Hi,
Recently I have had the problem that when the machines go into idle shutdown, not immediately, but after a while no IP address is assigned. Then the WOL no longer works. I thought that either the command would be sent to the MAC address or the hostname would be resolved accordingly.
Then I tried to switch to “FQDN” in the worker settings, but then even workers who are online don’t start the job. Any ideas? (Version: 10.3.1.3)
Hello
Thanks for reaching out. Are those IP address assigned through DHCP? And does it release the IP on shutdown?
Also do you have this option enabled in Monitor> Tools> Configure Repo Options> Worker settings:
Yes, i am using DHCP, but it is not setup , to release the IP on Shutdown.
I can try anable this, but i thought, this wouldnt work, because of no IP adress is given…
Enable “Workers IP Adress for remote…” didnt solve the problem
Please disable that setting I was interested to know if it was enabled or not it should not be enabled.
I think you should isolate the issue from Deadline. The error seems to be coming from the OS. I think a small test will be to check if the DHCP release the IP is to manually turn the machine on and check the IP.
You can try a software other than Deadline to to send Magic Packet to Wake the machine up on LAN to be sure if you get the same error.
Yes, strange, the computer with the IP address is in my firewall/DHCP (which is only released after 48 hours), and if I wake up the computer with a Powershell Script - WOL - MAC, then it also gets the IP that it had . On the server where the deadline server is running, I was also able to try to ping the computer by name, and the correct IP address was resolved.
If you right click the Worker in the Monitor after it’s missing the IP address and go to ‘Worker History’ is there any activity there? It’s the Worker application on the render node that populates the database with information, but it is possible for a user to make tweaks.
Nothing special there. If it is the worker, then an explanation would be that it no longer passes an IP during shutdown? hmmm… that would be new to me.
Shouldn’t you be looking at the dhcp server first? What are its settings?
If you try to nslookup PSLSRV74 what do you get?
1 Like
Server: psldc06.p5l.io
Address: 10.1.1.6
Name: PSLSRV74.p5l.io
Addresses: 10.1.11.74
10.1.10.74
Didnt find any problems at the DHCP Sever. The lost ips have not been released and are still set to the machines.
Could you then try
nslookup 10.1.10.74
We recently had a case, where the same address was handed to multiple machines, and thus we were losing connection to them (because it resolved to a machine we did not expect).
Not sure if this is the case here, but it might.
So the worker lives on two subnets?
Server: psldc06.p5l.io
Address: 10.1.1.6
Name: pslsrv74.p5l.io
Address: 10.1.10.74
Okay, but that doesn’t seem to be the problem. i using the 255.255.0.0 subnetmask.
Maybe one of the Thinkbox peeps can answer, but I think Deadline wants to associate one IPv4 address to one hostname. Not sure by which mechanism Deadline is using for a host lookup.
I was going to ask you questions about DNS round-robin, netmask ordering and DHCP reservations, but I think my first question is:
Is there a specific reason you have multiple interfaces on the render node?
For testing have you tried:
Overriding the MAC addr for WoL in the (local) Worker Properties?
OR
Is one of the interfaces on the node considered “primary”? If so, is that IPv4 metric set to be a higher priority over the other?
I have several worker instances to use 3d renderings (GPU) and comp renderings (CPU) separately and simultaneously if necessary.
Why should I overwrite the Mac address?
No.