Smart UPS doesn't follow delayed turn on schedule on power restore

[Anonymous user]

I have a Smart UPS (SMX3000LV).  I have it configured to do a staggered power on of the 3 outlet groups to allow a controlled startup of all the connected networking equipment, file server, and other servers.  This works nicely on initial power up.

Here is my problem.

• On power failure, the connected server directs the UPS to shutdown
  • this also works nicely, and the UPS goes into a mode where it says "Waiting for AC" on the display.
• When power is restored, the UPS turns back on...
  • BUT... instead of following the specific delayed turn on that is programmed for each outlet group, it just turns them all on immediately!
  • There is nothing in the menus on the UPS that talks about a restart delay, or gives me anything else to program.

Is there something I'm missing in the configuration that tells the UPS to ignore it's programmed power on delays on restart?

Why is this behavior even possible... the delays are programmed for a reason, why would it ever ignore them on power-on?

Thanks in advance for any insight the forum might have.

[BillP]

Hi,

Are you configuring a power on delay for outlet under sequencing or min return runtime? Also, how is the UPS being commanded off? 

[Anonymous user]

The delay was actually programmed as a "Power On Delay", but it was programmed (initially) from the front panel of the UPS, not from the powerchute software.  "Min return runtime" is used, as I understand it, to set the amount of charge the batteries must have before the UPS will power the outlet group back on.

The UPS is being commanded off by the usbhid-ups driver from NUT (network UPS Tools) in Linux.  This is how it reaches the "Waiting for AC" state where the LEDs sequence back and forth.

When I connect the UPS to a windows computer running powerchute business edition agent/server/console, I can see all the configuration that was made on the front panel.  In this case, when shutdown is initiated, it seems to "sleep" in a different state where it says "Outlet Group 2 on by USB in: 120s", and just sits there waiting for AC power.  When AC power returns, the 120s starts counting down, and it does a properly sequenced power up.

So... using powerchute, the behavior is as desired.  Unfortunately, I have a rack of linux machines that need to be controlled (not just one machine), and powerchute for business on Linux doesn't appear to have a USB option.  I'll explore this more closely.  Everything I have natively runs NUT.  Not all of the machines will be able to run powerchute.

So this question boils down to... On the SMX3000LV, when it reaches the "Waiting on AC" sleep state, why does it not follow the power-on delays when AC power returns?  How do I alter that behavior?

[BillP]

Hi,

You can verify the UPS is working properly by configuring load shedding on the outlet group and have a power on delay set. Example under Configuration for Group1 set the power on delay for 120  Loadshed: Time on Battery: Enabled, Loadshed: Time on Battery: 120. 

What should happen is when the UPS goes on battery after 120 seconds the outlet group will be powered off. When AC is restored the outlet group will wait 120 seconds and then restore power to the outlet group. If the UPS works as expected than the issue is with the shutdown command NUT is using. Unfortunately I am not familiar with NUT but if there is a users group for that app or a forum they may be able to help. 

[Anonymous user]

Since the process works correctly with PowerChute, the issue is indeed with the command issue by NUT... or more specifically, the response my UPS has to the command that is issued by NUT.  I can argue that there is a bug in the firmware of the UPS that is allowing the UPS to power on its outlets, ignoring the power on delays that are programmed into the UPS, when AC is restored.

Since I don't expect a fix, either from the USB driver in NUT or the firmware of the UPS, I've changed my setup to remove the power on delays and instead allow everything to power up simultaneously and I handle the staggering of boot times in software on the respective machines.  Not as elegant, but it is functional and possibly more robust as well.

[BillP]

Russ,

If your SMX3000 has a network management card and if the Linux version is supported you have the option of installing PowerChute Network Shutdown. You would then configure the shutdown and start up via the outlet groups. This way you can use the delays set at the UPS. 

[Anonymous user · Accepted solution]

Thanks for sharing your thoughts on this Bill.  I don't have the network card.  I didn't buy it because it wasn't likely to be supported by my linux platform.

The linux server that is actually directly connected to the UPS is a "FreeNAS" installation.  This is a version of FreeBSD that is specifically built to be a file server (think NetApp).  The build is very thin, and everything is set up via a GUI on a webpage.  It is extremely robust and powerful, but you don't want to deviate from what can be achieved in the GUI or it becomes very hard to maintain.

Using NUT, because it's a supported built-in function of FreeNAS, is a mandatory starting point for my environment.  It's also nice that the NUT clients are available for just about every platform so I can put any type of machine on the UPS and slave it's behavior, via NUT, to the computer controlling the UPS shutdowns.

I'm happy enough with the solution I sorted out.  I just wanted to comment here to talk about the behavior of the UPS to encourage APC to make the behavior more standard going forward and follow the power-on delays under all circumstance where the outlets go from off to on.