wget hangs when trying to download

[JuhazOne]

My wget seems to hang every time I try to download anything from the net. For example:
> wget www.ubuntuforums.org
--23:55:30-- http://www.ubuntuforums.org/
=> `index.html'
Resolving www.ubuntuforums.org...

And that's all. It hangs and doesn't respond to ctrl+c.
(Actually, my computer seems to have some weird problem that occasionally some processes hang so that even root can't kill them with kill -9 <pid>. This happens with wget, too. As far as I know, this is unrelated, though.)

I tried to give wget an ip only. Here's what happens:
> wget http://64.21.33.9/
--00:15:16-- http://64.21.33.9/
=> `index.html'

And wget hangs.

If it weren't for the problem with the ip's, one might suppose that there's a problem with DNS. My Opera browser also seems to lag quite a bit (up to 15 seconds) occasionally when I try to browse a new server.

Note that wget doesn't hang when it doesn't try to access the Internet. Typing wget --help or wget --version work just fine.

I'm using kernel 2.6.12-9-386. Wget is version 1.10 and apt-get says it's the newest version. My wget is normally aliased to wget -c -nc but unaliasing it has no effect. The wget binary is located at /usr/bin/wget if that's useful info for someone. I'm behind a NATted network and I use DHCP.


Any ideas what could be wrong?

[oskude]

hmm,

i get the "index.html" with "wget www.ubuntuforums.org"...
can you open www.ubuntuforums.org with firefox ?
what does ping and traceroute say for ubuntuforums.org ?

[JuhazOne]

Firefox works just fine. So does ping.

> ping www.ubuntuforums.org
PING www.ubuntuforums.org (64.21.33.9) 56(84) bytes of data.
64 bytes from wailuku.xlogicgroup.com (64.21.33.9): icmp_seq=1 ttl=48 time=126 ms
64 bytes from wailuku.xlogicgroup.com (64.21.33.9): icmp_seq=2 ttl=48 time=125 ms
64 bytes from wailuku.xlogicgroup.com (64.21.33.9): icmp_seq=3 ttl=48 time=125 ms
64 bytes from wailuku.xlogicgroup.com (64.21.33.9): icmp_seq=4 ttl=48 time=126 ms

--- www.ubuntuforums.org ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3002ms
rtt min/avg/max/mdev = 125.916/126.049/126.293/0.383 ms

traceroute hangs and becomes unkillable. It seems this problem is more common than I thought.

OpenOffice Writer ( /usr/lib/openoffice/program/soffice.bin ) hangs like this, too. It shows the splash screen, then hangs and the process becomes immortal.

Is this a bug in some library or something...?

[JuhazOne]

Problem solved by updating the kernel from 2.6.12-9-386 to 2.6.12-10-686.

[JuhazOne]

Oh crap... the problem is back again. And my kernel is still the same 2.6.12-10-686 that previously fixed the problem.

[JuhazOne]

So far the affected programs seem to be:

wget
nmap
traceroute
sudo
OO writer


Many of them are network-related but some are not.

Anyone got any ideas...?

[dradul]

Anyone got any ideas...?


Yes, all those are symptoms of a dying network interface. If your NIC is an integrated ethernet, it could even be the tell-tale sign of a dying mainboard. OR, your CPU is overheating... Sorry to be the bearer of bad news...

[shin]

Imho could be a problem with disk. I don't really know much about linux scandisks etc but maybe give it a try. Or maybe some kind of filesystem damage. I don't think it is network or cpu related problem.

[JuhazOne]

Yes, all those are symptoms of a dying network interface. If your NIC is an integrated ethernet, it could even be the tell-tale sign of a dying mainboard. OR, your CPU is overheating... Sorry to be the bearer of bad news...

It's possible in the sense that the NIC is old. More than five years at least, possibly even more.

Is there a way I could be sure without having to try another NIC?

[dradul]

It's possible in the sense that the NIC is old. More than five years at least, possibly even more.

Is there a way I could be sure without having to try another NIC?
You would have to run a hardware test. There are some benchmarking software packages that could help (I haven't used any in a long time, I don't recall the name of any either :-P ). Do make sure that you don't try something like pinging a public web site for a long time or something of the sort, you'd only get yourself in trouble (as in being the source of a DoS attack).

[JuhazOne]

Ok. So I got it all working again now.

I asked the people on #ubuntu (@freenode) about this and they (too) suggested I'd try another NIC. I decided to try the one that's integrated into my motherboard (haven't been using it since Windows doesn't like it). Well, the network adapter got detected and configured automatically and wget & friends work now. :)

It is unsure whether the problem was with the NIC itself or its drivers.

The NIC was Intel Pro/100+. When I last looked (a long time ago), Windows detected it as "Intel Pro/100+ Management Adapter 82559 10/100MB". I believe it was bought in March 2000.

[JuhazOne]

Oh crap........

Guess what? The problem is back once more.

This time I'm ditching my current installation of Ubuntu and starting from scratch again. Will see if it helps.

[stream303]

... If it weren't for the problem with the ip's, one might suppose that there's a problem with DNS. My Opera browser also seems to lag quite a bit (up to 15 seconds) occasionally when I try to browse a new server.

I'd take a look at your /etc/resolv.conf file

If you are running DHCP, I'll bet that the nameservers are pointing to an internal DNS of your router, which causes the delay as it times-out. Does it look something like this?

domain yourisp.net (or search yourisp.net)
nameserver 172.16.0.254

The private ip address is a giveaway that the system is timing out while trying to use the internal dns of a router, etc, and not your actual isp's dns first. We can fix this issue by editing /etc/dhcp3/dhclient.conf

I wrote a quick page about this issue at:
http://www.greertech.com/nixnotes/dhcpfix.html

Let me know if this works for you...