Re: Nut and RAID on FreeBSD 7.0

On Jan 11, 2008 7:19 AM, Derek Ragona <derek@xxxxxxxxxxxxxxxxxxxxxxxx> wrote:


At 01:43 AM 1/11/2008, Derrick Ryalls wrote:

On Jan 10, 2008 3:52 PM, Kurt Buff <kurt.buff@xxxxxxxxx> wrote:
>
> On Jan 10, 2008 3:14 PM, Bob Johnson <fbsdlists@xxxxxxxxx> wrote:
> > On 1/10/08, Derrick Ryalls <ryallsd@xxxxxxxxx> wrote:
> >
> > >
> > > Perhaps I need to re-evaluate my line of thinking. Light sometime
> > > flicker, but power almost never goes out. When it does it is either
> > > back on in less than 1 minute, or out for hours. If the UPS detects
> > > critical correctly and gives me at least a minute before death, then
> > > that should be plenty of time for the system to auto-shutdown. Guess
> > > I will have to do some experimentation tonight.
> >
> > While you experiment, keep in mind the following sequence of events:
> >
> > -- Power fails
> > -- UPS signals low battery
> > -- System shuts down
> > -- Power returns before UPS shuts itself down
> > --> System never reboots, because it never lost power.
> >
> > Getting around this is the tricky part. I haven't used NUT in about
> > seven years, but back then the recommendation was to shut down to
> > single user mode and run a script that delayed for some time longer
> > than the remaining battery life of the UPS, then rebooted the system.
> > There didn't seem to be an easy hook for running a script after
> > shutting down to single user mode (maybe there is now).
> >
> > I haven't looked at NUT recently, but I expect the various flags that
> > you are supposed to test are another way around this problem.
>

Trying to test out the scripts, I ran into a road block. I see that
upsmon is working and detecting the events I wanted to detect from
these sorts of entries in /var/log/messages:

Jan 10 23:28:57 frodo upsmon[80983]: UPS powercom@localhost on line power

Plus a similar message for going to battery power. However, the
notify executable is having issues and is dumping dozens of lines like
this in /var/log/messages:

Jan 10 23:28:09 frodo kernel: pid 81029 (upssched), uid 1005: exited
on signal 11
Jan 10 23:28:09 frodo kernel: pid 81031 (upssched), uid 1005: exited
on signal 11
Jan 10 23:28:10 frodo kernel: pid 81032 (upssched), uid 1005: exited
on signal 11
Jan 10 23:28:10 frodo kernel: pid 81033 (upssched), uid 1005: exited
on signal 11
Jan 10 23:28:11 frodo kernel: pid 81034 (upssched), uid 1005: exited
on signal 11
Jan 10 23:28:11 frodo kernel: pid 81035 (upssched), uid 1005: exited
on signal 11

I tried giving the user the user in question (nutmon) a shell of
/bin/sh instead of /sbin/nologin but that didn't help. Any clues on
how to fix this? Executing upssched from the command line it tells me
not to execute directly (similar to what the man page states), and
manually executing the upsched-cmd shell script does work and the
script itself uses full paths for commands.

What is in your notify command?

I set my NOTIFYCMD in upsmon.conf to a simple shell script I created to
send the message via sendmail, here is my script if that helps:

============================================
#!/usr/local/bin/ksh
#set -x
SENDMAIL=/usr/sbin/sendmail
MAIL=/usr/bin/mail
HOSTNAME=/bin/hostname

MYHOSTNAME=`$HOSTNAME -s`

echo $* | $MAIL -s "UPS Alert from $MYHOSTNAME"
upsmon@xxxxxxxxxxxxxxxxxxxxxxxx

================================================

-Derek


I was previously pointing it at the provided upssched process which is
supposed to handle timers and such and fire off upsshed-cmd which is a
provided shell script that I had modified. This is the 'documented'
way of doing it and gives the extra benefit of being able to add
timers so if the power goes out for 3 seconds, I won't be notified.
Using your script directly did work for lost/gained power events so I
guess I can go that route. I won't be able to do timers or anything,
but I will at least be notified when the power goes out. I'll be
checking the safe power down this weekend. Thanks.
_______________________________________________
freebsd-questions@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscribe@xxxxxxxxxxx"

Relevant Pages

  • Re: Nut and RAID on FreeBSD 7.0
    ... but power almost never goes out. ... -- UPS signals low battery ... There didn't seem to be an easy hook for running a script after ... Executing upssched from the command line it tells me ...
    (freebsd-questions)
  • Re: Nut and RAID on FreeBSD 7.0
    ... >> single user mode and run a script that delayed for some time longer ... >> shutting down to single user mode. ... Plus a similar message for going to battery power. ... Executing upssched from the command line it tells me ...
    (freebsd-questions)
  • Re: Power Management GPO?
    ... I suspect that the power management subsystem might not be started when the startup scripts runs or the SYSTEM context can't modify it, but I have some ideas about this and will do some testing later on today to work it out. ... I created a GPO and put the batch file in first as a startup script and when that didn't work I tried it as a logon script. ... A normal user can't change the power settings so a logon script like this won't work either. ... I can't remember if all power setting are per machine, but if some are per machine and some are per user then you could split it up into 2 scripts, one startup script and one logon script. ...
    (microsoft.public.windows.group_policy)
  • Re: Log repair from a script
    ... it's easy to do this via diag. ... there are some power tests. ... >> could not find a way to get this from the commandline or script. ... > Hallo ljm, ...
    (comp.unix.aix)
  • Re: RFC: powerd Patch & proposed future changes
    ... script on ac state change? ... battery falls below a configured threshold. ... A notebook ought to generate a battery event when capacity hits its factory set "warning" and "low" marks. ... As I've experienced with power saving, the moment you start doing more and more work to save power, you also end up consuming more power. ...
    (freebsd-current)