This is the weirdest thing I've ever seen. All agents could still take calls. The ones with "No Service" took calls after calls. The others appeared to happen sporadically. Sometimes, they were in ready state but did not get any calls. TAC engineer told me that it appears to be a firewall issue because I failed to telnet to port 59020 to the Chat server from an agent machine. But other machines didn’t have any problem. In fact, windows firewall is disabled on all machines in the OU and there are no other client firewalls installed on their machines at all. It took the TAC engineer several hours but he couldn't figure out why. Finally, we decided to bounce the node manager and it fixed the problem. Yep, that's it. We did not make any changes at all; restarting the node manager was all that we did. But there was another problem after I restarted the node manager, i.e. the timezone on master CRS engine got changed from Pacific to Mountain time unexpectedly. Thus it caused all the queues to close 1 hour earlier.
Has anyone ever seen the same issue? One thing I wanted to point out is that our servers haven't been rebooted for 8 months now. Some people in our IT department have suggested that I should regularly reboot them at least once every 1-3 months. What's your opinion on this?