In brief
- Manager settings are stored in the flamenco-manager.yaml file next to the executable.
- The Worker’s main configuration can be shared; credentials and the task update database are specific to each Worker.
- Use shared_storage_path to give the Manager and Workers access to shared job files.
- MQTT notifications can be enabled with an optional section added to the Manager configuration.
What you’ll learn in this guide
In this tutorial, you’ll configure the basics of a Blender Flamenco render farm. You’ll see where the Manager reads its settings, how Workers connect to the Manager, what the shared storage path is used for, and how to enable MQTT notifications. You’ll also learn the difference between Worker-specific credentials and the main configuration file that can be shared.
Flamenco Manager is the component that manages render jobs and Workers. Workers, in turn, execute the task types they are permitted to handle. Configuration files use YAML format. This guide focuses on configuring an existing Flamenco installation; it does not cover installation steps or make assumptions about system requirements not specified in the sources.
Requirements and file locations
The Flamenco Manager configuration file is named flamenco-manager.yaml and is located next to the Manager executable. If the file does not exist when Manager is first launched, Manager opens a setup assistant to create it. If the assistant is unavailable, you can create the configuration manually.
A Worker reads its main settings from the flamenco-worker.yaml file in its working directory. This file can be shared between multiple Workers. By contrast, flamenco-worker-credentials.yaml, which contains Worker credentials, and flamenco-worker.sqlite, which queues task updates, are specific to each Worker and are not designed for manual editing.
Before applying settings, make sure you can access the Manager and Worker configuration files. If you use shared storage, identify the path that the Manager and Workers will use to access the relevant files. The sources do not specify a particular Blender or operating system version as a general requirement. The oom_score_adjust option in the Worker configuration is available only on Linux.
Edit Manager and Worker settings step by step
Open or create the Manager configuration. You can launch Manager and let the setup assistant create the configuration file, or add
flamenco-manager.yamlnext to the executable. The example below includes basic settings as well as sample values for shared storage and the Shaman file storage system:
_meta:
version: 3
manager_name: Flamenco Manager
database: flamenco-manager.sqlite
listen: ":8080"
autodiscoverable: true
local_manager_storage_path: ./flamenco-manager-storage
shared_storage_path: /path/to/storage
shaman:
enabled: true
garbageCollect:
period: 24h
maxAge: 744hlisten sets the address and port that Manager uses for its web interface, API, and file transfers through Shaman. shared_storage_path points to the location of render outputs and .blend job files shared by the Manager and Workers. local_manager_storage_path is for files needed only by Manager, such as logs and the latest render images. Replace the /path/to/storage placeholder in the example with your actual shared storage path.
Set the Manager address and task types for Workers. You can add connection details and permitted task types to the
flamenco-worker.yamlfile in each Worker’s working directory:
manager_url: http://flamenco.local:8080/
task_types: [blender, ffmpeg, file-management, misc]
restart_exit_code: 47manager_url is the address of the Manager the Worker will connect to. If this setting is left empty or removed, the Worker will try to discover the Manager automatically on the network. The task_types list limits which task types the Worker can execute; the available task types depend on the job compiler scripts. Replace the example address with your own Manager address and assign the Worker only the types it needs to execute. When restart_exit_code is set to a non-zero value, the Worker is marked as restartable.
Set values that vary by operating system. The
variablessection in the Manager configuration lets you specify separate values for different operating systems. Under the variable name, create avalueslist; each entry contains a platform and the value for that platform. The platforms shown in the source arelinux,windows,darwin, andall, which covers all platforms. For example, if the Blender launch command differs by operating system, you can set an appropriate value for each platform in this section.Add optional MQTT notifications. Manager can send its internal events to an MQTT broker. Add the following section to
flamenco-manager.yamland restart Manager for the change to take effect:
mqtt:
client:
broker: "tcp://mqttserver.local:1883"
username: "username"
password: "your-password-here"
topic_prefix: flamencoThe broker address can use the tcp:// or ws:// format. If you use anonymous access, you can remove the username and password keys. topic_prefix sets the prefix for event topic names; with the example value flamenco, job updates are sent to the flamenco/jobs topic. Replace the example username, password, and broker address with your actual connection details. MQTT configuration is optional; if you enable it, you must restart Manager.
