0

Once per day the network on eth0 is going down and we need to do sudo ifdown eth0; sudo ifup eth0 to get the network up.

Here is the syslog:

Feb 11 12:48:01 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 12:52:35 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 12:56:23 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:00:28 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:04:29 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:09:16 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:13:53 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:18:16 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:22:25 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:26:52 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:30:44 www-tech-1 dhclient: DHCPREQUEST of address> on eth0 to 131.121.113.228 port 67
Feb 11 13:31:49 www-tech-1 dhclient: There is already a pid file /var/run/dhclient.eth0.pid with pid 3198
Feb 11 13:31:49 www-tech-1 dhclient: Listening on LPF/eth0/00:e0:81:49:fc:e0
Feb 11 13:31:49 www-tech-1 dhclient: Sending on LPF/eth0/00:e0:81:49:fc:e0
Feb 11 13:31:49 www-tech-1 dhclient: DHCPRELEASE on eth0 to 131.121.113.228 port 67
Feb 11 13:31:49 www-tech-1 dhclient: There is already a pid file /var/run/dhclient.eth0.pid with pid 134519072
Feb 11 13:31:50 www-tech-1 dhclient: Listening on LPF/eth0/00:e0:81:49:fc:e0
Feb 11 13:31:50 www-tech-1 dhclient: Sending on LPF/eth0/00:e0:81:49:fc:e0
Feb 11 13:31:52 www-tech-1 dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 8
Feb 11 13:31:52 www-tech-1 dhclient: DHCPREQUEST of 131.121.14.17 on eth0 to 255.255.255.255 port 67
Feb 11 13:31:53 www-tech-1 kernel: [265383.991682] eth0: no IPv6 routers present

Please check the last portion of this syslog.
Can anybody help me?

Charly
  • 321
  • 4
  • 6
  • 16
  • You say that this server has a static IP, yet it's obviously running dhclient. Do you really mean that it's your Internet gateway and it's asking for a DHCP address externally (eth0) and using a static IP internally (eth1)? – MikeyB Feb 11 '10 at 14:54

3 Answers3

1

I'll assume this is regarding a server since it's on SF.

You should really use static IP addresses for your servers. Try using netconfig (run from terminal) to set up static IP for your machine. You should also look into your DHCP-settings in whatever device you use as a DHCP-server. Make an exception for your servers new static IP.

Kiwi
  • 357
  • 1
  • 4
  • netconfig is only for Red Hat btw. If you have another dist i suggest configure manually in /etc/network/interfaces. – Kiwi Feb 11 '10 at 13:40
  • the server is using static IP – Charly Feb 11 '10 at 14:18
  • @Kiwi, Unless, of course, you're running Slack. Then you need to edit /etc/rc.d/rc.inetX.conf.... :) – jscott Feb 11 '10 at 14:18
  • and we are using Ubuntu Hardy 8.04 upgraded from Ubuntu dapper. Do you think it sis some problem with IPv6 – Charly Feb 11 '10 at 14:19
  • are you knowingly using IPv6? – Andy Feb 11 '10 at 14:36
  • No, but do you know the resaon for this log Feb 11 13:31:53 www-tech-1 kernel: [265383.991682] eth0: no IPv6 routers present – Charly Feb 11 '10 at 15:01
  • It's just looking for autoconfig IPv6 info. I don't recall ever seeing that message before, but letting it seek IPv6 autoconfig (even when not available, before we got IPv6 up and running here) has never caused me any issues. – Brian Knoblauch Feb 11 '10 at 15:41
1

You start off with IP address '131.121.113.228'. The requests for this this address seem to go unanswered, there is then a DHCP release and then you finally get address '131.121.14.17'.

Did you manually do a DHCP release? If not, you could have a couple of different things going on here:

  • How is the stability of the DHCP server? Is it responding to requests reliably?
  • The lease time advertised by the DHCP server may be abnormally short.
  • Your DHCP server may not have enough IP addresses available in its scope.
  • You may have more than one DHCP server handing out addresses. The first server the client received an address from fails to respond to the request for the original address which may be why it gets a new address finally.

It would also be a good idea to disable IPv6 if it is not in use on your network. This may improve performance.

Peter
  • 5,453
  • 1
  • 26
  • 32
0

What are you using for a DHCP server? I've had some really wonky issues with DHCP leases not always renewing properly when serving DHCP from GNU/Linux boxes. Haven'd had the same issues with BSD, Solaris, or appliance (which one would expect have GNU/Linux internally) DHCP servers. Either way, I tend to just avoid DHCP now for smaller, controlled networks. If it's not a necessary function (such as to support end user convenience) I recommend not deploying it.

Brian Knoblauch
  • 2,196
  • 2
  • 32
  • 48