From roger at rogerprice.org  Tue Dec  2 09:15:18 2025
From: roger at rogerprice.org (Roger Price)
Date: Tue, 2 Dec 2025 10:15:18 +0100 (CET)
Subject: [Nut-upsuser] Debian 13 (NUT 2.8.1) fixes upssched bug,
 but floods systemctl log
In-Reply-To: <CAJYg8vJk45NYOKMbjNzreqzayVjH0CCiDrCcUruCy7GJQ6=Avg@mail.gmail.com>
References: <fcb71f5d-7d87-8833-a794-f6b24a892602@rogerprice.org>
 <rmicy51uhl7.fsf@s1.lexort.com>
 <6eb8494f-e531-652f-f637-1dca6c45a4b7@rogerprice.org>
 <CAJYg8vJ4vGHno75dVh-X+nCqBYPr1-TU8UvORhESGpLyYKRT0w@mail.gmail.com>
 <befb53ac-dec2-a814-c460-4a62c1837a91@rogerprice.org>
 <CAJYg8vJk45NYOKMbjNzreqzayVjH0CCiDrCcUruCy7GJQ6=Avg@mail.gmail.com>
Message-ID: <7556ac6d-ad1b-75ce-6d6f-94b045d73e08@rogerprice.org>

On Sun, 30 Nov 2025, Jim Klimov wrote:

> Maybe then use double-quotes around the command, and escape the inner ones for 
> grep args. Or make a wrapper script in a separate file with equivalent 
> content, and call it as ExecStart.

I created a separate file /usr/local/bin/nut-monitor.ExecStart.sh

 #! /bin/bash -u
 # The following filter prevents upsmon from flooding the journal with
 # banner messages every 300 sec when upssched restarts the heartbeat.
 /usr/sbin/upsmon -F | grep -vE "Network UPS Tools upsmon 2\.8\.[012]"

In /usr/lib/systemd/system/nut-monitor.service I set 

 ExecStart=/usr/local/bin/nut-monitor.ExecStart.sh

I ran 

 systemctl daemon-reload
 systemctl restart nut-monitor 
 systemctl status nut monitor

and saw

     CGroup: /system.slice/nut-monitor.service
             ??10435 /bin/bash -u /usr/local/bin/nut-monitor.ExecStart.sh
             ??10438 /lib/nut/upsmon -F
             ??10439 grep -vE "Network UPS Tools upsmon 2\\.8\\.[012]"
             ??10440 /lib/nut/upsmon -F
             ??10444 /usr/sbin/upssched "UPS heartbeat at localhost: On battery"

There is no more flooding with banners.

I guess that I am one of the very few users of a heartbeat UPS, so this local 
fix is good enough.

Roger