I understand that servers shouldn’t reboot that often, if ever, however, I just had to reboot my server for maintenance reasons and it, of course, started the FlexLM service on a completely different port.
Which means none of my slaves are running and I have to change the ports manually for each of them. I only have 5 licenses, so it’s not such a big deal, however, one of the slaves refuses to start up even though I changed the lic port, even though I restarted it, even though the deadline service is running wihtout issues with admin priviledges (just like the others). The only way to get the slave working is to stop the service and start the slave app manually. Which is a pain and I can’t log off after that.
So, basically, two questions:
Is there a way to change the port on which deadline repository grants the license on the server?
How the hell do I make the slave to be online and functional without having to run the slave app manually?
All the slaves are the same on Windows 7 x64, same config, same version, same everything. Up until the restart of the server all was running smoothly.
If you don’t have these ports hardcoded in the license file, the I believe flexlm will use a random port between 27000 an 27009 when it starts up. In this case, when you set the license server setting on the slave, it should be in the form “@server”. It sounds like you have it set as “port@server”, which is only necessary when ports have been hardcoded in your license file. Please confirm if you have them hardcoded or not and we can go from there.
For the slave that isn’t starting up, do you know if the slave tries to start up but then fast-exits, or does it not start up at all? Are you relying on the “launch slave at startup” option to launch the slave, or are you using remote control through the monitor to try and start the slave?
Oh, I didn’t know I could have only specified “@serverIP” in the license field. Thanks for the tip! I changed it just now on all my slaves and so far it seems to be working! So, next time I’ll need to restart the server or Deadline service/flexlm service it should still be all fine and I won’t have to change the license path again? That’d be great! Btw: no I don’t have the port hardcoded in the license file.
As for the non-starting slave.
Since I had to restart the server, all went down, incl. Deadline.
Then as soon as the server started I changed the path to the license on the server (incl. the new port) and had the slaves running as they were using the Launcher service. I operate the machines via Remote Desktop in Windows.
The only single machine not wanting to actually report as “idle” is the very first render slave in my farm. Not really sure why. I tried restarting the machine, restarting the service, nothing works. Only when I run the Slave application manually it reports as idle and renders. But as a service, not.
The service is running, by the way, all seems to be fine, however, I see the slave in the monitor as Offline and it doesn’t accept any tasks.
The first thing we should check is if the slave tries to start up, but then exits. Use Remote Control in the Monitor to try and launch the slave. Wait a little bit (like 20 seconds), and then check the Deadline logs folder on the machine that the slave didn’t start up on. To find the folder quickly, just launch the Monitor and select Help → Explore Log Folder.
Look to see if there is a new slave log here (just sort by Date Modified). If there is, please post it and we’ll take a look. Also post the most recent Launcher log, and we’ll check to make sure it’s actually receiving the Remote Control command.
Unfortunately I can’t remotely jump start the Slave as it throws this error:
Machine Name Command Timestamp Status Results
Duber-slave-01 LaunchSlave 10/03 17:53 Failed The requested address is not valid in its context 255.255.255.255:5042
and even if I try to do the same on the local network, I get an error saying that the Remote Administrator has been disabled, even though I specifically for this enabled it…
However, I don’t have RAdmin installed and I’m not planning to as it’s not a free app and I really don’t need it. Is there any other way I can deal with this problem?
By the way, this is the latest log in the ProgramData log folder on the Slave machine:
2010-10-03 18:26:49: BEGIN - DUBER-SLAVE-01\loocas
2010-10-03 18:26:49: Start-up
2010-10-03 18:26:49: 2010-10-03 18:26:49
2010-10-03 18:26:49: Deadline Launcher 4.1 [v4.1.0.42706 R]
2010-10-03 18:26:50: Local version file: C:\Program Files\Prime Focus\Deadline\bin\Version
2010-10-03 18:26:50: Network version file: \messiah\deadline\bin\Windows\Version
2010-10-03 18:26:50: Comparing version files
2010-10-03 18:26:50: Launcher Thread - Launcher thread initializing…
2010-10-03 18:26:50: Perfoming remote admin check
2010-10-03 18:26:50: Launcher Thread - Remote administration is disabled
2010-10-03 18:26:50: Launcher Thread - Launcher thread listing on port 5042
2010-10-03 18:27:50: Perfoming remote admin check
Ok, there has to be something wrong on the Server side as after a reboot another render node can’t get the “idle” status in the Monitor and doesn’t accept jobs, until I manually disable the Deadline Service and manually fire up the Deadline Slave app.
Could it have anything to do with the license? As I just had a prbolem with the FLEXlm not assigning a license to the freshly restarted slave, even though I have 5 licenses, but I only had 4 machines running + the one restarted slave. I had to manually re-read the license file and stop/start the service on the Server.
Can you double check that error message? I would expect it to say Remote AdministraTION has been disabled, not Remote AdministraTOR. Remote Administrator is a remoting tool, like VNC, and isn’t needed to run any of the Remote Control commands (like Start Slave). To check if you have Remote Administration enabled (which is necessary to use Remote Control), look under the Launcher Settings in the Repository Options.
It just sounds like the slave application isn’t launching on these nodes (if it was, you’d see deadlineslave.exe in Task Manager). Once you’ve enabled Remote Administration, restart your nodes, and then send the Remote Control → Start Slave command to those that don’t launch the slave to see if it is able to run.
