![]() Apache License, Version 2.0, ( LICENSE-APACHE or )Īny kinds of contributions are welcome as a pull request.Jibri-pod-controller is licensed under either of If multiple jibri-pod-controller pods are running, one is elected to run the sweeper. A sweeper runs on a configurable interval to remove any expired jibri pods in case jibri fails to send the webhook for any reason. When jibri finishes recording or live-streaming, jibri-pod-controller will delete the pod.This isolates the jibri pod from the Deployment - the Deployment will immediately launch another jibri pod to replace it (thus keeping the required number of spare pods), and the isolated jibri pod will continue to run. When a jibri pod starts to record or live-stream, jibri-pod-controller will patch the pod's labels so that they don't match the Deployment's label selector.Configure jibri in single use mode, and configure it to send webhook requests to jibri-pod-controller.Deploy jibri-pod-controller in your cluster and give it RBAC permission to get/list/patch jibri pods.Set the Deployment's replicas to the number of spare jibri pods you want to run. Jibri-pod-controller can be used as part of an alternative approach: This is difficult to achieve with the standard Kubernetes horizontal pod autoscaler (HPA). When managing a large Jibri deployment, you usually want to autoscale using a strategy like "always keep N spare (not recording or livestreaming) jibri pods". We will update docs with that information soon.Jibri-pod-controller: A tool for managing the scaling of large Jibri deployments in Kubernetes. that is how through jitsu-workspace named volume jitsu and airbyte source manage to interchange files.dind-daemon container have /tmp/airbyte/ path mapped as jitsu-workspace named volume. ![]()
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |