Rendering stops and doesn't restart

[zach_ryder]

This has been happening for a while and I should have figured it out earlier.

About every 1-5 renderings completely fails in chaos rendering about halfway through but still takes cloud credit. The rendering thumbnail disappears and the render time continues to go up until I refresh the page, then it goes back to the time that it started. It's pretty frustrating.

Is there any fix to this problem?

[doychin_atanasov]

Would you give me examples of such failures? Maybe DM me links to the Cloud jobs so I could investigate them.

[zach_ryder]

I will send a screen recording when it happens again

[doychin_atanasov]

Even a URL to a past cloud job for which it happened would be enough. We do keep some diagnostics information for recently ran cloud jobs.

[doychin_atanasov]

I see that you've sent me one in a DM, thank you!

[ivaylocg]

The rendering process may experience occasional interruptions, which are implemented to optimize costs. During these interruptions, the rendering is temporarily halted and will resume once another available machine is ready.

Typically, these interruptions last for a few minutes. Once resumed, the rendering job picks up from the same point it reached before the interruption, utilizing V-Ray's resumable rendering feature.

It's important to note that no credits are charged during the interruption of the render job. However, please be aware that due to the minute-based reporting of cloud credit consumption, a minor uptick in credit usage — usually less than 0.1 credits — might appear following an interruption. This is merely a reporting delay and doesn't indicate actual credit deduction during the interruption period.

In cases where a job is paused or stopped by the user during an interruption, no output will be generated. It's worth mentioning that the pausing or stopping of a job will not happen immediately; rather, it will occur once a new machine becomes available or when a timeout is reached. Additionally, the job preview will be inaccessible throughout the interruption period. If you refresh the cloud page, the reported render time will be reset to the original time of the interruption. We are aware of these issues and we are actively working on fixing them.

For GPU jobs, interruptions can occur more frequently and may last longer, particularly during periods of high GPU demand. GPU resources are generally less available in the cloud compared to CPU resources.