Author Topic: DR Render won't restart after each job  (Read 1998 times)

2024-06-20, 23:38:56

Dionysios.TS

  • Active Users
  • **
  • Posts: 777
    • View Profile
    • Evolvia Imaging
Hi again guys,

I use the latest Corona Daily 12 RC 1 build.
I've noticed that by using Backburner, once a job is finished in DR mode, the next one when it starts it doesn't trigger the slaves to start loading the scene.
I need to close DR Server and open it again and then is fine.

What's going on?

Thanks,

Dionysios -

2024-06-21, 08:57:35
Reply #1

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 13154
  • Marcin
    • View Profile
Hi Dionysios,

Which version of 3ds Max and Backburner are you using?
What exactly do you mean by "using Backburner in DR mode"? Do you mean that you are distributing jobs through Backburner and rendering them with Corona DR over multiple nodes?

If you are using BB+DR this way, this is something we don't officially support. But if we can confirm that it used to work in V11 HF2 and stopped working now, then we will see if we can bring it back.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2024-06-21, 10:51:34
Reply #2

Dionysios.TS

  • Active Users
  • **
  • Posts: 777
    • View Profile
    • Evolvia Imaging
Hello Maru,

excuse me if my initial post wasn't clear.
First of all for this production I am working on now I use 3ds Max 2023 and the latest Backburner build.

I use Backburner to make the list of jobs to send to the Master PC while the DR is active and as usually once the jobs start, Corona should send the data to the DR Servers so they can contribute to the render.
But now I need everytime a job is finished on the Master PC, close the DR Server module and start it again to make it work. In a few words, the DR Servers are connected at the end of the job but they don't start once the next job is assigned and loaded on the Master PC.

2024-06-21, 16:23:25
Reply #3

Frood

  • Active Users
  • **
  • Posts: 1967
    • View Profile
    • Rakete GmbH
Hi,

But now I need everytime a job is finished on the Master PC, close the DR Server module and start it again to make it work.

Where exactly do you need to restart the dr server? On the dr slaves only? Or do you use an (unfortunate) setup where the master has a dr server running and is included as dr slave - and you have to restart it there? Can be found anything useful in the logs?

I tried to reproduce this, because I was quite frightened by your report (I'm using BB+DR regularly and would like to do so in the future). But everything seems to work fine here. Even when using the master as DR slave simultaneously (absolutely not recommended and evil of course).

This has been the test environment:

  • Loading V12 via multiloaders (not used the installer)
  • Manually replaced dr server by v12 dr server on slaves (because v11.2 is installed there)
  • Using dr server as windows service via nssm.exe
  • Having dr nodes configured as resolved names in "IP/Hostname" column of render setup -> system
  • Having "RestartMax = false" in DrConfig-1.5.txt on dr nodes
  • Tried "Search LAN during render" on and off
  • Versions were Max 2024, Corona V12 RC1, Backburner v2019.0.0.5

When loading the scene, it has properly configured dr settings (I dump them by default at loading time into Max log), and the master uses them accordingly. Tested jobs with single and multiple frames. If the master changes to the next job queued, dr slaves jump in as desired. I submitted the jobs to one node using backburner, while having three others configured as slaves. In case of "Search LAN" active, the slave list was empty and all slaves having dr server running were found.


Good Luck



Never underestimate the power of a well placed level one spell.

2024-06-22, 11:42:34
Reply #4

Dionysios.TS

  • Active Users
  • **
  • Posts: 777
    • View Profile
    • Evolvia Imaging
Hi,

But now I need everytime a job is finished on the Master PC, close the DR Server module and start it again to make it work.

Where exactly do you need to restart the dr server? On the dr slaves only? Or do you use an (unfortunate) setup where the master has a dr server running and is included as dr slave - and you have to restart it there? Can be found anything useful in the logs?

I tried to reproduce this, because I was quite frightened by your report (I'm using BB+DR regularly and would like to do so in the future). But everything seems to work fine here. Even when using the master as DR slave simultaneously (absolutely not recommended and evil of course).

This has been the test environment:

  • Loading V12 via multiloaders (not used the installer)
  • Manually replaced dr server by v12 dr server on slaves (because v11.2 is installed there)
  • Using dr server as windows service via nssm.exe
  • Having dr nodes configured as resolved names in "IP/Hostname" column of render setup -> system
  • Having "RestartMax = false" in DrConfig-1.5.txt on dr nodes
  • Tried "Search LAN during render" on and off
  • Versions were Max 2024, Corona V12 RC1, Backburner v2019.0.0.5

When loading the scene, it has properly configured dr settings (I dump them by default at loading time into Max log), and the master uses them accordingly. Tested jobs with single and multiple frames. If the master changes to the next job queued, dr slaves jump in as desired. I submitted the jobs to one node using backburner, while having three others configured as slaves. In case of "Search LAN" active, the slave list was empty and all slaves having dr server running were found.


Good Luck

First of all thanks for doing the tests!

No, the Master is not included as DR Server. :)
My setup is simple and logic, tha Master receives the Backburner job, it starts, and the DR Servers should start accordingly.
I never had issues lately so I was a bit frightened as well.

I'll try a clean install at this point to all the machine this weelend and I'll come back here to report.

Thanks again for your efforts,

Dionysios -
« Last Edit: 2024-06-22, 15:16:28 by Dionysios.TS »

2024-06-22, 12:16:20
Reply #5

Dionysios.TS

  • Active Users
  • **
  • Posts: 777
    • View Profile
    • Evolvia Imaging
I did a clean install everywhere and now it works!

I'll keep you updated guys ok?

2024-06-22, 13:59:29
Reply #6

Frood

  • Active Users
  • **
  • Posts: 1967
    • View Profile
    • Rakete GmbH
I did a clean install everywhere and now it works!

More than glad to read that :)


Good Luck



Never underestimate the power of a well placed level one spell.

2024-07-18, 10:33:22
Reply #7

Dionysios.TS

  • Active Users
  • **
  • Posts: 777
    • View Profile
    • Evolvia Imaging
I am coming back to this issue as I found the real reason of the problem.

I had Max 2023 + Max 2025 installed on the DR Server and it seems that having more than one releases installed on the same machine is the reason why Corona doesn't start automatically when a job in Backburner is finished and another one starts, so DR Server for a strange reason doesn't receive the next job starting message.

Uninstall Max 2025 resolved the issue, this is strange to me, I don't know if other users who have more releases of Max installed on the same DR Server have encountered this behavior...

Thanks,

Dionysios -

2024-07-18, 10:58:21
Reply #8

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 13154
  • Marcin
    • View Profile
Uninstall Max 2025 resolved the issue, this is strange to me, I don't know if other users who have more releases of Max installed on the same DR Server have encountered this behavior...

Perhaps the DR Server was launching the wrong version of 3ds Max. You can try either launching it with its UI and changing the Max version there (screenshot1) or going to C:\Users\_username_\AppData\Local\CoronaRenderer\DrData\DrConfig-1.5.txt and changing the version there (screenshot2).
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2024-07-18, 11:03:35
Reply #9

Dionysios.TS

  • Active Users
  • **
  • Posts: 777
    • View Profile
    • Evolvia Imaging
Thanks for the info.

I've changed the Max version in the UI and in effect the first job always started right, the problem was always produced to the following jobs in the Backburner's list.
Didn't try the *.txt solution, I'll give it a shot soon.