[Orca-users] Stopping 'file did exist and is now gone' message(s).
Blair Zajac
blair at orcaware.com
Sat Nov 23 17:24:03 PST 2002
"isaac r." wrote:
>
> --- Blair Zajac <blair at orcaware.com> wrote:
> > >
> > > How about using the localtime call and rewriting
> > the patch?
> >
> > Actually, thinking about it more, what I'd like Orca
> > to do is
> > to check if there are compressed versions of the
> > filename (i.e.
> > check if filename.{Z,gz,bz2} exists and if it was
> > created in the
> > last X minutes. Otherwise, complain. This way, if
> > there is a
> > real error around midnight, then it won't be
> > ignored. In fact,
> > most errors could occur at midnight if something
> > goes wrong with
> > the compression of the file.
>
> I thought about this; the algorithm would be indeed
> different if examination of compressed files would be
> addressed. One would be faced with verifying the
> directory where the compressed file lives (or assuming
> its the same directory).
Unless orcallator.se was modified, it compresses into the same
directory where it's output files are.
>
> But your point is well taken - we don't want real
> midnight-related errors to be ignored by a simple
> 'hour' check.
>
> If you are choosing a compression, you would have to
> verify the directory, make some assumptions about most
> commonly known compressed versions of the file.
If only one suffix was being checked, it wouldn't be much work to check
against the three most commonly used suffixes (.Z, .gz and .bz2).
>
> Any way you slice it, you're back at the real problem
> of
> no longer having your raw data file that your file
> descriptor points to.
As an FYI, until Orca closes the file, it reads till the end of
it, since Unix won't delete it until all open file descriptors
are closed.
> As time allows, I'll work something out. But as we all
> know, a good hack is ...just a good hack, until it
> gets a higher priority in the neverending queue of
> daily routines ....
So true.
Best,
Blair
--
Blair Zajac <blair at orcaware.com>
Web and OS performance plots - http://www.orcaware.com/orca/
More information about the Orca-users
mailing list