I'm having trouble getting PDG's Deadline scheduler to work. Each attempt returns the following error message:

Deadline submit job command failed!
Error: Could not connect to any of the specified Mongo DB servers. The following errors were encountered:

- The remote certificate is invalid according to the validation procedure.

- The remote certificate is invalid according to the validation procedure.

(FranticX.Database.DatabaseConnectionException

This happens if I use the Deadline scheduler. It also happens if I submit a job via DeadlineCommand.exe in a Python processor command using the local scheduler.

I can paste the same submission command into the command line and run it without errors, so I somehow need to validate the command when it comes out of PDG.

Thanks! You should really log a bug for that.
https://www.sidefx.com/forum/topic/61697/ [www.sidefx.com]

Are you specifying a Deadline repository for either command or using the default installed version (i.e. leaving the repository field empty)?

Run the following command via the Python processor:

deadlinecommand.exe GetRepositoryRoot

Then run the same command in your commandline and compare the output repository path. Could be due to environment differences.

Also might be due to permission issues with the certificate if you are using one for your repo. Are you running with different user privileges for Houdini/PDG vs your command line?

Hey! Thanks for the replies.

The Deadline repository that I get from calling ‘deadlinecommand.exe GetRepositoryRoot’ in my commandline is the same one I'm using on my PDG Deadline scheduler. When I run the command via a Python processor, I get the same connection error from above.

I tried running PDG as an Administrator, but I got the same results.

Are you actually using SSL certificate for your deadline repository setup? If so, are you able to disable it just to test that being the problem?

Where does your repository live? Local network, cloud? Is the repository mounted as a shared drive?

Are the jobs actually being registered in Deadline's Monitor? If so, is the failure happening on the client machines?

I wonder if the command line shell you are using automatically sets up environment to handle the certificate validation properly. Does your shell automatically run any launch scripts?

Need more info to debug this further.

Just to follow up, support helped me work this out. In my case, the Deadline repository field on the scheduler needed to be left empty.

joshua_krause
Just to follow up, support helped me work this out. In my case, the Deadline repository field on the scheduler needed to be left empty.

Thanks for the update. Had this exact same problem and now everything is working.

Just a quick question though: Is there any reason why the $HFS entries in the Job Parms tab of the DeadlineScheduler escaped like so: ‘\$HFS’?

Yes, the reason is to allow $HFS and any other variables escaped with \ to be evaluated on the client machine running the job, as opposed to evaluating it on the local machine submitting the job. This is needed when you have a mixed farm setup (windows, linux, mac machines) or more generally, for a farm where the client machines differ in configuration from the submitter's machine. Unfortunately that seems to be rarely the case, so in the soon to be refreshed version of the TOP Deadline scheduler, I plan on removing the \.

Similarly, the Repository field will be locked away in a toggle so that users don't need to worry about having to set it (and therefore just use the empty field).