Thank you Eric,

installing this version seems to fix the problem I have been having.

Sorry for not reporting earlier: Gmail decided to put this email
thread to my spam folder and I did not notice your replies until I
started wondering if I am on the list anymore :(

Only issue I have noticed with the upgrade was that my qmail-queue
symlink was pointing to qmail-dk and not to qmail-queue.orig like
before the upgrade.

Best,
Peter

On Thu, Apr 26, 2018 at 5:53 PM, Eric Broch <[email protected]> wrote:
> Hi Peter,
>
> This is not a Big DNS failure. It's a problem with CNAME lookup, and
> qmailtoaster is patched with the Big DNS patch.
>
> Dan Bernstein recommended the removal of the CNAME lookup portion of the
> code (patch below), which I did, a function in dns object module which is
> called by qmail-remote. There are new binaries with this patch at the
> following locations for respective CentOS version:
>
> CentOS 7:
> ftp://ftp.qmailtoaster.org/pub/repo/qmt/CentOS/7/development/x86_64/qmail-1.03-2.2.qt.el7.x86_64.rpm
>
> CentOS 6:
> ftp://ftp.qmailtoaster.org/pub/repo/qmt/CentOS/6/development/x86_64/qmail-1.03-1.1.qt.el6.x86_64.rpm
>
> CentOS 5:
> ftp://ftp.qmailtoaster.org/pub/repo/qmt/CentOS/5/development/i386/qmail-toaster-1.03-1.3.23.i386.rpm
>
> ftp://ftp.qmailtoaster.org/pub/repo/qmt/CentOS/5/development/x86_64/qmail-toaster-1.03-1.3.23.x86_64.rpm
>
> Eric
>
> CNAME lookup remove patch:
>
> <qmail-rm-cname.patch>
>
> --- qmailqmt/dns.c      2018-01-21 09:03:56.201694493 -0700
> +++ qmailqmt-new/dns.c  2018-01-21 09:06:40.696619489 -0700
> @@ -249,32 +249,7 @@
>  int dns_cname(sa)
>  stralloc *sa;
>  {
> - int r;
> - int loop;
> - for (loop = 0;loop < 10;++loop)
> -  {
> -   if (!sa->len) return loop;
> -   if (sa->s[sa->len - 1] == ']') return loop;
> -   if (sa->s[sa->len - 1] == '.') { --sa->len; continue; }
> -   switch(resolve(sa,T_CNAME))
> -    {
> -     case DNS_MEM: return DNS_MEM;
> -     case DNS_SOFT: return DNS_SOFT;
> -     case DNS_HARD: return loop;
> -     default:
> -       while ((r = findname(T_CNAME)) != 2)
> -       {
> -        if (r == DNS_SOFT) return DNS_SOFT;
> -        if (r == 1)
> -         {
> -          if (!stralloc_copys(sa,name)) return DNS_MEM;
> -          break;
> -         }
> -       }
> -       if (r == 2) return loop;
> -    }
> -  }
> - return DNS_HARD; /* alias loop */
> +   return 0;
>  }
>
>  #define FMT_IAA 40
>
> </qmail-rm-cname.patch>
>
>
>
> On 4/24/2018 12:48 AM, Peter Peltonen wrote:
>
> No ideas? From the archives I can see others have been struggling with
> the same issue...
>
> Peter
>
> On Wed, Apr 18, 2018 at 6:52 PM, Peter Peltonen
> <[email protected]> wrote:
>
> I am getting this error when sending to the tyks.fi domain:
>
> 2018-04-18 18:15:18.787618500 starting delivery 32313: msg 2232943 to
> remote ***@tyks.fi
> 2018-04-18 18:16:01.777845500 delivery 32313: deferral:
> CNAME_lookup_failed_temporarily._(#4.4.3)/
>
> I've been searching for this error and found the following:
>
> 1) known error for qmail + bind combination
> 2) fix is to either patch with qmailtoaster-big-dns.patch or use other
> recursor than bind
>
> Before I can proceed with 2) I have some questions though:
>
> * Why is the patch not installed by default in the toaster? I can see
> Shubert had it:
>
> https://github.com/QMailToaster/qmail/blob/master/qmailtoaster-big-dns.patch
>
> * As I understood it, the problem is a response too big that BIND
> cannot handle. I am a bit confused here, as the tyks.fi lookup does
> not return a big response and it does not have any CNAME records in
> it. Could this error be caused by something else?
>
> * If I would to change resolveer and not to patch, can I both run a
> BIND server and have a different resolver at the same time on the same
> server?
>
> Best,
> Peter
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>
> --
> Eric Broch
> White Horse Technical Consulting (WHTC)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to