This sort of “Could not find file” for temporary file, with SpillCollectorProcess on the stack seems to result from prior errors. Your earlier error might be below. What is this backend, and any idea why it errors you?
Are you able to do smaller backups successfully? If so, look in the <job> → Log → Complete log for RetryAttempts to see if you’re starting to get some random ones, and just survived due to default 5 number-of-retries managing to get through eventually. Any logs on the remote that would show 504?
You can also try getting the target URL from <job> → Export → As Command-line to edit with a new folder (the proposed test needs an empty folder) and run Duplicati.CommandLine.BackendTester.exe
It would be helpful to know what got 504. A way to find that (and maybe make a big file) is log-file=path along with log-file-log-level=retry to see what sort of things upload, how fast (timeout?), and what fails.
Are you running at the default 50 MB remote volume size, a.k.a dblock-size? How many source files? Assuming you’ve had at least one backup finish, the job log shows this under Source Files Examined, otherwise I guess you could do some sort of find command or something to try to get an estimate…
What I’m wondering is whether you’re making a dlist file that’s really large (lots of files), and timing out.