I also think this is getting out of hand.
This script is supposed to be a simplistic subsystem providing the
minimum. Intentionally.. Adding this option today, another option
tomorrow, eventually this turns into a ${} scribble-fest. And somewhere
around the time the script doubles in size, bugs start being introduced
accidentally. The dominant strategy against that future is to keep it
simple, and that is why this was kept simple and with one defined
purpose.
How about anyone who wants something different write their own seperate
script and maintain on their own?
Ingo Schwarze <schwa...@usta.de> wrote:
> Hi Nick,
>
> Nick Holland wrote on Sun, Oct 27, 2019 at 02:27:46PM -0400:
>
> > In version 1.78 of /etc/daily, the -i flag was added to the df output.
> > Apparently, some people run out of inodes.
> >
> > I only seem to run out of disk space, and too often, my eye skims
> > the daily report from a machine, looks at the last column,sees a
> > small percentage, and I decide, "all is good", even if I were
> > look a couple columns in, the actual disk space is low.
> >
> > To try to avoid bikeshedding and flopping this back and forth,
> > I offer this diff. With no change, daily df output is unchanged.
> > Those of us that don't worry about running out of inodes, we can
> > set DF_FLAGS in /etc/daily.local to be whatever we want, in my
> > case, I like "-hl" (currently, it's "-ikl")
>
> I don't object to the idea, but before commit, you would also have
> to document it in the daily(8) manual page.
>
> That said, i'm not enthusiastic either. I think sending mail every
> day even if all is obviously fine is not really useful in the first
> place; it is annoying even if you only have a handful of machines,
> and it doesn't scale at all for people supervising many machines.
> So i tend to think that VERBOSESTATUS=0 is the only setting that
> really matters in practice (by the way, why isn't it the default?)
> and in that case, your diff changes nothing.
>
> A real improvement that i would like to see if it can be done without
> excessive complication would be to print df(1) output if and only
> if it looks suspicious, instead of wrapping it in a dubious
> VERBOSESTATUS conditional.
>
> Yours,
> Ingo
>
> P.S.
> Do not send attachments, inline is just fine.
>
>
> > Index: daily
> > ===================================================================
> > RCS file: /cvs/src/etc/daily,v
> > retrieving revision 1.93
> > diff -u -r1.93 daily
> > --- daily 9 Sep 2019 20:02:26 -0000 1.93
> > +++ daily 27 Oct 2019 18:03:18 -0000
> > @@ -44,6 +44,10 @@
> > start_part "Running daily.local:"
> > run_script "daily.local"
> >
> > +if [ -z "$DF_FLAGS" ]; then
> > + DF_FLAGS="-ikl"
> > +fi
> > +
> > next_part "Removing scratch and junk files:"
> > if [ -d /tmp -a ! -L /tmp ]; then
> > cd /tmp && {
> > @@ -140,7 +144,7 @@
> > if [ "X$VERBOSESTATUS" != X0 ]; then
> > echo ""
> > echo "disks:"
> > - df -ikl
> > + df "$DF_FLAGS"
> > echo ""
> > dump W
> > else
>
