[Nut-upsuser] Early shutdowns and 2.6.5-1 package for Windows

FredericBohe at Eaton.com FredericBohe at Eaton.com
Mon Sep 17 09:23:56 UTC 2012

Hello Denis,

I have now understood what you mean about the event viewer and I have fixed it.
I will release a new binary package for Windows soon (hopefullt today) including all the fixes related to your reports (regression of upssched and regression of event viewer).

Thanks a lot.


Eaton Opensource Team - http://opensource.eaton.com



From: nut-upsuser-bounces+fredericbohe=eaton.com at lists.alioth.debian.org [nut-upsuser-bounces+fredericbohe=eaton.com at lists.alioth.debian.org] on behalf of Denis Serov [dns-srv at yandex.ru]
Sent: Saturday, September 15, 2012 3:23 PM
To: nut-upsuser at lists.alioth.debian.org
Subject: Re: [Nut-upsuser] Early shutdowns and 2.6.5-1 package for Windows

Hello Frederic!

>> upssched.conf
>> CMDSCRIPT "cmd.exe -C C:\\NUT\\1.cmd"

> I guess it's /C (not -C).

That is my misprint, slash is presented in original config file ;)

>> 1. NOTIFYCMD parameter is not specified NUT doesn’t start UPSSCHED
>> tool (as I remember, Linux version do it by default).

> upssched is called by upsmon when a NOTIFYFLAG set to EXEC happens. It is not started by NUT service (even on Linux).

I thought about "NUT" as product, certainly, UPSMON executes UPSSCHED =) But my original worry was wrong, Linux version does not start UPSSCHED as default tool too (if NOTIFYCMD parameter is not specified). Sorry for misconception.

>> Failed to connect to parent and failed to create parent: No error
>> [The system cannot find the path specified. ]

> In fact PIPEFN is ignored on Windows version, anyway you still must have something (anything) set for it.

>> There is no proper record in Event Viewer. 2.6.0-1 version writes different warnings and errors there. They helps to understand what is wrong.
>> That is quite difficult now because of total silence.

Well, I have meant something else. I sent a screenshot to you. You will find a lot of records which has been created by 2.6.0-1 version. Non-existing PIPEFN property was just a test to raise an error. Version 2.6.5-1 does not write to Windows Event Log anything except notifications about starting/stopping service/upsd/upsmon. That is not useful for troubleshooting.

>> got [] Timer daemon started
>> addchar: discarding invalid character (0x00)!
>> addchar: discarding invalid character (0x00)!

>> Sometimes I see the strings with garbage inside, like repeated using of released memory.

> Good point Denis.
> There is a regression on Windows version. It is related to a security issue fixed a while ago. Some Linux modifications did not make it to Windows code.
> Can you try this binary :

> fbohe.free.fr/upssched.exe

> Let us know if this fix your issue.

I have very good news: the early shutdowns are working properly. Only one instance of UPSSCHED (for my configuration) is started and stopped, no errors like "read confirmation failed, trying again read confirmation failed" and CMDSCRIPT is executed. Here is UPSMON output:
  15.153257     do_notify: ntype 0x0005 (COMMBAD)
  15.168857     Communications with UPS upsa at lost
  15.168857     Current power value: 1
  15.184457     Minimum power value: 1
Timer daemon started
New timer: upslost (2 seconds)
New timer: upslost (2 seconds)
Event: upslost

C:\NUT>echo upslost  1>C:\NUT\1.txt
Execute command "cmd.exe /C C:\NUT\1.cmd upslost" OK
Event: upslost

C:\NUT>echo upslost  1>C:\NUT\1.txt
Execute command "cmd.exe /C C:\NUT\1.cmd upslost" OK
  20.198341     pollups: upsa at
  20.198341     get_var: upsa at / status


Thank you very much,

Nut-upsuser mailing list
Nut-upsuser at lists.alioth.debian.org

More information about the Nut-upsuser mailing list