[Nut-upsdev] [nut-commits] svn commit r1837 - trunk/clients

[Arnaud Quette]

guys,

you seem to be standing on your position, each one with valid points.
how about adding another option to close this log?
that way, you satisfy both needs without breaking the existing behavior by
default...

Arno

2009/5/15 Daniel O'Connor <doconnor at gsoft.com.au>

> On Fri, 15 May 2009, Arjen de Korte wrote:
> > Citeren Daniel O'Connor <doconnor at gsoft.com.au>:
> > >> Well, one thing is that it *will* cause existing installations to
> > >> break, where upslog is started as 'root' and the NUT user doesn't
> > >> have write permissions to the log file. Obviously, send upslog a
> > >> SIGHUP in such case would effectively kill it (because it can't
> > >> reopen the log), but I know of at least one distribution (SuSE)
> > >> that had a default setup exactly like that. We've seen people
> > >> complain about this behavior before, so in all likelihood, it is
> > >> being used in the field.
> > >
> > > That seems broken anyway, either it expands forever and you run out
> > > of space or you try and rotate it and upslog stops.
> >
> > If you log a few dozen bytes to the logs once every five minutes
> > (which boils down to about 1 kB/hr), realistically this may never
> > happen, so it is questionable that this is ever going to cause a
> > problem.
>
> <shrugs> depends on your /var.
>
> If it WAS a small flash embedded system 60Mb a year would not be pocket
> change. We still have some systems in the field with 256Mb /var's that
> are >50% full.
>
> > > IMO the SuSE package maintainer should change it to either run
> > > upslog as root or make a new directory iun /var/log owned but the
> > > NUT user..
> >
> > The SuSE package maintainer did the only sensible thing here and
> > removed upslog from the default startup script and leave the
> > configuration to the user. One shouldn't run upslog without a reason.
>
> I guess so, but sensible defaults are a good idea.
> (ie you shouldn't have to do more than enable it and pick the format to
> get it running).
>
> --
> 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 --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/nut-upsdev/attachments/20090515/ea8faaa3/attachment.htm>