Jeffrey Janner wrote:
-----Original Message-----
From: Jeffrey Janner [mailto:jeffrey.jan...@polydyne.com]
Sent: Friday, April 04, 2014 12:04 PM
To: 'Tomcat Users List'
Subject: RE: AW: AW: tomcat-connectors-1.2.39-windows-x86_64-iis does
not work
-----Original Message-----
From: Christopher Schultz [mailto:ch...@christopherschultz.net]
Sent: Friday, April 04, 2014 10:23 AM
To: Tomcat Users List
Subject: Re: AW: AW: tomcat-connectors-1.2.39-windows-x86_64-iis does
not work
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256
Jeffrey,
On 4/4/14, 10:50 AM, Jeffrey Janner wrote:
-----Original Message----- From: André Warnier [mailto:aw@ice-
sa.com]
Sent: Thursday, April 03, 2014 5:27 PM To:
Tomcat Users List Subject: Re: AW: AW:
tomcat-connectors-1.2.39-windows-x86_64-iis does not work
Christopher Schultz wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
André,
On 4/3/14, 3:34 PM, André Warnier wrote:
Alten, Jessica-Aileen wrote:
-----Ursprüngliche Nachricht----- Von: André Warnier
[mailto:a...@ice-sa.com] Gesendet: Donnerstag, 3. April
2014 15:36 An: Tomcat Users List Betreff: Re: AW:
tomcat-connectors-1.2.39-windows-x86_64-iis does not work
Alten, Jessica-Aileen wrote:
A bit guessing here :
You have :
worker.ajp13w.host=localhost
and
jk_open_socket::jk_connect.c (735): connect to
0.0.0.0:8009
failed
(errno=49)
is "localhost" == 0.0.0.0 ?
From the point of view of mod_jk/isapi, should it not be
"127.0.0.1" ?
Your answer points to the right direction. 0.0.0.0
means: any configured IPv4-Address on this computer, see
http://serverfault.com/questions/78048/whats-the-difference-
betwee
n-
ip
-addre ss-0-0-0-0-and-127-0-0-1
In principle this is ok at first. The Ajp13 Connector was
configured in server.xml to listen at any IPv4 address on
port
8009 - which is the default setting.
But the connector can't find any suitable
address.
The problem is: The new Tomcat-Connector can't parse
"worker.ajp13w.host=localhost", instead localhost must be
replaced
with "127.0.0.1", this works!
In my eyes this is a big fat bug, because most documentation
on workers use "localhost". localhost is actually the default
for
the
"host" connection directive.
The new worker directive "prefer_ipv6" doesn't change this
behavior.
Hi.
Can you please really check this ?
Open a command window on that server, and do "ping localhost".
It
should tell you what it understands by "localhost". Copy and
paste the result here :
ping localhost
Ping wird ausgeführt für xyz.uv.local [127.0.0.1] mit 32 Bytes
Daten: Antwort von 127.0.0.1: Bytes=32 Zeit<1ms
TTL=128 Antwort von 127.0.0.1: Bytes=32 Zeit<1ms TTL=128
Antwort
von 127.0.0.1: Bytes=32 Zeit<1ms TTL=128 Antwort von 127.0.0.1:
Bytes=32 Zeit<1ms TTL=128
Ping-Statistik für 127.0.0.1: Pakete: Gesendet = 4, Empfangen =
4,
Verloren = 0 (0% Verlust), Ca. Zeitangaben in Millisek.:
Minimum
=
0ms, Maximum = 0ms, Mittelwert = 0ms
That /is/ bizarre. As far as I know, to resolve hostnames in
its
configuration, mod_jk/isapi is using the OS's resolver library,
the
same as the one "ping" should be using. On the other hand, you
say that if you have
worker.ajp13w.host=localhost
it doesn't work (mod_jk cannot connect to tomcat), but when you
change this to
worker.ajp13w.host=127.0.0.1
then it works fine.
