RE: internal error from libopendkim

> -----Original Message-----
> From: opendkim-users-bounce_at_lists.opendkim.org [mailto:opendkim-users-bounce_at_lists.opendkim.org] On Behalf Of Rolf E. Sonneveld
> Sent: Monday, November 15, 2010 1:29 PM
> To: opendkim-users_at_lists.opendkim.org
> Subject: internal error from libopendkim
>
> Hi,
>
> running:
>
> opendkim: OpenDKIM Filter v2.1.3
> Compiled with OpenSSL 0.9.8e-fips-rhel5 01 Jul 2008
> Supported signing algorithms:
> rsa-sha1
> rsa-sha256
> Supported canonicalization algorithms:
> relaxed
> simple
> Active code options:
> USE_ARLIB
> USE_DB
> _FFR_DKIM_REPUTATION
> _FFR_DNSUPGRADE
> _FFR_REPORT_INTERVALS
> _FFR_STATS
> _FFR_VBR
> libopendkim 2.1.3: dkim_reputation dnsupgrade libar
>
> (Yes, hopefully soon to be upgraded)
>
> Today I noticed quite a number of DKIM error messages in my mail log
> file.
>
> Nov 15 21:46:43 hydrogen opendkim[3712]: 0LBY00O651PU6U00: dkim_eoh():
> internal error from libopendkim: ar_addquery() for
> `cs._domainkey.cs.tcd.ie' failed
> Nov 15 21:46:50 hydrogen opendkim[3712]: 0LBY0003S1Q1SG00: dkim_eoh():
> internal error from libopendkim: ar_addquery() for
> `cs._domainkey.cs.tcd.ie' failed
> Nov 15 22:17:05 hydrogen opendkim[3712]: 0LBY00O6D34G6U00: dkim_eoh():
> internal error from libopendkim: ar_addquery() for
> `cs._domainkey.cs.tcd.ie' failed
> [...]

Oops, I already replied to this privately. So, sorry, Rolf, for the duplicate.

Are you able to attach a debugger to it to get some other details? Also, can you use "ps" to see what the memory size of the process is?

I'm working on a bug report from another site using libar, as you are; in his case, a deadlock was reported.

I'm also running it in my home environment hoping to see it reproduced, but so far I haven't managed to see it. That could be because I don't get enough volume.

I'm reviewing the code for libar/ar.c right now and although there are a few minor issues I've found, none of them explain the conditions being reported.

If you are up to the debugger task, I'd love to see a breakpoint set on ar_addquery() followed by "next"ing through that function to see where it's failing. Then I can provide some additional commands to dig deeper.

The ar_addquery() function fails in the way described for any of several reasons, from invalid arguments (e.g. a negative timeout value or an impossible query string) to a malloc() failure. It can also fail if the resolver has given up talking to the nameserver because of a complete shutdown in communication (e.g. the nameservers have all gone away).
Received on Mon Nov 15 2010 - 21:48:25 PST