AWS Thinkbox Discussion Forums

failed to import module

Whenever we reboot rendernodes, we tend to get this error on a large percentage of them when trying to import an internal module:

Traceback (most recent call last):
File “none”, line 19, in
ImportError: No module named scanline.Paths

The weird thing is, another modules load just fine, from the same site-packages folder a couple of lines later…

Any ideas?

Edit: Also, the script thats trying to load the failing “not found” libraries would not even run if it could not run some of our central libraries earlier… Its very odd.

We run a site customize script on python init that loads some scripts off a network drive to initialize the environment. And by the time this error pops, that process has finished successfully (otherwise none of the other libraries would load)

Not too sure why this would happen. I’m guessing scanline.Paths is in a network location? From our experience with Deadline 5, we’ve seen cases where parts of the repository wouldn’t be accessible for a brief period of time, while other parts were still accessible. We fought with that issue for years, and it was just another reason to move to a database. I wonder if you’re seeing a similar issue where some modules are visible at a certain time and others on the same share are not…

It could be… we seem to be blaming samba for that right now. This error is hard to detect, as it pops randomly, but sadly due to the lack of encapsulated python environments, it causes slaves to be incapacitated for periods of time.

Privacy | Site terms | Cookie preferences