Potential Issues

../../_images/ovas_api_banner_thin.png

There are two primary issues that are frequently seen:

Black screen on start-up

This can be caused by shader compilation during start-up, which can take 5-10 minutes, especially on a new cluster where the shared cache hasn’t been populated yet. This delay can also occur when new shaders need to be compiled for the first time. Although it may look like the session has failed, the application should start rendering once the shaders are compiled and cached. Future sessions will load significantly faster. An easy test is to wait 10-15 minutes, allowing the shaders to compile, and then seeing if the problem persists.

An alternative cause could be if the GPU Worker Nodes are unable to utilize the shared shader cache. This could be due to a network, configuration, or service issue, preventing the shaders from being properly cached and accessed across the cluster.

Network Routing

It is critical that you ensure there is a viable network route between the GPU Worker Node and the client. This is complicated by the use of UDP for the media and data communication. Many corporate firewalls block UDP traffic, so special rules may be required.

This can also present itself as a black-screen as typically the issue is that while the signaling communicating may be routable, the UDP traffic might be blocked.

Discussing the intricacies of Network Routing are outside the scope of this guide. However, the inability for the client to communicate with the Kit Application, in particular the UDP WebRTC traffic, is the leading cause of issues.

You can use the developer tools in modern browsers to help diagnose potential network issues. There are many helpful guides online.