Cloud render asset server can't find input file

Deadline
1

Hello!

I’m using deadline 10 and want to work with cloud rendering.
My setup:

  • Monitor on a Win10 machine
  • Remote connection server, Asset Server, Portal Link on a Linux server
  • Network drive with the input data which is mounted to the Linux server

I have a custom DL plugin, which reads the .ass file before rendering to do some magic. However this always fails, since it runs on the Spot fleet machine and can’t reach the file on my Network drive.

I think the asset server should come in the picture for this somehow,but I’m not sure how.
Can anyone help me with this?

2

Hello @pw_smn

Please refer to our docs link here, it explain how the network locations can be used by asset server.

You will need to use the path accessible by the machine running the asset server.

3

Thanks, I’ve already read this. The asset server is configured and can should have access to the files. The issue still remains.

4

You should check the logs of the asset server, central on the gateway, and slave on the worker.

Portal central maps paths from the local share to “fake” filesystems that sync files from s3. Asset server uploads these files to s3 to make them available. Generally a file should be transparently transferred when it is requested.

Did you correctly set path mapping for the aws workers?
Did portal link report that it correctly connected to the gateway?

5

For a totally different thought - given it’s a custom made plugin, is it possible you’ve forgotten to path-map the files you’re working with before trying to use them?

If the errors you’re seeing are referring to the path on your on-premise network / storage I’d suspect that’s the case.

Check the Arnold plugin, it’s pretty simple and you should be able to lift the pathmapping calls from there.

6

Good advice, thanks. Already done it though. Also tested with the vanilla Arnold plugin and got the same results unfortunately.

What I noticed is in the S3 bucket logs it says ACCESS DENIED in every few minutes. My bucket is as available as it gets now, but still the same error. Possibly related to the original error.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "PublicRead",
            "Effect": "Allow",
            "Principal": "*",
            "Action": "*",
            "Resource": "arn:aws:s3:::bucket-name-XXXXXXXXXXXXXXXXXXXXXXXXXXX/*"
        }
    ]
}
7

Let’s take a step back and please provide more info on your set up since you now mentioned your s3 bucket.

  • When you installed the AWS Portal Server, did you let the installer create IAM resources automatically or did you run the IAM creation scripts manually?

  • Are you creating a spot fleet through deadline or are you launching your own on-demand ec2-instances?

8
  • When you installed the AWS Portal Server, did you let the installer create IAM resources automatically or did you run the IAM creation scripts manually?

We tried both.

  • Are you creating a spot fleet through deadline or are you launching your own on-demand ec2-instances?

Creating it through deadline.

9

Good news!

The issue lied in our Path mapping.
Cloud machines have their own, generated path mapping, but ours were in effect, ruining the proper path mapping. After figuring this out and honing the path mapping rules, it’s working.
Thanks for the feedback!

2 Likes