AWS Thinkbox Discussion Forums

Limit Render Slaves to certain Plugin types?

I might have missed something here but is it possible to limit a Render Slave to just one or two plugin types without using Groups?

Not really. That’s the main reason groups exist. That, and hardware capabilities.

What were you thinking here? Is there something groups aren’t able to do? Limits are a pretty good secondary option as you can apply more than one to a job at a time (RAM + 3D Software + Renderer for example).

I want to have one computer that is dedicated to a certain type of plugin… which we do via groups, but then I also want it to be able to do Draft and Tile Assembly jobs.

Interesting thought… It is a bit of work to make groups or limits just for one machine…

It’s not uncommon though to have simpler 2D jobs and processing to work on a single machine, but we don’t have an easy way to filter out job types for one machine.

We dream of one day having a scriptable queuing logic that could accommodate this, but there’s no ETA on that one.

Why not create a new “poolName” and allocate it this one machine? Send any job types to this one pool.

This isn’t about sending one job to one machine…

We have our custom plugin that is in a group of one machine. But this plugin is only used 4 or 5 times a day… so that machine is idle for a lot of the time, but I don’t want it to get locked up for hours on end with long render jobs so I only want it to pick up say TileAssembly jobs… but I don’t want it to do all of the tile-assembly jobs…

What I’d like to be able to do is add this machine to the other groups but then restrict is so that it can only do Tile Assembly jobs.

I’m not sure I fully understand your latest logic description Dave. Looking back at a previous thread reply:

So, we now have the ability to apply 1 or more limits at the app “plugin” level:
docs.thinkboxsoftware.com/produc … ic-options

these plugin limits are configured via the normal limits system, so you could create a limit for your custom plugin job type, which perhaps whitelists this 1 Slave to only run these custom plugin jobs:
docs.thinkboxsoftware.com/produc … #new-limit

then this free’s up your current “pools” and “groups” for any further job filter selection during the dequeuing evaluation. For example, this one Slave can also be a member of a “pool” or a member of several “pools”, which would allow it to pickup other Draft or Tile Assembly jobs.

If you need this one Slave to also be able interrupted for these Draft/Tile Assembly jobs, then ensure you apply job “interruptible” settings to those jobs which should be allowed to be interrupted if they are running on this one Slave:
docs.thinkboxsoftware.com/produc … erruptible

Further use the job “interruptible %” or “time remaining” settings to dial in more granular control.

Does that help?

The only way I can see to do it now is to make a group called “TileAssembly” which all machines are part of and force that to be the group used for all tile assembly… then my one machine can be a member of that group and the special plugin group.

But what happens if your one machine gets stuck processing for a long time via the TileAssembly group? Are we not back to the original issue now? I was previously answering the original thread title of how to limit 1 or more Slaves to a certain Plugin type but without using pools or groups.

Tile Assembly tasks don’t take very long thankfully… And we have our plugin on a couple of other machines as backup too, but it’s restricted to workstations for now.

Privacy | Site terms | Cookie preferences