Smart-UPS Causes Power Loss to Connected Devices

[Anonymous user]

Hello, I'm hoping someone can help me with an issue I've encountered with one of our SURTD3000XLI models. The UPS unexpectedly rebooted the other day, causing the 10 Cisco 2960 access switches connected to it to reboot. My understanding is that these online models should 'fail on', thanks to their automatic internal bypass feature, meaning that the access switches should not have rebooted.

My initial question is therefore, is my understanding of my model's expected behaviour correct?

I should add that the UPS was in online mode at the time of its unexpected reboot (i.e. the manual bypass switch at the back of the unit was in the 'Online' position).

I'd be grateful if anyone could share their expertise of this UPS model with me. Whilst I have now replaced the suspected faulty 3000XLI unit with a 3000RMXLI unit, I have a number of other 3000XLI models out there, supporting our networking estate. I therefore urgently need to understand more about the expected behaviour of them during UPS fault scenarios.

Thanks in advance.

Andy

[voidstar_apc]

The UPSes are supposed to go to bypass during a fault providing the input voltage is within the acceptable bypass limits. If there is a problem with the bypass system, then a subsequent UPS fault could drop the load.

Big picture, we buy UPSes to prevent downtime, but UPSes themselves can fail causing downtime. This trade-off is a win if you have frequent power failures, but suppose you're aiming for even greater reliability... what to do? There are a few options:

1) The Symmetra line of UPSes has redundant power modules so both the redundant modules and the bypass system would have to fail. Although this is much more reliable there are still ways this could fail, so...

2) Often critical equipment is dual-corded having redundant power supplies. So you put a separate UPS on each supply. The nice thing about this arrangement is both power paths are active, so you're more likely to detect failures before they become a problem. It looks like Cisco 2960 switches can have redundant supplies.

3) Sometimes even that isn't enough, so instead of spending money keeping a piece of infrastructure up, you can allow it to go down and automatically failover to redundant infrastructure.

[Anonymous user]

Hi Voidstar. Thanks very much for taking the time to respond. Really appreciate it.

Do you think its worth checking the bypass limits of my other UPSes, in case the current thresholds in play would prevent a bypass during a fault? (I'm assuming this check, and any subsequent adjustments, can be done via the serial interface, or remotely via our StruxureWare software).

I also intend to remotely reboot of the UPSes out of hours, to see whether the process causes the access switches to reboot. Everything I read about the UPS model involved suggests that a reboot Should NOT affect connected devices, but I'd still like to see it for myself.

[voidstar_apc]

You can certainly check the bypass limits. Note that the UPS will throw an alarm "Bypass out of range" when the input is outside the bypass limits so presumably StruxureWare would already be telling you when this is the case.

I'm not entirely clear what you mean by "reboot". If you mean reboot the network management card (NMC), then yes, the UPS is unaffected -- the NMC is it's own separate processor. On the other hand, you can also reboot the UPS output under control->ups or control->outlet groups. Unless the load has redundant power supplies or it's own backup battery, it will be affected.

[Anonymous user]

Well, we didn't see anything in our StruxureWare logs, so its likely that the thresholds set are okay.

I did mean a reboot of the device. One of our network engineers was adamant that power to connected devices is not affected by a reboot, whereas previous experience of colleagues suggests otherwise. Its likely that the network engineer is basing his opinion on another UPS model (perhaps one with redundant power supplies).

Anyway, I think I have enough to go on so, thanks once again for your input.

[voidstar_apc]

Some devices have enough energy storage to survive a short loss of power. Maybe the network engineer is familiar with a switch like that? It's a bit of a problem if you want the device to actually reboot, so the UPS has a "reboot delay" setting that can be increased if necessary to ensure the device turns off before turning it back on.

[Anonymous user · Accepted solution]

Faulty electrical relays are the ones not responding at expected milliseconds of time.

Under such issues UPS become slow to respond, and anything connected will suffer from power loss.