I am having trouble getting more than two servers to stay connected to the Manager. The Manager in this case is a physical Linux machine running a 64 bit version of Scientific Linux (Redhat clone). The servers are running Windows XP-64. Everything has gone fine with the software setup until now.
Steps to reproduce:
• Launch Server on each render node and configure the settings to connect to the Manager. Server confirms connection to Manager and the Server is displayed in the Web Monitor
• Repeat this process on Server #2. Server shows up in Web Monitor and confirms connection to the Manager
• Repeat this process on Server #3
• Either Server 1 or Server 2 get disconnected and Server 3 now shows up in Web Monitor. If I check on the Server that got dropped the Server dialogue says:
If I wait a while and come back to Web Monitor and hit refresh a different Server gets dropped, with one of the missing servers taking its place. It is all very strange. They are all connected to the same Gigabit Ethernet switch so communication should not be a problem. SELinux and IPTables have both been disabled on the Linux machine running Manager and Web Monitor.
Check your XP event logs first and make sure you are not reaching a concurrect connection limit.
I have made zero progress with this problem. Since my last post I have:
• Swapped out the switch everything is plugged into
• Set up another server on a different piece of hardware to function as the Manager
• Set up a Linux VM on one of our ESXi hosts to function as the Manager
• Set up half a dozen Windows 7 (x64) workstations as "servers"
No matter what I do I can only maintain 2-3 concurrent connections to the Manager. The servers all report intermittent connections to the Manager. Kind of stuck here. The only thing I have not changed yet is the server OS itself. I'm going to try OpenSUSE (another RPM-based distro) and see if that works. If that doesn't work I'm going to set our render farm on fire.
Any ideas? Thanks in advance for your input...