AWS Thinkbox Discussion Forums

Using pools to utilise farm ressources

We have a renderfarm of different types of computers, some newer and considerably faster than others. We have a wish for utilizing the ressources of our farm in the best possible way, and therefore have the faster computers take jobs before the slower ones.

The way we’ve tried to set it up, is by using pools and their priority. We’ve made a pool for all the fast computers, simply called “fast” and setup all submissions to use “fast” as primary pool and “none” as the secondary pool.

All submitted jobs
Pool: fast
Secondary Pool: none

Render farm example
Fast server #1 - Pools: fast, none
Fast server #2 - Pools: fast, none
Fast server #3 - Pools: fast, none
Fast server #4 - Pools: fast, none
Slow server #1 - Pools: none
Slow server #2 - Pools: none
Slow server #3 - Pools: none
Slow server #4 - Pools: none

The theory in the above setup, is that that if every slave is idle, and 4 jobs are submitted, all 4 jobs should start on the fast servers, and when subsequent jobs are submitted, they should recognize that the primary pool is full and fill op the slow servers.

However this is not the case. New jobs submitted, is still picked up by slow servers, even though a fast one is available.

Do I misunderstand how pools work, in regards to primary and seconday pools? It is also worth noting that we changed the Job Scheduling Order to: “Priority, Pool, First-In First-Out”, will this affect the desired outcome, when Pool isn’t first priority?

I actually found the answer myself. I was initially looking at the wrong documentation for 6.0

docs.thinkboxsoftware.com/produc … =secondary

“Note that the Secondary Pool feature was designed for Job Scheduling Orders that have Pool listed first, and might not work as expected otherwise. For example, if Priority is listed first, a job with lower priority that’s found during the initial Primary Pool scan will be preferred over a job with higher priority that’s found during the Secondary Pool scan. This is because the Secondary Pool scan is only performed if no jobs are found during the initial Primary Pool scan.”

I also found explanation on how secondary pools work.

Hello,

While there is no real way to prioritize based on speed outside of pools/groups currently, this has been asked for. I am glad you found a solution in the meantime, though.

Privacy | Site terms | Cookie preferences