When I try to access certain websites in middleman
1.8.2 (as well as 1.8) I get a "Connection to
<website>:80 failed" message after a pregnant pause.
The logs indicate...
Mon 11 13:35:34 [26633] header: error reading header
from www.ananova.com
Mon 11 13:35:34 [26633] template: sent noconnect
Two sites I have found that do this are
"www.ananova.com" and "www.hsamuel.co.uk", which are
reachable through squid.
The DNS cache indicates they have been resolved correctly.
Any ideas?
Logged In: YES
user_id=432781
Is there any delay before the error shows up? I can't think of any
reason for this to happen besides the website taking too long to
respond to the request, causing a timeout... I can't reproduce this on
my system. Also, did squid use a cached copy? If you're using
konqueror, you can figure this out by right clicking on the page and
going to "View document information" and checking the X-Cache
header.
Logged In: YES
user_id=835361
There is what is approx a 10 sec delay before the error
appears which corresponds quite nicely to my connect
timeout in middleman. The two sites under squid bring the
text in under a second and take about 5 seconds to fill in the
graphics. These are not cached (fully less than a second
when cached).
I also use a disk cache size of 0 (memory only) - just in case
that is important.
Logged In: NO
Tried both sites, using MiddleMan 1.8.2. Just standard
install using ./configure, make and make install.
Both sites worked fine... Although I must say there ARE some
sites which I cannot connect to. I thought this was due to a
header isue or something. Like when you click "compose" in
hotmail, the textbox is very small. This is because hotmail
has another page for IE then for mozilla. Due to the
header_replace, hotmail thinks I am using IE.
I'll keep my eyes open, if I find more sites to which I'm
unable to connect to, I'll post them.
Roel Brook
Logged In: YES
user_id=835361
I knocked up the connection timeout to 100 seconds, and it
does exactly the same, but 100 seconds later, rather than
10, so I'm starting to suspect it isn't even trying to connect,
rather that it is timing out.
I'll try it without any filtering functions later if I have time.
Logged In: YES
user_id=835361
I built a different machine with middleman 1.8.3. Some of the
sites formally inaccessible now work (and some are still
inaccessible). This would start to point the finger at the first
machine (assuming nothing has changed since 1.8.2 in that
department).
Strange.