Author Topic: DR crashing render just before denoising  (Read 1765 times)

2019-02-05, 18:59:03

nerfherder

  • Active Users
  • **
  • Posts: 22
    • View Profile
Hi.
Over the last week or so distributed renders have, more often than not, been crashing at the end of the render, just before denoising starts and while gathering data from the nodes. All systems are on v3 hotfix 1, Max 2018.4.
The render log says 'Sending sampling mask to slaves timed out. Try to increase synchronization time or decrease maximum pixels per transfer.'
The DRLog on the node also says 'Sending file to remote side failed'.
Nothing has been changed with the setup and I've never had to touch the synchronization and pixels per transfer values before, so I'm not sure what are 'good' values to use. I'm assuming its a problem caused by something else in my system. If anyone can point me in the direction of a solution that would be most appreciated, as I can't reliably use DR at the moment.
Cheers

2019-02-05, 19:19:57
Reply #1

TomG

  • Administrator
  • Active Users
  • *****
  • Posts: 6136
    • View Profile
I have no idea if this is related - is Bloom and Glare enabled? This can cause Max to hang, so I am thinking if maybe denoising kicks in, Bloom and Glare recalculates and shows the status bar, and then Max freezes, the nodes would then be unable to reach it and conclude it was switched off / crashed / otherwise not there.
Tom Grimes | chaos-corona.com
Product Manager | contact us

2019-02-06, 00:01:49
Reply #2

nerfherder

  • Active Users
  • **
  • Posts: 22
    • View Profile
Thanks for the suggestion - I'll look into it. If I remember it correctly, bloom and glare are enabled on the Lightmix element but perhaps not in the main settings.
1 other thing I discovered when testing it before - on the Master machine it appeared to hang at the end of the render, so I went to the only node running and shut down DRSlave - the master then immediately started to denoise and finished the render.

2019-02-08, 11:45:46
Reply #3

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 13722
  • Marcin
    • View Profile
Sounds like some issue with the communication between master and slaves. If you want to try with sync interval and max pixels, you can adjust the values by half, and then see if it helps. So sync interval 60s > 120s, and max pixels 500 000 > 250 000.

If something is crashing or freezing, the best idea is to capture a minidump from 3dsmax.exe and send it to us. In case of DR issues, you can also capture a minidump from drserver.exe. Here is how to do it (check out points 1 and 2 depending on the case) - https://coronarenderer.freshdesk.com/support/solutions/articles/5000524006

If it's happening only in some specific scene(s), then you can send it to us and we will try reproducing it.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us