When using WebClient, "WebClient httpGet: 'http://google.com'" succeeded, but "WebClient httpGet: '<a href="http://localhost:8090'">http://localhost:8090'" reported "ConnectionTimedOut: Cannot connect to 127.0.0.1:8090"

The address looked fine from Safari.  What gives?

Thanks.
Sean

Log:
ConnectionTimedOut: Cannot connect to 127.0.0.1:8090
28 October 2010 6:23:58.673 pm

VM: Mac OS - Smalltalk
Image: Squeak4.1 [latest update: #9957]

SecurityManager state:
Restricted: false
FileAccess: true
SocketAccess: true
Working Dir /Users/sean/Squeak/Main Working Images/Squeak4.1
Trusted Dir /foobar/tooBar/forSqueak/bogus
Untrusted Dir /Users/sean/Library/Preferences/Squeak/Internet/My Squeak

[] in Socket>>connectTo:port:waitForConnectionFor:
        Receiver: a Socket[unconnected]
        Arguments and temporary variables:
<<error during printing>
        Receiver's instance variables:
                semaphore: a Semaphore()
                socketHandle: #[52 74 202 76 0 0 0 0 64 114 35 0]
                readSemaphore: a Semaphore()
                writeSemaphore: a Semaphore()
                primitiveOnlySupportsOneSemaphore: false

Socket>>waitForConnectionFor:ifTimedOut:
        Receiver: a Socket[unconnected]
        Arguments and temporary variables:
                timeout: 45
                timeoutBlock: [closure] in Socket>>connectTo:port:waitForConnectionFor:
                status: 0
                deadline: 672266
        Receiver's instance variables:
                semaphore: a Semaphore()
                socketHandle: #[52 74 202 76 0 0 0 0 64 114 35 0]
                readSemaphore: a Semaphore()
                writeSemaphore: a Semaphore()
                primitiveOnlySupportsOneSemaphore: false

Socket>>connectTo:port:waitForConnectionFor:
        Receiver: a Socket[unconnected]
        Arguments and temporary variables:
                hostAddress: #[127 0 0 1]
                port: 8090
                timeout: 45
        Receiver's instance variables:
                semaphore: a Semaphore()
                socketHandle: #[52 74 202 76 0 0 0 0 64 114 35 0]
                readSemaphore: a Semaphore()
                writeSemaphore: a Semaphore()
                primitiveOnlySupportsOneSemaphore: false

Socket>>connectTo:port:
        Receiver: a Socket[unconnected]
        Arguments and temporary variables:
                hostAddress: #[127 0 0 1]
                port: 8090
        Receiver's instance variables:
                semaphore: a Semaphore()
                socketHandle: #[52 74 202 76 0 0 0 0 64 114 35 0]
                readSemaphore: a Semaphore()
                writeSemaphore: a Semaphore()
                primitiveOnlySupportsOneSemaphore: false


