On Tue, 14 Apr 2009, Paul Heinlein wrote:

> First, if it won't run manually, then you've got a better chance of figuring 
> it out. Run it in strace to see if the error is visible:
>  strace -o /tmp/pflogsumm.trace /usr/local/bin/pflogsumm
> The output of strace is cryptic, but the failure should be visible.

Paul,

   I mis-communiated the location of the problem. /usr/local/bin/pflogsumm
runs. It is normally called by /etc/cron.daily/1pflogsumm which is a shell
script that defines the variables (e.g., time period, log file location) for
the perl script.

> I assume that the stuff in /etc/cron.daily is launched via a utility
> called run-parts. If so, check the manpage. The new Debian version of
> run-parts places restrictions on the names of scripts that can be run. My
> centrally managed cron scripts all have a .cron filename extension, but
> that violates the new restrictions so they wouldn't run on Debian. Wierd.

   My cron scripts do not have extensions. And the inconsitency when scripts
run and when they won't suggests the issue is not as simple as a run-parts
namespace change. Unfortunately.

> Fourth, you can always try putting a bunch of "print STDERR ..." stuff in
> your Perl script to see how far it gets. :-)

   T'ain't the perl script, but the shell script.

   After initially reading this message of yours, I manually invoked
/etc/cron.daily/1pflogsumm from within a virtual console. Today it ran and
mailed me the report. Yesterday it would not run when I tried the same
thing.

   Screwy, eh? Perhaps it's related to the state of the economy?

Rich
_______________________________________________
PLUG mailing list
[email protected]
http://lists.pdxlinux.org/mailman/listinfo/plug

Reply via email to