UPS Master/Slave reconnect observation.


Mar 16, 2013
Thanks to a couple of switch failures, I have experienced two situations where UPS Slave Clients must be stopped and manually restarted in order to re-establish connection to the master. This has happened through several version updates.

Hardware: Five TrueNAS servers with LACP connections to several smart switches connected to pfsense CARPed routers backed up via 100's of pounds of batteries and 12,000W of inverters. The smart switches are not redundant like the router. This setup supports multiple PC's where each have their own, tiny individual consumer UPS. One TrueNAS server has a USB connection to a consumer CyberPower UPS plugged into mains power and nothing else - purely to sense and log a loss of mains power. The remaining 4 TrueNAS servers function as clients to the host having the USB connection. All the PC's on the system are also running WinNUT clients which connect to the one TrueNAS host on the network.

This is the second year when one of the smart switches developed a hardware failure and prevented the master TrueNas host from being accessible to some of the client/slave machines. None of the machines ever lost power. All the clients reported the failed UPS connection properly (both TrueNAS and WinNUT). All clients failed to automatically reconnect to the host upon replacement of the switch even though other NAS functions, like replications and backups had automatically resumed. The UPS services had to be stopped and restarted from GUI to clear the UPS faults. It also required stopping and restarting WinNUT clients on the Windows machines to reconnect.