Re: can't read SMFIC_BODY reply packet header: Success

From: Murray S. Kucherawy <msk_at_blackops.org>
Date: Thu, 24 Nov 2011 16:57:00 -0800 (PST)

Hi Rolf,

On Fri, 25 Nov 2011, Rolf E. Sonneveld wrote:
> Nov 25 00:00:52 vds postfix/cleanup[19573]: warning: milter
> inet:localhost:9999: can't read SMFIC_BODY reply packet header: Success
> Nov 25 00:00:52 vds opendkim[18165]: terminated with signal 6, restarting

Signal 6 is SIGABRT, which is usually an assertion failure inside the
code. This is typically a "should be impossible" scenario, which means
there's something happening we've not seen before. It might also be a bug
in one of the subsidiary libraries we'll have to work around somehow (one
of these just came up with libdb yesterday).

Did it produce a core dump from which you can get a stack trace? If not,
you can use EnableCoreDumps to request such.

If it's reproducible, can you produce a message that triggers it?

> What does:
>
> warning: milter inet:localhost:9999: can't read SMFIC_BODY reply packet
> header: Success

That's typically an error that indicates a conflict between the milter
version postfix is using and the one with which the filter was compiled.
When it's coupled with SIGABRT though, it might be a side effect of the
crash.

> And does OpenDKIM crash and restart during this session? (it seems it does?)

It looks like it did, and that maches the configuration you included.

-MSK
Received on Fri Nov 25 2011 - 00:57:15 PST