Intermittent "no route to host" error on Linux

Question

I have 2 computers on the same network:
10.0.0.4
10.0.0.5

If I leave things idle for about 10 minutes or so, when I attempt to connect between them, I get:

$ ssh user@10.0.0.4
$ ssh: connect to host 10.0.0.4 port 22: No route to host

But if I retry immediately, it works. Leave it for 10 mins again, and I get the error. This also happens with ping, so I don't think it's related to ssh.

Any ideas?

Additional info: Both machines are Ubuntu

**10.0.0.4:**
$ ifconfig
ath0      Link encap:Ethernet  HWaddr 00:1f:e2:86:65:60  
          inet addr:10.0.0.4  Bcast:10.0.0.255  Mask:255.255.255.0
          inet6 addr: fe80::21f:e2ff:fe86:6560/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:3778 errors:0 dropped:0 overruns:0 frame:0
          TX packets:1531 errors:1 dropped:1 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:516476 (516.4 KB)  TX bytes:209053 (209.0 KB)

$ netstat -rn
Kernel IP routing table
Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
10.0.0.0        0.0.0.0         255.255.255.0   U         0 0          0 ath0
169.254.0.0     0.0.0.0         255.255.0.0     U         0 0          0 ath0
0.0.0.0         10.0.0.1        0.0.0.0         UG        0 0          0 ath0

$  arp -n
Address                  HWtype  HWaddress           Flags Mask            Iface
10.0.0.1                 ether   00:11:50:1e:61:02   C                     ath0
10.0.0.5                 ether   00:26:5e:11:97:13   C                     ath0

**10.0.0.5:**
$ ifconfig
eth1      Link encap:Ethernet  HWaddr 00:26:5e:11:97:13  
          inet addr:10.0.0.5  Bcast:10.0.0.255  Mask:255.255.255.0
          inet6 addr: fe80::226:5eff:fe11:9713/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:9388 errors:0 dropped:0 overruns:0 frame:24978
          TX packets:10300 errors:6 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:4785464 (4.7 MB)  TX bytes:1978774 (1.9 MB)
          Interrupt:17 

$ netstat -rn
Kernel IP routing table
Destination     Gateway         Genmask         Flags   MSS Window  irtt Iface
10.0.0.0        0.0.0.0         255.255.255.0   U         0 0          0 eth1
169.254.0.0     0.0.0.0         255.255.0.0     U         0 0          0 eth1
0.0.0.0         10.0.0.1        0.0.0.0         UG        0 0          0 eth1

$  arp -n
Address                  HWtype  HWaddress           Flags Mask            Iface
10.0.0.1                 ether   00:11:50:1e:61:02   C                     eth1

Please provide the output of `ifconfig`, `netstat -rn`, and `arp -n` from both computers. Thanks — Charles Hooper, Nov 09 '10 at 16:46
Since it's a desktop distro install, my first guess: are your NICs set to power down when not in use? If so, you may want to look at http://en.wikipedia.org/wiki/Wake-on-LAN, or alternatively set them to be on all the time. — Zayne S Halsall, Nov 09 '10 at 17:15
@Zayne: Where would I go to check that? I don't see anything under power management. — Hollister, Nov 09 '10 at 17:30
The massive amount of frame errors on 10.0.0.5 indicates duplex mismatch. What does 'ethtool eth1' say for this host? — flashnode, Nov 09 '10 at 18:06
Well, there goes my expected "replace the ethernet cable" response :P — Charles Hooper, Nov 09 '10 at 18:21
@Hollister: I'd start looking in the BIOS of the machine (assuming the NIC is on the main board). If not, any other info would be valuable - such as the relevant output from lspci -vvv. Another question - when did this begin happening? Have these laptops always been on wireless? Has anything changed recently? — Zayne S Halsall, Nov 10 '10 at 15:35

score 1 · Answer 1 · answered Nov 09 '10 at 18:27

1

At this point, knowing that both computers have wireless NICs then I wouldn't rule out interference. However, given that only one of the hosts is experiencing frame errors, I want to point my finger at that particular host.

Recommendations

Ensure that there isn't anything near the hosts (particularly 10.0.0.5) that may cause interference. Possibilities: Power cords, microwaves, other broadcasting sources (cordless phones, etc)
Ensure 10.0.0.5 is using the correct wireless driver.
Borrow a wireless card for 10.0.0.5 to determine if your current one is fubar

answered Nov 09 '10 at 18:27

Charles Hooper

1,520
9
8

@Charles: Hm. Well, they're both laptops on my desk about a foot from each other, with as few wires nearby as possible. I use 10.0.0.5 all day everyday and see no other issues than this one. Swapping the NIC would not be easy at all. – Hollister Nov 09 '10 at 18:32
Well, if you're confident that 10.0.0.5 is OK, then it's possibly that it's 10.0.0.4 that is spewing garbage out of its radios. Turn .4 off and use .5 for awhile. Do the number of frame errors listed in `ifconfig` continue to increase? – Charles Hooper Nov 09 '10 at 18:44
5 is the one I use all the time, and the frame errors have doubled since I posted the number here. This is without any communication to 4. – Hollister Nov 09 '10 at 19:42
Then I can only re-iterate my current recommendations. Good luck! – Charles Hooper Nov 10 '10 at 03:55
@Charles: To try to isolate the frame error problem, I took the 5 laptop to the room where the wifi AP is, well away from any interference sources. I still get hundreds of errors a minute. What else would cause those errors? Thanks for your help, I'm a software guy, not a hardware guy. – Hollister Nov 10 '10 at 12:56
Then I'm going to bet on a bad driver or bad wireless card (which is why those are recommendations #2 and #3.) Borrow a PCMCIA or USB wireless card and see if those get frame errors. – Charles Hooper Nov 10 '10 at 15:52

score 0 · Answer 2 · answered Nov 09 '10 at 16:51

0

Try to trace the problem using network sniffer like tcpdump or wireshark.

You can also try to ssh from the 10.0.0.4 to 10.0.0.5.

answered Nov 09 '10 at 16:51

Khaled

36,533
8
72
99

Intermittent "no route to host" error on Linux

2 Answers2