Tried to upgrade to just-released Nginx1.4. TCP 3-way hand-shake
aborted by server’s ACK+RST packet, but netstat shows server
is listening on that port. Any config has been changed since Nginx 1.2
to 1.4 in this regard?
Thanks,
- Alder
Hello!
On Wed, May 01, 2013 at 10:13:34AM -0700, Alder N. wrote:
Tried to upgrade to just-released Nginx1.4. TCP 3-way hand-shake
aborted by server’s ACK+RST packet, but netstat shows server
is listening on that port. Any config has been changed since Nginx 1.2
to 1.4 in this regard?
There are lots of changes in 1.4.0 compared to 1.2.x, see
http://nginx.org/en/CHANGES-1.4.
In this particular case I would recommend checking if nginx is
listening on the port, the address, and the protocol in question.
Note that since 1.3.4 ipv6only listen option is on by default, and
if you have
listen [::]:80;
in your config, it no longer implies IPv4 addresses regardless of
your OS settings.
–
Maxim D.
http://nginx.org/en/donation.html
netstat -pln shows the server is waiting on that port.
Yes, I have been using in server section
listen [::]:80;
What is supposed to be for IPV4 now?
I’ll go over the changelist later, Thanks,
- Alder
Just for clarity, I want to be listening on both IPv4 and IPv6 on the
same
port.
Hello!
On Wed, May 01, 2013 at 11:17:10AM -0700, Alder N. wrote:
Just for clarity, I want to be listening on both IPv4 and IPv6 on the same
port.
You have to write
listen 80;
listen [::]:80;
to listen on both IPv4 and IPv6.
I’ll go over the changelist later, Thanks,
Tried to upgrade to just-released Nginx1.4. TCP 3-way hand-shake
if you have
nginx mailing list
[email protected]
nginx Info Page
nginx mailing list
[email protected]
nginx Info Page
–
Maxim D.
http://nginx.org/en/donation.html
I tried
listen [::]:80 ipv6only=off;
and the TCP connection went through, but got
an http 400 bad response.
It would be nice to have a corresponding config
upgrade manual as well. Thanks,
On Wed, May 1, 2013 at 2:07 PM, Maxim D. [email protected] wrote:
listen [::]:80;to listen on both IPv4 and IPv6.
Doesn’t that require ipv6only=on?
listen 80;
listen [::]:80 ipv6only=on;
Hello!
On Wed, May 01, 2013 at 02:18:44PM -0700, Paul N. Pace wrote:
listen 80; listen [::]:80;to listen on both IPv4 and IPv6.
Doesn’t that require ipv6only=on?
listen 80; listen [::]:80 ipv6only=on;
As of nginx 1.3.4+, it’s on by default:
[…]
Note that since 1.3.4 ipv6only listen option is on by default, and
if you have
[…]
–
Maxim D.
http://nginx.org/en/donation.html
Although [::]:80 ipv6only=off; does work as advertized (including for
localhost sockets), [::1]:80 ipv6only=off; fails to respond to v4
connections.
Which is expected, since ::1 is an ipv6 address.
Lukas
“LT” == Lukas T. [email protected] writes:
Although [::]:80 ipv6only=off; does work as advertized (including for
localhost sockets), [::1]:80 ipv6only=off; fails to respond to v4
connections.
LT> Which is expected, since ::1 is an ipv6 address.
No it is not expected.
Everything else sees conenctions to anything in 127.0.0.0/8 when
listen(2)ing to ::1 with bindv6only off.
-JimC
James C. [email protected] OpenPGP: 1024D/ED7DAEA6
If this is a linux box you could simple use [::]:80 and it should, by
default, responde to both v4 and v6…
"In Linux by default any IPv6 TCP socket also accepts IPv4 traffic using
the IPv4 to IPv6 mapped address format, i.e., ::ffff:. E.g., ::ffff:192.168.0.27 maps the IPv4
address
192.168.0.27 to an IPv6 address.
When you enable the address [::]:80, binding port 80 using IPv6, in the
listen directive, in Linux, by default, the IPv4 port 80 is also
enabled.
Meaning that nginx listens for both IPv4 and IPv6 incoming traffic.
Therefore if you erroneously specify also a IPv4 address you’ll get an
already bind address error when reloading nginx configuration." (
http://wiki.nginx.org/HttpCoreModule) "
So I guess ipv6only=off should do the same… and it should work… I
can
be 100% sure since I don’t have any box with nginx 1.4 + ipv6 yet!
“AN” == Alder N. [email protected] writes:
AN> I tried
AN> listen [::]:80 ipv6only=off;
Although [::]:80 ipv6only=off; does work as advertized (including for
localhost sockets), [::1]:80 ipv6only=off; fails to respond to v4
connections.
-JimC
James C. [email protected] OpenPGP: 1024D/ED7DAEA6
Hi Jim,
Everything else sees conenctions to anything in 127.0.0.0/8
Not sure what you mean by everything else, but I don’t think
thats the case.
See this example:
“LT” == Lukas T. [email protected] writes:
LT> Hi Jim,
Everything else sees conenctions to anything in 127.0.0.0/8
LT> Not sure what you mean by everything else, but I don’t think
LT> thats the case.
Some time ago a message on debian-devel noted that deb was going to
start
defaulting to bindv6only=1. What is your /proc/sys/net/ipv6/bindv6only?
While looking into this, I found that, when given ::1, nc(1) explicitly
listens to both ::1 and ::ffff:127.0.0.1. (Although it has a bug which
causes it to close all v4 connections right after accept(2)ing them. 
My expectation was that it would work, but it seems some applications
force it to work by Doing What I Mean behind my back.
Evidently I’m going to have to write a test app to confirm things.
Probably later today.
-JimC
James C. [email protected] OpenPGP: 1024D/ED7DAEA6
Even when explicitly setting the socket option IPV6_V6ONLY to 0
(man 7 ipv6) - and thus ignoring cat /proc/sys/net/ipv6/bindv6only
this doesn’t work.
While looking into this, I found that, when given ::1, nc(1)
explicitly listens to both ::1 and ::ffff:127.0.0.1.
The behavior here is exactly the opposite. Perhaps you saw this
with an older/buggy kernel or netcat release?
terminal 1
lukas@ubuntuvm:~$ nc -vv -l ::1 8080
Connection from 0.0.0.0 port 8080 [tcp/http-alt] acceptedasd
lukas@ubuntuvm:~$
terminal 2
asd
^]
telnet> quit
Connection closed.
lukas@ubuntuvm:~$
Regards,
Lukas