Strange message !

[Glenn Steen]

On 11/03/2008, Rick Cooper <rcooper at dwford.com> wrote:
> Well I guess I should have read the next one before my last post. Actually
>  that line is commented by default. And yes, while mailwatch is doing
>  something during or just before the scan output (which comes after the scan,
>  from the parser) the trace log hasn't been opened properly so the output is
>  headed to, I think, stdout (might be stderr as that is redirected iirc) and
>  it's getting fed to the parser as clamd output.
>
>  If the original poster would comment out the trace line in MailWatch.pm I
>  believe his issues would go away too. There are a lot of incidental warnings
>  and undef errors that DBI doesn't choke on unless -w is used and don't think
>  MailScanner uses that switch a all, or at least not for the most part. The
>  trace log would show those errors however.
>
>
>  Rick
>

A "fun" addendum here is that since the tracing is at the DBI level,
it affects all systems that use that... So teh SA cahce will also
start barfing things into the trace.
So... Moral of this story is: Never do DBI-tracing unless you actually
have a DBI-related problem. If the file is writable (which that path
would be on most Linux systems, but perhaps not on most FreeBSD ones?
Even if run as user postfix...), it'll fill up pretty fast, so not
that good tohave enabled.

(snip)

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se