Date: Fri May 02 2003 - 12:30:48 CDT

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Fri, 2 May 2003 Victor.Duchovnimorganstanley.com wrote:

> On Fri, 2 May 2003, John Farnsworth wrote:
>
> > Hopefully this will be a much simpler question: I have a pipe calling a
> > filter script (simple, based upon the sample filter script). Script seems
> > to be processing ok, and all the return values from inside it, and it are
> > 0. In trying to figure this out, I dug through pipe_command.c, and
> > noticed it looks like it has an internal corruption handler. Does anyone
> > know what causes this to 'trigger'? If i'm understanding this message
> > right, it means the script exited with return value 0 (errno 0 from
> > vstring_sprintf expansion of %m), retval is not EPIPE (32) so that
> > condition wouldn't hit, so the only one left is write_status, which
> > appears to be set from mail_copy's 'corruption' tests. The filter worked
> > as-is in 1.1.10, i'm wondering if something went foul when compiling (OS
> > is SGI IRIX 6.5). The setup is based off of the Advosys paper
> > (http://advosys.ca/papers/printable/postfix-filtering.html). Any thoughts
> > as to what conditions can cause the pipe handler to think something is in
> > error?
> >
> > May 2 03:15:45 6C:challenger postfix/pipe[31308]: 635568460B9:
> > to=<sichallenger.darkness.nu>, orig_to=<sidarkness.nu>, relay=filter,
> > delay=175, status=deferred (Command failed: Error 0:
> > "/usr/local/anomy/filter.sh")
> >
>
> Please show all log entries for queue id 635568460B9 and any other log
> entries from pipe process 31308 near the event in question, over-trimming
> the logs in an effort to save bandwidth can raise the final bandwidth use.
> Especially show (panic|fatal|error|warning): log entries if any.
>
> The most likely cause is that the command timed out, but without the other
> log entries, one can only guess.
>
> --
> Viktor.
>

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]