[Nut-upsuser] TrippLite OMNIVS1500XL Segmentation Fault

Charles Lepple clepple at gmail.com
Sat Aug 11 15:12:51 UTC 2007


On 8/11/07, Junk Mail <junk at chassit.com> wrote:
> On Sat, 11 Aug 2007, Charles Lepple wrote:
>
> >> input.voltage: 99.33
> >> input.voltage.nominal: 120
> >
> > input.voltage looks a little low.
>
> Is this something to worry about? It seems to fluctuate right around 100.
> Could it be an error in the UPS?

Possibly an error in the UPS, or maybe our conversion is off. I only
did a sanity check for protocol 1001, and yours uses protocol 2001
(code for which was contributed by another user).

> >> upsd[12452]: Data for UPS [tripplite] is stale - check driver
> >> tripplite_usb[12450]: Reconnect attempt #1
> >> tripplite_usb[12450]: Successfully reconnected
> >
> > I forget how MAXAGE works, but if you can fix this by increasing
> > maxage, you probably want to figure out the time difference between
> > the first tripplite_usb error above, and the "Sucessfully reconnected"
> > message.
> >
> > Beyond that, debugging can get messy. It may not be worth it on these
> > models - they do occasionally miss sending back responses.
>
> I guess I shouldn't have trimmed the timestamps... those log messages all
> appear with the same timestamp... so they happen within one second. (and
> this is the same every time)
> Increasing MAXAGE has not helped. The period between failures ranges from
> about one minute to over an hour (and counting).

If you decide you do want to track it down, just run the driver with
the debug options like in your first post, then gzip and send the logs
to the list. As I mentioned, it seems to be some sort of mismatch
between what the UPS is sending back, and what we expect it to send
back.

> I'm not really concerned about the warnings... My only real need for nut
> is to shut down (several machines) cleanly.

You still may want to test the LB (low battery) signalling to be sure
that the shutdown works. It "should" work, but it never hurts to plug
your NUT master server into another UPS or the wall, then plug a dummy
load into the UPS that the master is monitoring, and pull the power
cord to the UPS. (Details are in the NUT documentation.)

Since we don't have official documentation for any of the Tripp Lite
protocols, I don't know how well the UPS shutdown command will work.
(On the OMNIVS models, they have to be on battery when you send the
shutdown command.) You may have to manually cycle power in certain
cases.

-- 
- Charles Lepple



More information about the Nut-upsuser mailing list