Topic: Internal network stops working

[holck]

My SME server is setup in "server and gateway" mode, with two network interfaces: one connected to the outside world, and another to my internal network. Usually everything is fine, but recently I've two times experienced that the internal network just stops working.

I can connect an Ethernet cable physically to my server, and the LED lights up. The interface (eth0) is up, dhcpd is running. But when I try to connect a workstation to the network, nothing is registered in /var/log/messages or /var/log/dhcpd/current, and the workstation doesn't receive an IP address. Restarting dhcpd doesn't help.

The external network doesn't seem to be affected.

If I restart the server, everything seems to be working fine again, but of course this isn't very convenient doing every other day...

Any suggestions or pointers? I will happily contribute with more information from logs etc. I run SME 8.0.

Jesper, Denmark

[antystein]

I had the same problem last year. I am not using DHCP but similarly the server was not visible on the network at times. It turned out to be one of the network switches hanging under heavy traffic. I replaced it and the problem disappeared.

[janet]

holck

htop
top -i
ps -aux
to see what processes are happening.
Is sme server busy doing virus scanning, or busy processing & scanning incoming email messages or doing some other task ?

You need to isolate the source of the problem, ie is it sme server hanging, or just busy doing tasks, or is it an external hardware problem.

Try swapping the LAN NIC, cables, switch/hub etc etc etc
Is it only with one workstation ?
Does the problem occur with different workstation OS's ?

[CharlieBrady]

If I restart the server, everything seems to be working fine again, but of course this isn't very convenient doing every other day...

There are a few network drivvers which have had bugs which manifest like that. Google for your NIC driver name and RHEL/CentOS and see what you find. Replace with a different/better NIC if you can. Check your cables and switch.

[holck]

Charlie and Janet, thanks for your suggestions - I have digged a bit more into the problem

The problem occurred once more, just an hour ago. Nothing peculiar going on at the server. Devices already connected to the server, wireless or by cable, continued to work, but new devices suddenly couldn't connect anymore. The dhcpd process was running, ifconfig didn't show anything special, both eth0 and eth1 were up, no errors.

I stopped the dhcpd service, started dhcpd from the command line, and all was well.

Code: [Select]

/sbin/service dhcpd stop
/usr/sbin/dhcpd -d -cf /etc/dhcpd.conf -lf /var/lib/dhcpd/dhcpd.leases eth0
Killed the dhcpd process, and started the service - no errors, but dhcpd wasn't working.

Once more, stopped the dhcpd service, started dhcpd from the command line, and everything is well again.

Strange, isn't it?

[holck]

SORRY - my bad...

The problem seems to have occurred as a result from my editing /var/service/dhcpd/log/run . I some weeks ago I added the line 

Code: [Select]

/usr/local/bin/setuidgid smelog \, as I was trying to limit the amount of dhcpd logging in /var/log/messages, and then I forgot about it ...

Sorry for the trouble,
Jesper