Re: SMTP Looping and RBL problem (separate issues)

[Matti Aarnio <mea@nic.funet.fi>]

On Tue, Apr 03, 2001 at 07:06:47PM -0300, zmailer wrote:
> Hi,
> 2 things we found today using 2.99.54 on Solaris 2.6.
...
> SMTP Looping:
> Today we found a spike on our DNS server (more or less 300 queries 
> per second). All the queries were for the A record of the same name.
> 
> The problem was an smtp process. The message being processed was 
> for XX@qmx.net (apologies for publishing the domain, but it may help)
> The MX for this domain (ICE.QAX.net) didn't have an A record.
> 
> After enabling verbose we have the text at the end.
> this are just a couple of lines from a file which grove up to 3Mb in less
> than a minute.
> 
> BTW: the same address tested with 2.99.54 bounce saying:
>   Diagnostic texts:
>     smtp; 500 (DNS: MX host does not support SMTP: qmx.net)

SMTP: Connecting to host: qmx.net firstmx=0 mxcount=? noMX=0
 getmxrr(qmx.net)DNS lookup reply: len=120 rcode=0 qdcount=1 ancount=1 nscount=2 arcount=2 RD=1 TC=0 AA=0 QR=1 RA=1
 -> (1981s) MX[0] pref=5 host=ICE.QAX.net
  mx[0] mxtype=--(0) host='ICE.QAX.net'
  getaddrinfo('ICE.QAX.net','0') (PF_INET) -> r=-2 (name or service is not known), ai=(nil)
...

	Damn..  Somehow my glibc (2.2.2-5) has again f*ed the
	gethostbyname2() return codes.  SERVFAIL should mean
	EAI_AGAIN (-3), not EAI_NONAME (-2)...

	Or then it is compiler problem somewhere.

	No, NSCD f*cs it.  Damn beast.
	Ok, so the code works correctly, and will correctly retry
	the message after a delay, etc.


> Nicolás Baumgarten
> Pert Consultores 
> nico@pert.com.ar

-- 
/Matti Aarnio	<mea@nic.funet.fi>