[Nut-upsuser] help with CyberPower 425HG

[James Northcott / Chief Systems]

On 11-10-16 01:51 PM, nut at johnea.net wrote:
> On 2011-10-16 12:06, Ron wrote:
>> Rebooting worked for me too.  Wish I had read this post earlier
>>
> Hi Ron,
>
> I've been struggling with nut and CyberPower USB devices for over a year now.
>
> You may find connection to the UPS on boot to be unreliable, sometimes it will
> connect, other times it won't.
>
> > From the thread: "Re: [Nut-upsuser] USB CyberPower"
> -----------------------------------------------------------
> Hi James,
>
> I've worked extensively on getting a similar CyberPower UPS (same IDs) to play
> well with nut.
>
> The conclusion I ultimately came to is that if this UPS is not connected to by
> it's driver within about 20sec, it drops off the USB bus and re-enumerates. This
> causes an endless cycle of the UPS attaching and detaching. In archlinux, I was able
> to make it work by placing the nut daemon at the beginning of the rc scripts,
> before everything else. This caused it to be attached to before it dropped from
> the bus. Once it was attached, everything worked.
>
> IMHO, If you can, return it and get a different unit.
> -----------------------------------------------------------
>
> As the text above indicates, the time between the system powering up, and the
> driver trying to connect to the UPS is critical. If this time is too long
> (greater than about 20sec) the UPS will have already disconnected from the USB
> and the connection will fail. Thereafter the UPS will continuously re-enumerate,
> wait about 20sec then disconnect again, and continue an endless cycle of
> re-enumerate/wait 20sec/disconnect.
>

To add some more information, this cycle (re-enumerate/wait) will also 
occur if the driver is unloaded for any reason.

This means that the stock scripts for shutting down in case of power 
failure won't work on CyberPower UPSs, at least they didn't for me.  I 
had to adjust my shutdown scripts so that nut is never shut off, and the 
nut driver process is never sent the TERM/KILL signals during shutdown.  
Then, the UPS shutdown command works correctly.  Otherwise, as soon as 
the driver is unloaded the UPS goes into that cycle, and you cannot send 
a UPS command, so the machine shuts down normally, which is not what you 
want.

> One major factor in the length of the delay on power up, was my computer's DHCP
> delay. I found if I could get the nut driver to load prior to my computer going
> out for DHCP, then it would reliably connect on boot.
>

This matches my experience too, I actually moved NUT to the very first 
item in the startup sequence.

> If you have a static Network IP address, then this DHCP delay is not a factor
> for you.
>
> Anyway, try to connect to the UPS first, before any other boot time activity.
>
> Thanks for your post! Please pass along your ongoing experiences...
>
> johnea
>
> _______________________________________________
> Nut-upsuser mailing list
> Nut-upsuser at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/nut-upsuser
>