I’m trying to save power in the building using Deadline.
So I took one rendernode of our renderfarm and tried to set it up.
I can shutdown the machine in deadline as a super user, wake-up-on-lan is configured, as I can start the machine also using deadline as a super user.
I configured yesterday deadline in a way as you can see in the attachments.
The machine did shutdown, but never came up this morning. Why is that? Can it be, it didn’t came up because there wasn’t a render submitted? Or how can I test this better? Did I configure something wrong?
Yup, that would be the reason. The WOL feature only wakes up machines as they are needed. For logging information while power management is running, check the Power Management tab in Pulse. This will show you which machines are candidates for being woken up and shut down (if there are any).
Thanks for your respons. I’ve made a limit group with only this renderfarm inside. And submitted a fusion flow onto that limit group, but the machine didn’t boot.
So, I checked the power management tab in deadline pulse , but there I can only see results of loggings?
In the limit group, was the machine blacklisted or whitelisted? If it was blacklisted, that would explain why no attempt was made to start it up.
One thing to try would be to start up the slave normally, and then make sure it picks up the job that you’ve submitted. If it does, then shutdown that machine and resubmit the job and see if Pulse starts it up.
The slave is for sure in the whitelist. I’ve resubmitted the job into that specific limit group and it ran.
I’ve stopped the job and shutted down the machine, and resubmitted the job.
In deadline pulse I receive this message:
Checking job “185_239_Rabbits_Kinepolis_LV_Vs001.comp” (002_090_999_277e0da5)
only slaves with assigned pool “fusion51” should be woken up
job is not part of a group, any slaves should be woken up
slaves that are candidates to be woken up based on pool and group: Rf021
checking slave Rf021
waking up offline slave Rf021 because it is required to render job “185_239_Rabbits_Kinepolis_LV_Vs001.comp”
And indeed, after a while, it booted.
Apparently I wasn’t patient enough… I thought the slave would boot immediately after the submission of the job…
Cool, glad it’s working for you now! By default, the power management check is only made every 5 minutes. This operation can be pretty heavy at times, hence the delay. In the future, for testing purposes, you can select Control -> Perform Power Management Check in Pulse to have it perform the check immediately.
I thought the slave would boot immediately after the submission of the job…