--- The full stack ---
[] in Socket>>connectTo:port:waitForConnectionFor:
Socket>>waitForConnectionFor:ifTimedOut:
Socket>>connectTo:port:waitForConnectionFor:
Socket>>connectTo:port:
 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
SocketStream class>>openConnectionToHost:port:
SocketStream class>>openConnectionToHostNamed:port:
WebClient>>connect
WebClient>>sendRequest:contentBlock:
WebClient>>sendRequest:content:size:
WebClient>>sendRequest:
WebClient>>httpGet:do:
WebClient>>httpGet:
WebClient class>>httpGet:
UndefinedObject>>DoIt
Compiler>>evaluate:in:to:notifying:ifFail:logged:
[] in SmalltalkEditor(TextEditor)>>evaluateSelection
BlockClosure>>on:do:
SmalltalkEditor(TextEditor)>>evaluateSelection
SmalltalkEditor(TextEditor)>>inspectIt
SmalltalkEditor(TextEditor)>>inspectIt:
SmalltalkEditor(TextEditor)>>dispatchOnCharacter:with:
SmalltalkEditor(TextEditor)>>readKeyboard
[] in TextMorphForEditView(TextMorph)>>keyStroke:
TextMorphForEditView(TextMorph)>>handleInteraction:fromEvent:
TextMorphForEditView>>handleInteraction:fromEvent:
TextMorphForEditView(TextMorph)>>keyStroke:
TextMorphForEditView>>keyStroke:
TextMorphForEditView(TextMorph)>>handleKeystroke:
KeyboardEvent>>sentTo:
TextMorphForEditView(Morph)>>handleEvent:
TextMorphForEditView(Morph)>>handleFocusEvent:
[] in HandMorph>>sendFocusEvent:to:clear:
[] in PasteUpMorph>>becomeActiveDuring:
BlockClosure>>on:do:
PasteUpMorph>>becomeActiveDuring:
HandMorph>>sendFocusEvent:to:clear:
HandMorph>>sendEvent:focus:clear:
HandMorph>>sendKeyboardEvent:
HandMorph>>handleEvent:
HandMorph>>processEvents
[] in WorldState>>doOneCycleNowFor:
Array(SequenceableCollection)>>do:
WorldState>>handsDo:
WorldState>>doOneCycleNowFor:
WorldState>>doOneCycleFor:
PasteUpMorph>>doOneCycle
[] in Project class>>spawnNewProcess
[] in BlockClosure>>newProcess

On 29/10/10 12:34 AM, Sean P. DeNigris wrote:
>
> When using WebClient, "WebClient httpGet: 'http://google.com'" succeeded, but
> "WebClient httpGet: '<a href="http://localhost:8090'">http://localhost:8090'" reported "ConnectionTimedOut:
> Cannot connect to 127.0.0.1:8090"
>
> The address looked fine from Safari.  What gives?

Some ideas:

- It's really 8090 (that's a nine), not 8080?
- Are there any proxy configurations involved
- Could it be an IPv6 issue

Yanni Chiu wrote

- It's really 8090 (that's a nine), not 8080?

Yes, I kept changing it to make sure it wasn't a problem with that particular port

Yanni Chiu wrote

- Are there any proxy configurations involved

I don't know what that means, but for more context, I started a Ruby TCPServer on that port, and then tried to "get" from Squeak, then from Safari

Yanni Chiu wrote

- Could it be an IPv6 issue

I don't know what that means, either, lol.

Sean

On 10/28/10 11:01 PM, Sean P. DeNigris wrote: Can you "get" using Squeak on any port from the Ruby TCPServer?  
Different scripting libraries can have oddities that the more mainstream
c-based ones don't have. I once found an error in the Mac's Python TCP
client code that randomly prepended header data into  xmlrpc content.
Its possible that Ruby is doing something that Safari libs are robust
enough to compensate for, or its possible that Ruby is just fine and
there's an obscure bug in the Squeak lib that you happen to be the first
to find.

Lawson

Lawson English-2 wrote

Can you "get" using Squeak on any port from the Ruby TCPServer?  

I think not.  I tried several different ports in the 80xx range with the same result.

Sean

On 2010/10/29 08:01, Sean P. DeNigris wrote:
It means that you could try access http://127.0.0.1:8090/. If that
works, then it's likely that Safari's actually trying to access
http://[::1]:80/ - and you might not be running anything there.

