Deadline 5 Overview
Deadline 5 is the latest version of Thinkbox Software's render farm management system. It offers a variety of work flow improvements, helping you streamline your production pipeline. Updated Monitor options allow artists to save their layout presets and the new slave availability filter helps them troubleshoot their jobs faster. Farm administration is made easier with increased security for authenticating Deadline users and streamlined configuration options improve system roll out times. Click here for printable one-sheet.
Deadline 5.1 introduces several new key features to help improve usability and performance, and extends upon some of the new features we added in 5.0. These features include Shotgun integration, native Python support, running multiple Slaves on one machine, and storing Deadline statistics in a database.
New Features In Deadline 5.1
New License Server
Deadline now users the new thinkbox license server, instead of the previous frantic server. The process for setting up the new server and configuring Deadline to connect to it hasn't changed. However, you will need a new 5.1 license file. Please be sure to get your new license in place before upgrading to avoid any problems.
The thinkbox license server has auto-configuration built in. If a slave starts up and can't connect to the license server, it will send a broadcast to the network. If the thinkbox server is running, it will reply to the broadcast with the license server settings. Note that this is only support in IPv4 at the moment, but studios using IPv6 can still use Deadline's auto-configuration feature to set the license.
The Deadline/Shotgun integration enables a seamless render and review data flow. When Deadline starts a render, a version is automatically created in Shotgun with key metadata. When the render is complete, Deadline updates Shotgun with a thumbnail image, paths to frames, render stats, and playback links. Shotgun then dispatches targeted notifications with links back to the work. Studios can view versions in various contexts, create reports, and organize work into playlists for review sessions where they can quickly take notes with the Shotgun Note App.
Native Python Support
Deadline now comes bundled with a native installation of Python 2.6, and the new Python.NET integration allows Deadline to run native Python scripts and load CPython modules. Any existing IronPython script can be converted to a Python.NET script, allowing studios to integrate their custom Python libraries with ease.
Multiple Slaves on One Machine
Users can now launch and configure an arbitrary number of Slaves on a single machine. Each Slave instance can be given a unique name, and can be assigned its own list of pools and groups, which allows Slaves to work on separate jobs. Now a single high performance machine can process multiple 3D, compositing, and simulation jobs simultaneously.
Slave Stability Improvements
More system libraries are now used to gather data like CPU and memory usage, which greatly reduces the number of external processes the Slave starts up. We've also fixed the bug that could cause the Slave to run out of process handles, which prevented it from starting new processes (including the rendering process).
There was also a known issue on Linux and OSX where a Slave could grab all the tasks for a job and leave them in the rendering state with no Slave actually assigned to them. We had originally thought this was a permissions problem, but it turned out to be a file system issue. We have found a different way to move the task from the queued to the rendering state which should prevent this problem from occurring again.
Another known issue was when a network hiccup could cause the Slave to think its task had been requeued and it would move on. This would leave the task in a rendering state with no Slave actually working on it. Now, the Slave counts the number of task files it can actually "see" and compares then to the job's known task count. If the numbers are different, the Slave will assume there is a network problem because it is not seeing all the task files it should be. This should greatly reduce the chances of this problem occuring in the future.
Users can now opt to store Deadline statistics in an existing Postgres database, allowing studios to write their own queries to gather the information they need. Any existing Deadline statistics currently stored in the repository can be easily imported into the database. In addition, if the database goes offline, Deadline will write the statistics to the repository so that they can be imported into the database when it is available again.
Draft is a tool that provides simple compositing functionality. It is implemented as a Python library, which exposes functionality for use in python scripts. Draft is designed to be tightly integrated with Deadline, but it can also be used as a standalone tool.
Using Deadline's Draft plugin, artists can automatically perform simple compositing operations on rendered frames after a render job finishes. They can also convert them to a different image format, or generate Quicktimes for dailies.
Active Deadline subscribers are entitled to Draft licenses at no additional cost. Note that Draft is currently still in beta, so active Deadline subscribers can request a Draft beta license by emailing firstname.lastname@example.org.
New Features in Deadline 5.0
Simplify farm administration by configuring slave settings like the repository path and license server in a single location. If the repository path or license server changes, just update the auto configuration settings instead of applying the changes to each machine individually.
New User System:
Administration features are optionally password protected and all modifications to jobs, tasks and slaves are logged and easily traceable. Deadline’s Access Control System prevents users from inadvertently disrupting other rendering tasks. Additionally, administrators can prevent user impersonation by configuring Deadline to use the system user as the Deadline user for even greater security.
Slave Availability Filter:
Allow users to quickly diagnose a failed job. Click on the job to filter the slave list and display those slaves available to render the selected job. This feature factors in Pools, Groups, White/Blacklisted slaves, and any Limit Groups the job uses.
Flexible licensing model:
Bump up the size of your farm to handle larger loads during crunch time. Our Surge Rental Program makes it easy for users to quickly increase the capacity of their render farm on a weekly basis at affordable rates.
Create plug-ins to have Deadline update in-house pipeline tools such as databases when jobs start and finish rendering. Write custom scripts to submit jobs to the farm or to perform custom pre/post rendering tasks as the job is rendered. All scripts can make use of Deadline’s Iron Python script-based SDK, resulting in endless possibilities.
Increased accuracy for slave and repository statistics when specifying any given time frame. These statistics can now be displayed in a new detailed graphing system that offers improved controls for viewing and saving graph images to disk.
This feature allows companies with multiple facilities to improve performance and latency when controlling and managing massive global render farms remotely.
Updated for iPhone and Now Android: makes it easy to keep tabs on your render farm by letting you monitor your Deadline jobs from anywhere. With the simple touch of a screen, you can view your render jobs in real time and check on their status within seconds.