DR seems to keep rendering after completing passes

[Dippndots]

I'm having an odd problem when at the stage "gathering remaining data from slaves". The render just seems to keep going. It completes the 90 required passes, now 15min later it's still gathering remaining data and is at 157 passes...

What can I do to troubleshoot this?

[Dippndots]

These lines show up in backburner's server log:

Code: [Select]

2015/11/27 09:08:54 INF 3dsmax adapter : CORONA : Pass 88/90
2015/11/27 09:08:54 INF 3dsmax adapter : CORONA : Pass 89/90
2015/11/27 09:08:54 INF 3dsmax adapter : CORONA : Saving + Cleaning up
2015/11/27 09:08:54 INF 3dsmax adapter : ***WARNING*** CORONA ERROR: DR error: Accumulating EXR from slave '10.0.7.134' failed. Please report this to our support: ERROR: Iex::BaseExc: Error reading pixel data from image file "{memory}". Data decompression (zlib) failed.


[Ondra]

this is really interesting. Does this happen every time? Do you have multiple slaves, if yes, is the problem in all of them, or just one?

[Dippndots]

It happens at least 4 out of 5 times. Multiple slaves report the same problems (we have 25).

[Dippndots]

I should add that it also just locks up similarly mid render at random passes and the server log has the same lines

[Ondra]

ok, I will add more info to the error message and we will debug the problem together. Can we continue on skype? If yes, can you send me your screen name in PM?

[Dippndots]

I'm not at work anymore unfortunately :( Happy to Skype, but the earliest I could would be next Thursday

[Ondra]

sure, no problem

[Nekrobul]

I have one idea about this problem.

I have encounterd this when i had 2 distant nodes and had to use VPN connection to fake local area connection. I found out that lowering default settings fixes it. I usualy used numbers like 3sec sinc interval na vax 50k pixels to transfer at once.

[Dippndots]

Those are the settings I've currently got, I pushed it out to 10sec with no difference. All the computers are on the same local network.

[Ondra]

I will close this as it is too old, but if you experience it with 1.4 please let us know. There is now improved errror reporting, so it should point you to a file you should send us

[Dippndots]

Hey Ondra, I think I figured out the cause of this. When rendering a 5000px image between 26 slaves, i currently have DR setup so it can only transfer 50,000px every 15sec. I noticed that after it had reached the pass limit, these 50,000px .exrs were still being generated by the slaves (up to 200 of them) to send to the master and because it could only transfer every 15sec, it seemed like it was taking forever.

I will test again some time today with the default pixel amount and see if that doesn't solve the problem.