Ok, another check in a command window (and I assume that you
open
this command window *on the server itself* where mod_jk and
Tomcat are running, right ?)
test :
1) telnet localhost 8009
2) telnet 127.0.0.1 8009
Any difference between these 2 cases ?
If not, then indeed it looks like a mod_jk/isapi_redirect
1.2.39 problem.
In any case, you cannot "connect to" 0.0.0.0, as this log line
would suggest :
jk_open_socket::jk_connect.c (735): connect to
0.0.0.0:8009
failed
Could this be an interaction between IPv4 and IPv6? Try:
C:> nslookup localhost
You might get only 127.0.0.1 or you might also get :: (or
something equivalent). I'm not sure why it wasn't happening with
earlier versions of mod_jk (which?).
(versions : her first post mentioned the versions she was
comparing)
I previously asked Jessica-Aileen to do a "ping localhost" on the
machine, see results above. It definitiely pings 127.0.0.1 ..
(assuming it was done on the same machine)
And I don't think that nslookup uses the local resolver. When I'm
doing that on my Windows laptop, for "localhost" it responds "not
found" (in many more German words)
C:\Dokumente und Einstellungen\aw>nslookup localhost Server:
fire-see.localdomain Address: 192.168.245.253
*** localhost wurde von fire-see.localdomain nicht gefunden:
Non- existent domain
On the other hand, it does this (spot the difference..):
C:\Dokumente und Einstellungen\aw>nslookup localhost. Server:
fire-see.localdomain Address: 192.168.245.253
Name: localhost Address: 127.0.0.1
(But that of course could be the "localhost" of my DNS server,
since it happens to be a Linux box running dnsmasq, and it has
that
name
in
it's own hosts file.)
This result is understandable, as the nslookup tool is a DNS
resolver tool. It's designed to query the DNS system directly,
avoiding the systems resolver and any caching. Not exactly sure why
it resolves "localhost.", but adding the final period tells it not
to append the default domain. In other words, localhost. Is a top-
level domain.
Perhaps there is a special case built into the DNS system for that.
The difference here is that the resolver is going to search DNS and
the local hosts table, usually with the local hosts table
(/etc/hosts in *nix) taking precedence. I've not followed the
complete thread,
but
if the server is a *nix implementation, the better diag tool might
be dig. And yes, I would not expect the address 0.0.0.0 on a client
to connect to the localhost. That is a special case address
meaning
"local network".
If anything, it would be sending packets out the NIC card, not via
loopback.
0.0.0.0 means "all IPv4 interfaces available" and only applies for
binding a server socket. You can never connect to "0.0.0.0" as a
client.
Chris -
It actually has a different meaning based on use. For binding to a
socket in the local IP stack, it means what you say.
In the routing table, it means the default route. In
firewalls/routers, it probably means something completely different.
When used as a destination address, it means what I said. How the IP
stack/hardware deals with it is dependent on the implementation. The
RFCs specify that it should be treated the same as the broadcast
address, but local network only, and not routable. That may be for
received packets only, as I've seen other references that it should
never be used on-the-wire, unless as the source address in protocols
like DHCP.
In any event, definitely not expect the 0.0.0.0. address to get any
response, either local host or otherwise.
For the OP's specific problem, s/he need to see how "localhost" is
resolving. Most systems define it in the local "hosts" file, either
/etc/hosts (*nix) or c:\Windows\system32\etc\hosts. Not sure for other
systems.
Jeff
Make that C:\Windows\system32\drivers\etc\hosts.
I did a test and it appeared that ping didn't rely on the entry being there,
but it could have been a cached result.
Jeff
Summary of what went on previously :
I did ask before, in the thread, to Jessica to do a "ping localhost" on the Apache/Tomcat
server where the issue arises. It does ping "127.0.0.1".
I presume that "ping" does resolve the name "localhost" by using the local
resolver.
Yet mod_jk's log seems to show that it is trying, as a client, to connect to "0.0.0.0",
although it's worker's host address is specified as "localhost".
That is what's puzzling us all so far.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
