Network render scheduling

[wyzass]

Having started to dabble with network rendering, we plan to use the shiny MAC's in our Graphics department as slaves.

Unfortunately, the Graphics team aren't too happy with us pulling all of their CPU when we push jobs through...

What would be good is to be able to set a schedule so that the Slave gets started at say 9pm and is then stopped at 8am - this way the renders can be run when the machines are sitting idle without the users needing to remember to start and stop the service...

[guest84672]

This is on the roadmap.

[PhilippeV8]

Is it also on the roadmap to add the option to sync it with HR-database, so that when people are on a day or half-day leave from work, their CPU can be used ? 

[DriesV]

I have been using a network rendering trial for 1.5 weeks now. I have configured 4 machines in our office as slaves, including my own workstation (slave + master).
I find that if you set the slave service priority to 'low' (as by default) the performance hit for users on the slave machines is quite minimal. I guess this will depend also on what software/workload is run on these machines. I think with quad core machines as slaves you should be save with respect to potential slowdowns.

[wyzass]

Agree with the Low priority having quite a low performance hit on Slave machines - mine is set to slave and I'm still running ProE and other stuff.

Strangely, we have another machine set as Slave while the user is in China on business yet my machine rendered way more tiles than his machine - even though his machine is doing nothing but act as a Slave...thoughts on this would be useful...

[wyzass]

Hi Thomas,

Do you know how far ahead on the roadmap the scheduling option is?

Are we talking weeks or months?

I currently have our IT guys playing with some code to switch the slave funciton on and off but with slightly mixed results so far.

Would be useful to know whether to tell them to stop or work harder