Hi,
We’ve recently experienced some weird issues with tasks activating with no slave attached causing the job to stall until we requeue the task. Is this a problem anyone else has experienced? It generally seems to happen when multiple slaves pickup small jobs at the same time. We’re runninng a 60 machine farm with pulse setup and working. Any suggestions on why this might happen?
Nick
Hey Nick,
This is actually another known issue that should be resolved in 5.1. It happens when you are running pulse, and there is a bit of network latency. Essentially, the slave doesn’t “see” the task file on disk in the repository after pulse has assigned it, so the slave thinks the task was requeued and moves on. This is why it is orphaned like this. In 5.1, the slave now waits to confirm it has received the task before starting the render.
Cheers,
Haha! I’m just going to assume from now on that anything weird we have going on is fixed in 5.1! I guess we could sign up to the Beta to fix these things but if it’s only a short time until the full release I’d just as happily wait… Do you have a firm date yet for the public release of 5.1?
Cheers
Nick
As long as nothing catastrophic happens in beta at this point, we are looking at a release date in the first or second full week of December.
It definitely doesn’t hurt to join the beta now though. The first release candidate should be available either today or tomorrow.
Cheers,