🚚 database backend, and move to something more efficient (such as Redis).
0️⃣ Eventlet is no longer used as the default worker for gunicorn, and thus is no longer installed by default. If you're using the udp worker, or
send_fake_datayou MUST 👷 install eventlet yourself. The recommended worker class for gunicorn is now
geventas it corrects some issues with async queries in psycopg2.
The queue has been changed from direct Kombu, to utilizing Celery entirely. If you were ⬆️ already using the queue, the upgrade simply requires you to change "sentry start worker" to "sentry celeryd".