Common mistakes and things to watch for
Don’t confuse shared storage with local storage. Job files and render outputs are content that the Manager and Workers need to share; their location is set with shared_storage_path. Manager’s own logs and latest render image files are stored under local_manager_storage_path. These settings serve different purposes.
Don’t mistake Worker-specific files for shared configuration. The flamenco-worker.yaml file can be shared, but credentials and the Worker’s task update database are specific to each Worker. If you need to reset a Worker, first shut it down, then delete these files and remove the old Worker record from Manager’s web interface as well.
Understand timeout and failure settings before using them. Manager’s task_timeout and worker_timeout are used to determine whether a Worker is considered to have an issue or to have timed out. blocklist_threshold can prevent a Worker that has failed tasks of a particular type in a job from receiving more tasks of that type in the same job. task_fail_after_softfail_count determines how many Workers can fail a task before it is considered definitively failed. Understanding what these settings change is important when investigating issues in a render farm.
Don’t add MQTT if you don’t need it. The official configuration guide recommends leaving the relevant section empty if you’re unsure what MQTT is for. When MQTT is enabled, Manager can send events about its lifecycle, farm status, jobs, tasks, Workers, and latest render images to the configured topics.
Impact on architecture, interior design, and archviz workflows
Architecture and visualization teams rendering Blender scenes across multiple machines should first clarify the shared file path and Worker task permissions when configuring job distribution with Flamenco. Using shared storage for job files and outputs is important so the Manager and Workers can access the same job content. The task_types setting helps determine which types of work a Worker can take on.
oom_score_adjust in the Worker configuration is an optional setting on Linux that controls how Worker subprocesses are treated in an out-of-memory situation. According to the source, this setting can take a value from 0 to 1000. Raising the value is intended to make Blender more likely to be terminated before the Flamenco Worker if memory runs out during rendering. Teams should consider this option only if they use Linux and the described behavior is suitable for their workloads. The guide does not provide hardware purchasing recommendations or license pricing.
Next steps
First, check the Manager-Worker connection, then verify the shared storage path. Next, decide which task types to assign to Workers and confirm that the addresses and sample values in the configuration match your actual environment. If you need MQTT, add the broker access details and restart Manager. Clients listening for events should take the configured topic prefix and the topic names used by Flamenco into account.
Sources and license
This tutorial was adapted into Turkish based on Blender Flamenco’s official Manager Configuration, Worker Configuration, and Manager Configuration: MQTT guides. All three sources are identified as license=free in the supplied source information.
Sources
3 sourcesSource texts are not republished; short quotes are marked, everything else is our own summary and commentary.
For small and medium-sized visualization teams in Turkey, Flamenco’s configuration flexibility can be useful for assigning tasks across existing machines. Defining the shared storage path and Worker permissions correctly from the outset can help make render jobs more predictable across the team.
However, the guide does not provide information about hardware costs or license pricing, so decisions should be based on existing workloads and infrastructure. MQTT is not necessary for every team, either. A more cautious approach is to first verify the basic Manager-Worker connection and file access, then enable MQTT if monitoring needs call for it.
Frequently asked questions
Where is the Flamenco Manager configuration file located?
Manager reads its settings from the `flamenco-manager.yaml` file. It is located next to the `flamenco-manager` executable.
Can the Flamenco Worker configuration be shared with other Workers?
The `flamenco-worker.yaml` file in the Worker’s working directory can be shared. Files containing credentials and task updates are specific to each Worker.
How do you enable MQTT in Flamenco?
Add an `mqtt` configuration section to `flamenco-manager.yaml` and restart Manager. The broker address can use the `tcp://` or `ws://` format.



