CGRU Forum

I've got a configuration challenge I'm not sure how to approach.

Currently the wake-on-lan task blocks are ran by any render node, Is this possible to run from the server instead? either through a configuration change, or setting the server up as a render node only taking wake-on-lan type jobs?

Our network configuration doesn't allow broadcast packets from client machines, only from the datacentre.

Not sure how to approach this!

Thanks as always!

Solved! I've set up extra patterns in the farm.json file:
global defaults set without any services
new group with regex mask for all render nodes - has all services except wol
new group with server in it, has only the wol service.

I love the control of Afanasy once you get your head around it!

One small point- how do I enable AF to automatically wake render nodes when needed?
I've read mentions of the capability of AF "Deciding" to wake nodes, but not sure how. Waking manually is working great!

Thanks!

Hi,
AF should do it by default.
If not - there is some bug.
The only configuration directive for this is:
https://github.com/CGRU/cgru/blob/maste ... .json#L235

These are definately not waking up, I've also set af_wolwake_interval to 1, just incase.
There aren't any wake-on-lan jobs being spawned into the system job queue, if that's relevant?
Can't see any error in the logs (other than the JSON ones we spoke about a number of weeks ago.)

What are the criteria for them waking? Do they need to have been put to sleep, or is "offline" state ok?

thank you

Rich

Hi.
afrende node should be in "Sleeping" state to be wake able.