[Nut-upsuser] Synology NAS is shutting down Ubuntu servers after very brief power outage (fwd)

Roger Price roger at rogerprice.org
Fri Aug 14 21:52:23 BST 2020


On Fri, 14 Aug 2020, Todd Benivegna wrote:

> https://forum.synology.com/enu/viewtopic.php?f=19&t=73960&hilit=ups+slave 
>
>       I have the latest version of the Windows port of NUT (A UPS management package) installed on my Windows 7 machine. The UPS I am using is a
>       very old APC BK650M which is controlled and sends its status via a special serial cable to the Windows machine.
> 
> I also have my Synology NAS (which uses NUT for its UPS management) connected over the local network as a slave to the Windows NUT master. The
> Synology slave is configured to immediately go into "safe mode" whenever an FSD (system shutdown message) or OB+LB (On-Battery and Low Battery)
> indication is received from the Windows NUT master.
> 
> The Windows master machine responds correctly to OB+LB events, correctly shutting down the master machine. The Synology box also responds to the
> OB+LB events from the master correctly and goes into "safe mode", the expected behavior.
> 
> The problem comes when using the command "upsmon -c fsd" on the Windows NUT master to force a UPS system shutdown to test the system. The Windows
> master shuts down correctly, but the Synology NAS slave does not go into "safe mode" when it receives the FSD notification.
> 
> I was wondering if there is a bug in the Windows port of NUT such that the FSD notification is not sent to the slaves when "upsmon -c fsd" is
> executed.
> 
> I then looked into the Synology NAS NUT configuration more closely.
> 
> When configured as a NUT slave device, the Synology responds to an FSD notification from the NUT master by calling a script defined in the
> Synology's /usr/syno/etc/ups/upssched.conf with an "fsd" argument. The script is /usr/syno/bin/synoups.
> 
> Before causing the NAS to go into "safe mode", the script first retrieves the status of the NUT master with upsc. If the status is "OL" (on-line),
> the script aborts any action for the received FSD notification. Since the UPS is actually on-line and not in a low battery state when the test FSD
> is sent from the Windows master, the Synology will not enter "safe mode" - it does nothing.

Not quite true.

Script /usr/syno/bin/synoups calls command /usr/sbin/upsmon -c fsd and then 
calls function UPSSafeMode fsd.  This function reads the UPS status ups.status 
into variable $St.

If $St is exactly "OL" the function logs "UPS is On-Line" and exits.

But if $St = "OL FSD" then the function goes into Safe Mode, stops all services 
and unmounts volumes.

> If the device was in OB (on battery), LB (low battery), or OB+LB mode, the OL check would fail and the Synology would enter "safe mode", as
> expected.
> 
> I tried commenting out the OL status check in the Synology script. The Synology slave then enters "safe mode" when a "upsmon -c fsd" is issued from
> the Windows master, as expected.
> 
> Interesting that Synology disables the test functionality on the NAS.
> 
> Roger, do you know where in /usr/syno/etc/ups/upssched.conf the OL status check is as mentioned above?  I'd like to comment that out and see if it works. 
> I'd really like to test everything out without waiting for the battery to run out, but if not then I will do that; will just take about 45 minutes or so. 
> I would never have guessed that it would do a check like that, which basically prevents the "upsmon -c fsd" test!

In /usr/syno/bin/synoups function UPSSafeMode you need to change

 	if [ "$St" = "OL" ]; then
 		SYSLOG "WARNING: UPS is On-Line"
 		return

to something like

         if [[ "$St" =~ "OL" ]]; then
 		SYSLOG "WARNING: UPS is On-Line"
 		return

This assumes that the NAS supports the =~ operator of Bash 4.  To reduce the 
long wait: in the NAS you need to execute

  upsrw -s battery.charge.low=85 -u <upsuser> -p sekret <myups>

Better is to set this value in the NAS ups.conf.

Roger


More information about the Nut-upsuser mailing list