Flamenco Manager
The Flamenco Manager is launched as a k8s Service within the kubernetes cluster and the Gateway Device NGINX Reverse Proxy allows computers on the Home Network to connect to this Web Console.
Manager Deployment
The Flamenco Manager is run on the k8s-control-plane-node as an exception to the rule so that it remains responsive when other (worker) nodes are heavily taxed rendering projects. If the Flamenco Manager becomes unstable and jittery.
The Flamenco Manager presents a Web Console and API on port 8080. It also requires SocketIO v2 (Websockets) to function properly as it uses websockets to poll for any current job status.
Configuration File
The Flamenco Manager uses a pre-created manager configuration file. The Flamenco Manager database and configuration is written out to the cluster NFS file store.
POD /media/shared/flamenco/flamenco-manager-storage
->
NFS /nfs/shares/flamenco/flamenco-manager-storage
The cluster also has Flamenco Workers use the internal cluster address of the Flamenco Manager service at http://flamenco-service.flamenco-manager.svc:8080.
Task Timeouts
The Flamenco Manager constantly polls and monitors how long tasks are taking and whether Flamenco Worker has checked in.
This is the most likely problem using default values in the Flameco Manager configuration. This will depend on several factors, not least, how long does your project scenes take to render and indirectly, how slow is your slowest Worker.
You can provision very large task_timeout values in the event you expect a Worker to spend hours on a scene… maybe because of sub par hardware performance. It is not advisable to have very large worker_timeout values. Workers may become unresponsive while working on rendering a scene but you want to know if they have never checked in and gone AWOL.