[Nut-upsdev] upsd flapping in the breeze

Daniel O'Connor doconnor at gsoft.com.au
Wed May 27 13:52:36 UTC 2009


On Wed, 27 May 2009, Arnaud Quette wrote:
> > Have you also tried the newmge-ups driver? Being based on the same
> > core as the usbhid-ups (USB) driver, it may be a little more
> > forgiving in case of timeouts (although I'm not very familiar with
> > it's predecessor, mge-shut).
>
> CAUTION: newmge-shut has still a major issue I haven't yet solved
> (which explains why it has not yet superseded): a loss of
> synchronization tends to happen.

OK.
Does it recover, or just stay out of sync?

> The SHUT protocol is *very* verbose and uses a slow 2400 baud link,
> so you
>
> > want to give it lots of time before giving up (reading a 1500+
> > bytes report descriptor will take more than 10 seconds). You may
> > want to increase MAXAGE in upsd.conf to at least double that, to
> > prevent PINGing the driver to death.
>
> I still have a point in my TODO list which is the baudrate
> negotiation. newer units supports it.
> there are also rooms for improvement on the startup sequence.

Sounds like it would be nice :)

> Daniel: I'm interested in debug trace of these startup failure. just
> to see how it fails, and what should be done.

OK, I'll have to see - I assume mge-shut -D -D -D is what you need?

I'm not sure when I'll be able to test it though since the code could 
potentially leave a remote system shut down with no way to turn it back 
on :(

-- 
Daniel O'Connor software and network engineer
for Genesis Software - http://www.gsoft.com.au
"The nice thing about standards is that there
are so many of them to choose from."
  -- Andrew Tanenbaum
GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 188 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.alioth.debian.org/pipermail/nut-upsdev/attachments/20090527/643b0086/attachment.pgp>


More information about the Nut-upsdev mailing list