[Nut-upsuser] NUT Client Shuts Down After Brief Power Loss

[Jim Klimov]

On August 9, 2017 4:53:11 AM GMT+02:00, Charles Lepple <clepple at gmail.com> wrote:
>On Aug 8, 2017, at 6:34 PM, Todd Benivegna <todd at benivegna.com> wrote:
>> 
>> OS name and version:  MacOS 10.9.5 on 2011 Mac mini
>> NUT version:  2.7.4-1
>> NUT installation method:  Package installed via FINK and
>FinkCommander
>> Exact device name:  APC Back-UPS NS 650M1 (brand new)
>> 
>> Issue:  
>> 
>> I have the NUT client installed and running properly; two processes
>running as root and the other as regular user processes.  The NUT
>server is my Synology NAS (DS416).  Client is communicating with server
>(confirmed by running "sudo upsc UPS@[Your_IP]”).  It receives events
>when I manually pull the power plug and shuts down properly when I do a
>forced shutdown (by doing “sudo upsmon -c fsd” via telnet on the NAS).
>> 
>> The problem is when there is an actual brief power outage of a few
>seconds, maybe 5 - 10 seconds max.  The NAS stays powered up, however
>the Mac mini (NUT client) shuts down.  When booted back up it
>immediately shuts down again.
>
>When it is in that state, what do you get from "upsc UPS@[IP]
>ups.status"? (Not a big deal, but upsc doesn't need root privileges -
>it just connects over a TCP socket.)
>
>> It will keep shutting down as soon as it boots up in MacOS and will
>keep doing that until I reboot the NAS.  I’m not sure if the problem is
>the client or server, however I’ve been talking to Synology Tech
>Support and this was their last message:
>> 
>> "Can you please check the NUT settings on your Clients and make sure
>that there is enough of a delay/wait period after it gets the signal.”
>> 
>> What delay are they talking about?  I have looked through the
>upsmon.conf and am not sure if anything there is the appropriate delay
>to adjust.  Does that even seem right?  Is the problem with the client
>or the server?  I can paste my upsmon.conf and any logs if necessary. 
>Any help would be greatly appreciated.  Thank you.
>
>Unless the ups.status value is "OB LB", a basic upsmon configuration
>won't shut down a client system for just a brief excursion onto battery
>power. Beyond that, things get complicated, so we would need to see how
>things are configured. Be sure to blank out any passwords, and you can
>also strip out lines with comments in the configuration ("grep -v '^#'"
>or something similar).
>
>Be sure to CC the list on the reply - other people on the list are more
>familiar with the "shut down after N minutes" style of shutdown than I
>am.
>_______________________________________________
>Nut-upsuser mailing list
>Nut-upsuser at lists.alioth.debian.org
>http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/nut-upsuser

It may also be that the FSD flag stays on. I saw this behavior in larger systems, where the UPS is expected to provide enough juice for graceful shutdown of business software that can take ages - so timing/percentage configs in NUT and in UPS-side FSD are tuned to that. 

After an outage there is little left in the battery so it can not guarantee the, say, 15 mins - so it shuts down the server which came up too early. By design (of such setup, anyway), which may be inconvenient when loads with different requirements about shutdown timing and guarantees are mixed into a single most restrictive setting.

See if this may be the case? E.g. the flag gets raised and stuck on NAS upsd (or perhaps UPS itself, afraid of unstable wall power)?

Hope this helps,
Jim Klimov
--
Typos courtesy of K-9 Mail on my Android