You can check by running netstat -anp tcp and looking for the relevant
line. Something like

tcp4       0      0  *.8090   *.*       LISTEN

without a line like
tcp6       0      0  *.8090   *.*       LISTEN

for instance.

frank

I tested in a mac (10.6.4) using pydoc -p 8090 as the server and works ok:

> WebClient httpGet: '<a href="http://localhost:8090'">http://localhost:8090'

> WebResponse(HTTP/1.0 200 OK
> server: BaseHTTP/0.3 Python/2.6.4
> date: Fri, 29 Oct 2010 16:02:12 GMT
> content-type: text/html
>
> )
On 29/10/2010, at 00:36, Yanni Chiu wrote:

On 29/10/10 2:01 AM, Sean P. DeNigris wrote:
>
>> - Are there any proxy configurations involved
>>
> I don't know what that means, but for more context, I started a Ruby
> TCPServer on that port, and then tried to "get" from Squeak, then from
> Safari

I don't use WebClient, but I recall some posts about configuring it for
proxies. If you're not on a corporate network, you probably don't have a
proxy.

Previously, a newly imaged corporate machine would land on my desk. The
IE browser configuration already had various domains configured to
bypass the proxy. We developers had to test with Firefox, so we had to
configure the proxy in Firefox. If "localhost" were not added to the
bypass the proxy list, then we would never reach the app running on our
local machine, because it would try to go through the corporate proxy to
filter the "localhost" access.

>> - Could it be an IPv6 issue
>>
> I don't know what that means, either, lol.

In the /etc/hosts on my Mac, I have:

127.0.0.1       localhost
::1             localhost

I can't remember whether I added the 127.0.0.1 entry, or it was already
there. I do recall having to add it on some machine though. The
127.0.0.1 is and IPv4 address, and (I'm no IPv6 expert) the ::1 is an
IPv6 address.

Frank Shearar wrote

>> - Could it be an IPv6 issue
It means that you could try access http://127.0.0.1:8090/. If that
works, then it's likely that Safari's actually trying to access
http://[::1]:80/ - and you might not be running anything there.

Thanks.  I got:
~$ netstat -anp tcp
Active Internet connections (including servers)
Proto Recv-Q Send-Q  Local Address          Foreign Address        (state)
tcp6       0      0  ::1.8090               *.*                    LISTEN
...

That was the only entry for that port.  Is that what I should be seeing?

Sean

Yanni Chiu wrote

If you're not on a corporate network, you probably don't have a
proxy.

I'm not

Yanni Chiu wrote

In the /etc/hosts on my Mac, I have:

127.0.0.1       localhost
::1             localhost

Me too.

Thanks.
Sean

How do I debug this to get a clearer idea of what is going on?  For the moment I can use files instead of sockets, but I'll need to use this server approach quickly.

Thanks.
Sean

On 2010/10/29 19:23, Sean P. DeNigris wrote:
It means that Safari might be resolving "localhost" as the IPv4 address
127.0.0.1 and trying to contact that IP/port, and of course failing.
Your next step, I suspect, will be to fire up tcpdump and confirm that
that's the case.

Start tcpdump with something like: sudo tcpdump -i fw0 -v port 8090 and
then trying to access the site through Safari.

frank

If you send the ruby script you're using for running the TCPServer?

Cheers,
   - Andreas

On 10/28/2010 11:01 PM, Sean P. DeNigris wrote:

Frank Shearar wrote

Start tcpdump with something like: sudo tcpdump -i fw0 -v port 8090 and
then trying to access the site through Safari.

Exciting!  WebClient is communicating, just not successfully.

After a little playing around, running "sudo tcpdump -i lo0 -vnS port 2002" (I keep changing the port) produced the following packet info:
  Safari's first packet:
    08:44:10.841003 IP6 (hlim 64, next-header TCP (6) payload length: 44) ::1.54889 > ::1.2002: Flags [S], cksum 0x9345 (correct), seq 2994624526, win 65535, options [mss 16324,nop,wscale 2,nop,nop,TS val 1034896177 ecr 0,sackOK,eol], length 0
  WebClient's first packet:
    08:39:48.796371 IP (tos 0x0, ttl 64, id 52777, offset 0, flags [DF], proto TCP (6), length 52, bad cksum 0 (->6e98)!)
    127.0.0.1.54859 > 127.0.0.1.2002: Flags [F.], cksum 0xfe28 (incorrect -> 0xa0fd), seq 964011037, ack 1862845032, win 65535, options [nop,nop,TS val 1034893560 ecr 1034892868], length 0

Any packet guru know what this means?

