Bridged Linux on Linux drops network

[axelmasok]

Hi everyone,

Anyone know why a certain vbox cguest I have keeps dropping off the network when using bridged mode networking?

Host: Centos 5.2
Guest: vTiger Jumpbox (I think it's Debian Linux)

Vbox v2.1.4

The host has dual gigabit which we only use eth0 or one interface. I have other guests on the same machine but don't leave them running like vTiger.

In a nutshell, the guest OS picks up an IP from our LAN dhcp server, can ping to and from OK for hours and sometimes days and then all of a sudden it's not contactable via it's IP in any direction. To get it back online I either :
1) ifconfig eth0 down and back up (on guest)
2) save the session and start it again
3) reboot the guest

Getting a bit tired of doing this, there must be something to try and stop this. I am currently trying to leave another guest OS on the same host running to see if it happens to another guest.

[vbox4me2]

Check out the DHCP lease time, the Guest must aquire a new one once that period is passed, the Guest may have a problem doing that.

[axelmasok]

I'll force the IP instead of DHCP. I'm pretty sure I have tried this already. Will report result.

[axelmasok]

OK, since last post setting the IP to be a fixed IP, I managed to have increased the reliability.

Something is still not right though...

Just upgraded to 2.2.2 and found the virtual machine was unreachable. I VNC'd to it and run (Debian) /etc/init.d/networking restart - to bring it back up.

Another infuriating problem is whilst the Centos host can ping the Debian guest (and vice versa) they can't ssh to each other either way.
All I get is:

Code: Select all

[root@ci-nas ~]# ssh admin@10.10.10.6
Connection closed by 10.10.10.6

(the guest)
If I netcat to port 22 I get:

Code: Select all

[root@ci-nas ~]# nc 10.10.10.6 22
SSH-2.0-OpenSSH_4.7p1 Debian-8ubuntu1

What the?

[axelmasok]

Forgot to mention: ping and other services between the guest and "other" machines on the network work perfectly. It's only host to guest.

There seems to a few other people on the forum with similar problems: bridged network and bad or impossible connections between guest and host (and only ping works)...

[baf]

Another infuriating problem is whilst the Centos host can ping the Debian guest (and vice versa) they can't ssh to each other either way.
All I get is:

Code: Select all

[root@ci-nas ~]# ssh admin@10.10.10.6
Connection closed by 10.10.10.6

(the guest)
If I netcat to port 22 I get:

Code: Select all

[root@ci-nas ~]# nc 10.10.10.6 22
SSH-2.0-OpenSSH_4.7p1 Debian-8ubuntu1

What the?

As nc gives a reasonable value I'd say that this is a configuration problem with your sshd. Check /etc/hosts.{allow,deny} and see that they are ok.
If that doesnt help try
ssh -vvv ....
and see what that gives.

Also look at the UseDNS setting in /etc/sshd_config (approximate path, Differs between distros).

[axelmasok]

Hi Baf,
Appreciate the help.
Sshd is configured fine because I can ssh to the guest VM from ANY machine on the network except the host.
hosts.allow/deny are normal on both machines

I am 100% positive its a networking issue between the host and guest interfaces.

I have also booted up a WindowsXP VM on the same host. Same problem. I can't ssh to the host and I can't use smb/cifs between the two either.
Have just tried using different net card drivers for the guest and also tried interface 2. No joy.
Will have to wait for this to be fixed I guess....

[vbox4me2]

You sure the Host is not firewalled?

[axelmasok]

You sure the Host is not firewalled?

Very:

Sshd is configured fine because I can ssh to the guest VM from ANY machine on the network except the host.

Code: Select all

[root@ci-nas scripts]# iptables -L -n
Chain INPUT (policy ACCEPT)
target     prot opt source               destination

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination
[root@ci-nas scripts]#

The guest doesn't even have iptables userland tools installed.
There is a bug somewhere and it's either vbox or the Centos kernel or something in this combination.

[axelmasok]

Done some more testing. Problem seems isolated to this particular server/hardware.

I just tried systemrescuecd on my laptop here (host kubuntu 9.04, guest systemrecuecd) and bound the guest's lan interface to my host wlan0 (Intel PRO/Wireless 3945ABG): works fine ssh both ways.

Back to the problem server, it has x2 1000T interfaces on a supermicro motherboard. Eth0 in use, Eth1 not used.
The LAN interfaces are: 06:00.0 Ethernet controller: Intel Corporation 80003ES2LAN Gigabit Ethernet Controller (Copper) (rev 01)

Code: Select all

[root@ci-nas ~]# uname -a
Linux ci-nas.ci-domain 2.6.18-128.1.10.el5.centos.plusPAE #1 SMP Mon May 11 07:51:33 EDT 2009 i686 i686i386 GNU/Linux

[root@ci-nas ~]# ifconfig
eth0      Link encap:Ethernet  HWaddr 00:30:48:65:FE:4E
          inet addr:10.10.10.15  Bcast:255.255.255.255  Mask:255.255.255.0
          inet6 addr: fe80::230:48ff:fe65:fe4e/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:2528311 errors:0 dropped:0 overruns:0 frame:0
          TX packets:693668 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:3220550616 (2.9 GiB)  TX bytes:101280784 (96.5 MiB)
          Memory:da320000-da340000

[root@ci-nas ~]# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
10.10.10.0      0.0.0.0         255.255.255.0   U     0      0        0 eth0
169.254.0.0     0.0.0.0         255.255.0.0     U     0      0        0 eth0
0.0.0.0         10.10.10.2      0.0.0.0         UG    0      0        0 eth0

I just created a new VM guest with Damn Small Linux iso. I see the same problem: I can't make a "complete" ssh between the guest and host (or visaversa). Ping works however.

Here is the output from the DSL guest ssh to the host and the host's netstat output:

Code: Select all

..snipped....
debug1: Connecting to 10.10.10.15 [10.10.10.15] port 22
debug1: Connection established.
...snipped....

Code: Select all

tcp        0     20 ::ffff:10.10.10.15:22     ::ffff:10.10.10.163:1031    ESTABLISHED

So the connection is established but won't complete. This is why I can netcat the port too as per previous post. BTW, I have tried smb (smbclient) which also fails to progress.
The problem I believe is Centos 5.2 and/or the LAN driver or it's configuration.

[Sasquatch]

What if you use ssh -4, instead of normal ssh. With the -4 parameter, you force SSH to use the IPv4 protocol. Now, using the IP itself as IPv4 should have been enough, but with your netcat output, the ::ffff:.10.10.10.15:22 seems like a conversion from IPv6 to IPv4.

[axelmasok]

Thanks for the tip. That didn't make much difference however.
Regards

[axelmasok]

OK, testing suspended as Vmware server in bridged mode (unfortunately) solves all the above problems.
Same hardware/host.
Same virtual machines.
Same LAN settings.
Same SSH settings.
Again, it's something to do with Vbox and Centos 5.2 on Supermicro hardware.
If anyone wants to test further or has a solution to this problem with Virtualbox then send me an email/PM.
Regards