UE5.5 jobs re-rendering frames — output check racing ahead of file write/sync

[Henrik Bach Christensen]

Setup:
- Royal Render 9.1.16
- Render app: Unreal Engine 5.5 (Movie Render Queue)
- Output goes to local NAS.

Problem:
Some jobs render their frames a second time. The frames render very fast, and my read of it is that the rrServer's output check runs before the freshly written frames are fully flushed/visible on the share. The server sees them as missing (or too small), flags those frames, and reissues them to a client — so they render twice.

Questions:
1. Is there a way to make the rrServer wait a configurable number of seconds before performing the frame existence/size check after a client reports a segment done? If so, is it a per-job setting or a global rrServer/rrConfig value in 9.1.16, and where exactly is it?
2. Is the file-size check the more likely trigger here, and what's the cleanest way to set/relax the expected minimum size for UE5 MRQ output?
3. Is there a recommended UE5-specific setup for this (e.g. local render-out + copy) to avoid the file-visibility race entirely?

I can pull the frame log showing the re-send reason if that helps narrow down whether it's the existence check or the size check firing. 

Thanks!

H

[RR, Schoenberger]

Hi

1. Is there a way to make the rrServer wait a configurable number of seconds before performing the frame existence/size check after a client reports a segment done?

If you have one fileserver only and no other process that syncs files to the NAS, then the rrServer will not check frames before they are available.

In Detail:
If the rrServer has send frame 1-10 to a clients, then these frames are checked once the client finishes all frames.
For Unreal the rrClient waits until Unreal is closed before it sends the rrServer that the frame segment was finished.

But if you want to test it:
RR has a setting like this globally for all jobs.
It is intended for multiple (worldwide) caching/sync fileservers that take 20 seconds to sync the metadata.
rrConfig/files/fileserver:

>2. Is the file-size check the more likely trigger here

Please take a look at the job in rrControl, tab Job Info, server log.
There is a column with frames done.
And the rrServer should report if frames are deleted because of the file size.

But if the file size is the issue, then the job would never finish as a re-render writes the same file size.

What is the smallest file size and which file format?

Have you overridden the “min file size” in rrJob settings Limits?

>3. Is there a recommended UE5-specific setup for

As this did not happen before, we do not have any recommended settings.

 

> I can pull the frame log showing the re-send reason

At best, please export a debug info.
Then I can take a look at the render logs and the jobs server log.

Please select the job (or multiple) in rrControl.

Then execute "Export debug info - selected job(s)" via the menu "Debug".

Upload via www.RoyalRender.de/upload/

 

 

 

regards,
Holger Schönberger

Craftsman and Keeper of the Royal Render Flame

Please use the rrKnights Tavern
or our support system for new questions.