AWS Thinkbox Discussion Forums

Custom event plugin getting locked preventing editing.

I’ve found that if I edit a custom event plugin, what can happen is that a rendernode might error/crash due to saving the plugin with some kind of error. Instead of the slave closing the process gracefully, I end up with a broken event plugin that is locked and I can’t fix the error without rebooting the entire farm (or finding a way to discover which slave and process ID is locking the file.)
This is on a windows based renderfarm and a pixstor file server.

a) is this expected behaviour. I don’t remember having this problem in the past.
b) how do I avoid this problem?
c) is there a guide/tips on how best to stage/test plugins and event-plugins without breaking anything? Or is the only way to setup a sandbox repo?

By edit the plugin are you changing params from the UI or are you referring to editing the script? What event callbacks are you using? What is the error you are getting?

Regards,

Charles

Hi Charles,
by editing I do mean editing the code.
It can be any event plugin, but in this case both the shotgun event plugin and our own custom event plugin.
I don’t have the errors to hand. It’s most likely a syntax error in the python that causes the script to simply not load. Then, once that happens, I’m unable to correct the error because the file becomes locked by a deadline process on arbitrary rendernode or workstation on the farm.

Sandbox Repo is the best way. It’s quick, and I run three to four of them on my machine without even noticing. (Currently at three MongoDB instances are taking up ~190MB between them while Firefox eats ~900MB)

If something does go wrong, I’d expect it to be the the Python sandbox. Digging up which machine is using the file is up to the file server, but considering you’re blocked when the file is open for read, I’m guessing it’s Windows at the moment. You can hunt down who’s got what open by following a few of the answers here:

stackoverflow.com/questions/581 … work-share

If the Python Sandbox is locking up, we need to fix that and having a problem script would be great if you can figure out how to reproduce it reliably.

Privacy | Site terms | Cookie preferences