Exim slow on my ipv6 connection

Hm, thanks for trying.

So:

- my internet -> my server -> slow
- my internet -> other server -> fast
- work -> my server -> fast
- you -> my server -> fast

Starting to think there is a problem with my internet connection, however why are other servers fast? I've tried several other random IPv6 servers and everything is fast :(

edit:
Just thought of something.

Code: 
root@raspberrypi:/home/pi# date && telnet 2a05:e1c0::38 587
Mon 21 Nov 18:44:21 CET 2016
Trying 2a05:e1c0::38...
Connected to 2a05:e1c0::38.
Escape character is '^]'.
220 vps.mattie-systems.nl ESMTP Exim 4.87 Mon, 21 Nov 2016 18:44:41 +0100

The "Connected to ip" is fast, but the welcome message takes a long time. I can also type BEFORE the welcome message, and then after the "long wait" I got disconnected due to SMTP sync error:

Code: 
2016-11-21 18:45:34 SMTP protocol synchronization error (input sent without waiting for greeting): rejected connection from H=[2001:470:1f15:787:35f:76a4:ecfd:d12f] input="haha\r\nhaha\r\n\r\n"

So, I am connected in that case?
 
Last edited:
Did some more debugging, and I can no longer assume it is an ISP problem.

SERVER:
Code: 
(without exim running)
root@vps:~# nc -6 -l -p 587
test

CLIENT:
Code: 
root@raspberrypi:/home/pi# time nc -6 2a05:e1c0::38 587
test
^C

real    0m1.978s
user    0m0.000s
sys     0m0.030s

Now with exim running

Code: 
root@raspberrypi:/home/pi# time nc -6 2a05:e1c0::38 587
220 vps.mattie-systems.nl ESMTP Exim 4.87 Mon, 21 Nov 2016 19:02:17 +0100
^C

real    0m22.434s
user    0m0.010s
sys     0m0.010s

How can this be my ISP? It has to be SOMETHING with firewall/ratelimit/trying to resolve the rDNS (what should work btw)

Another update:
Set Exim to also accept poort 1337
From me -> SLOW
From other location -> FAST

So it is also not port number related....
 
Last edited:
So, I found the problem :D

I always assumed that the rDNS was correctly and yes it was for my SERVER, just not for my CLIENT. I got an entire /64 block from my tunnel provider and that is delegated to my own nameserver, so when I manually add a rDNS record for my client it works like a charm!

So:
Either I add an record for ALL ip's (my preference) see http://forum.directadmin.com/showthread.php?t=54027&p=277005
Or I'll disable the rDNS lookup assuming that is possible
 
Yeah, strange right :)

Think I'll disable the lookup (if possible). I can fix my own rDNS (works on powerdns with *) however what if some server connects without rDNS, I can't do anything on it and it will be slow in that case. As I can connect fast to other servers I'm guessing that it is "not bad" to stop resolving.
 
I guess is there more for security actually, i wouldn't remove it, better to write a script to add rDNS to your /64 subnet as i wrote in your other thread ;)

Best regards
 
Yes that was my first idea, but that does not solve the (potential) problem with other hosts without rDNS. But hey I'm not sure now much that would be.
 
I don't think there will be that many that use IPv6 so far, and for the ones that does with their ISP they would already have an rDNS because the ISP should already take care of it :)

I wouldn't bother that much honestly with that ;)

Best regards
 
Bind supports wildcards in PTR. That can I say for sure. I used to support a subnet with PTR in Directadmin.

Sure, I will check the thread.
 
You must log in or register to reply here.
Back
Top