Thanks for the support so far - I'm really wanting to get this project up and this seems like progress!!!
Sean

Duh, I kept thinking that and forgot...

I tried a few different examples:
  1. Starting with the one from the pickaxe book:
        server = TCPServer.new('localhost', 8091)
        while (session = server.accept)
                puts "Request: #{session.gets}"
                session.print "HTTP/1.1 200/OK\r\nContent-type: text/html\r\n\r\n"
                session.print "<html><body>

#{Time.now}

</body></html>\r\n"
                session.close
        end

  2. server = TCPServer.open(2002)  # Socket to listen on port 2000
    loop {                         # Servers run forever
        client = server.accept       # Wait for a client to connect
        client.puts(Time.now.ctime)  # Send the time to the client
        client.puts "Closing the connection. Bye!"
    client.close                 # Disconnect from the client
    }

  3. server = TCPServer.new('localhost', 8091)
        while (true)
          begin
              sock = server.accept_nonblock
          rescue Errno::EAGAIN, Errno::ECONNABORTED, Errno::EPROTO, Errno::EINTR
              IO.select([server])
              retry
          end
          sock.print "sdflkdsjfslkdj"
          sock.close
        end

All three work in Safari and fail in WebClient.

Thanks.
Sean

On 2010/10/29 20:48, Sean P. DeNigris wrote:
It looks like Safari's resolving "localhost" to ::1, IPv6's loopback,
and the address on which your server's running, while WebClient isn't.

WebClient, instead, is resolving "localhost" to 127.0.0.1, IPv4's
loopback. There's nothing running there, so the connection fails.

Both behaviours are correct, or at least partially so. It's a policy
decision whether to try IPv4 addresses or IPv6 addresses first. After
that, it makes sense to try all the addresses (switching from 4 to 6 or
vice versa) before failing. I don't know how WebClient's name resolution
works, so we'll have to wait for Andreas, I think.

frank

On 2010/10/29 21:04, Sean P. DeNigris wrote:
I'll put my neck on the block and boldly claim that if you change
'localhost' to '127.0.0.1' WebClient will work.

frank

Frank Shearar wrote

I'll put my neck on the block and boldly claim that if you change
'localhost' to '127.0.0.1' WebClient will work.

#$%!@$ *@#$ &!*!# $*!%! &$!&$!&%$ *!#$ *!#$!#!!!!!!!!

It was that freaking easy?!?!?!

Thanks.
Sean

On 2010/10/29 21:39, Sean P. DeNigris wrote:
Haha! I'm glad we sorted out your problem.

I think NetNameResolver might have to be updated to support IPv6:

NetNameResolver addressForName: '::1' timeout: 5 --> timeout
NetNameResolver addressForName: '127.0.0.1' timeout: 5 --> #[127 0 0 1]

(NetNameResolver calls down to the SocketPlugin for the heavy lifting.)

frank

On 29.10.2010, at 22:09, Frank Shearar wrote:

> I think NetNameResolver might have to be updated to support IPv6:
>
> NetNameResolver addressForName: '::1' timeout: 5 --> timeout
> NetNameResolver addressForName: '127.0.0.1' timeout: 5 --> #[127 0 0 1]
>
> (NetNameResolver calls down to the SocketPlugin for the heavy lifting.)
>
> frank

In current Etoys on a Mac:

NetNameResolver addressForName: '::1' timeout: 5
==> ::1(localhost),0(0)

NetNameResolver addressesForName: 'localhost'
==> an OrderedCollection(
  ::1(localhost),0(0)-inet6-stream-tcp
  fe80::1%lo0(localhost),0(0)-inet6-stream-tcp
  127.0.0.1(localhost),0(0)-inet4-stream-tcp)

- Bert -