I wanted to write a follow-up for a dkim-filter 2.4.0 crash report
that I sent a few weeks ago.

To recap, we were running dkim-filter 2.4.0 on Centos 5, with sendmail
8.13.  dkim-filter crashed very frequenly, sometimes as often as every
15 minutes.  I'd captured a set of core dumps that showed the crash
occurring (very consistenly) in openssl library calls.

That the crash-prone dkim-filter had been compiled on a machine with
openssl 0.9.7 (not 0.9.8), so, I recompiled a new version of
dkim-filter (version 2.4.2), linked against openssl 0.9.8b.

How are things working out with dkim-filter 2.4.2 and openssl 0.9.8b?
Better, but with hiccups.

We run dkim-filter on outbound mail relays; dkim-filter signs outgoing
mail, but does no signature verification.  These relays are
"newsletter boxes"; they send the same message to a list of
recipients, and the message bodies differ only slightly from recipient
to receipient.

We've done two mailings with the new dkim-filter.  During the first,
we had crashes, but fewer of them (a dozen or so).  During the second
mailing, dkim-filter was _completely_ stable, no crashes whatsoever.
Between the two mailings, the rate of delivery has been about the
same.

Unfortunately, I had core dumps disabled during the first mailing
(ulimit -c 0), so I don't have any stack traces.

Between the two mailings, the only real difference is message size.
In mailing #1 (crashes), we had multipart/alertnative messages of
around 51K.  In mailing #2 (no crashes), we had multipart/alternative
messages of around 5k.

I'll try to develop a test case that reproduces the crash, and if I
can get a stack trace, I'll post it.

Steve

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
dkim-milter-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dkim-milter-discuss

Reply via email to