[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: dhclient keeps sending old hostname [solved]



On Tue, 2011-06-21 at 12:59 -0700, Ross Boylan wrote:
> I have a kvm virtual machine whose disk was initially a copy of another
> virtual machine.  The original machine's hostname was vm-lenny00. I
> changed the new machine to vm-migrate07
> in /etc/hostname, /etc/hosts, /etc/dhcp3/dhclient.conf (send host-name
> "vm-migrate07";), /etc/dhcp/dhclient.conf, and other places (e.g., mail
> and backup system).  I tried several reboots of the vm, although it is
> possible I have only edited some of those files since the last reboot.
> In particular, /etc/dhcp/dhclient.conf is a recent change.
> 
I was able to get the name right by doing
nmcli con down id "Auto eth0"
and then the corresponding con up command.  This got me anew dhclient,
which did use the current, correct, host-name in dhcp.conf.

I speculate that my previous efforts failed because I had not changed
the (correct) dhclient.conf before reboot.  Presumably before upgrade
there was only one such file (under dhcp3/).

NetworkManager, not ifupdown, is managing my network connection.  Using
the latter produced
# ifdown eth0
Ignoring unknown interface eth0=eth0.

It remains unclear to me exactly how dhclient was invoked, or whether I
could have restarted it outside the networkmanager machinery.

I guess dhclient3 is still running because it was running before the
upgrade.

Ross

P.S.  I used VNC to keep connected while the interface went down.  Maybe
someone brave could get away with 
nmcli con down id "Auto eth0"; nmcli con up id "Auto eth0"
> The new vm has had a distinct MAC address from the start.
> 
> dnsmasq on my host (lenny) system continues to report
> Jun 20 23:46:48 markov dnsmasq[4243]: DHCPREQUEST(tap0) 192.168.8.88 52:54:a0:12:07:01
> Jun 20 23:46:48 markov dnsmasq[4243]: DHCPACK(tap0) 192.168.8.88 52:54:a0:12:07:01 vm-lenny00
> This is a mess, since the real vm-lenny00 is also talking (below is from
> a reboot of that vm):
> Jun 20 23:30:32 markov dnsmasq[4243]: DHCPRELEASE(tap0) 192.168.8.140 52:54:a0:12:01:01
> Jun 20 23:30:53 markov dnsmasq[4243]: DHCPDISCOVER(tap0) 192.168.8.140 52:54:a0:12:01:01
> Jun 20 23:30:53 markov dnsmasq[4243]: DHCPOFFER(tap0) 192.168.8.140 52:54:a0:12:01:01
> Jun 20 23:30:53 markov dnsmasq[4243]: DHCPREQUEST(tap0) 192.168.8.140 52:54:a0:12:01:01
> Jun 20 23:30:53 markov dnsmasq[4243]: DHCPACK(tap0) 192.168.8.140 52:54:a0:12:01:01 vm-lenny00
> 
> /var/lib/dhcp/dhclient-xxxxx-eth0.lease (on vm-migrate07) shows only
> entries on with
>   option host-name "vm-lenny00";
> 
> Since I'm accessing the vm via ssh I'm reluctant to bring networking
> down.  I do have vnc setup if necessary.  Since I'm in the middle of an
> upgrade, I'm also reluctant to restart the vm.  At one point after the
> upgrade the disk filled.
> 
> I used the vm to test upgrade to squeeze, and so it is now basically
> running squeeze.  I tried to reset the name (and did reboots) before the
> upgrade.  vm-migrate07 has network-manager installed, and it appears the
> both the old (dhcp3) and new (dhcp) clients are running:
> # ps -Alc | grep dhc
> 5 S     0  1052     1 TS   19 -  1593 -      ?        00:00:00 dhclient3
> 4 S     0 12286 12276 TS   19 -  1687 -      ?        00:00:00 dhclient
> 
> # netstat -nlp | grep dhc
> udp        0      0 0.0.0.0:68              0.0.0.0:*                           12286/dhclient
> udp        0      0 0.0.0.0:68              0.0.0.0:*                           1052/dhclient3
> 
> So I'm not entirely sure who is in control.  I assume the vm-lenny00
> name in the dnsmasq exchanges is coming from the client, but perhaps
> it's not.
> 
> I tried using omshell on the vm to  talk to dhclient, but it says
> "dhcpctl_connect: no more" when I try to connect.  The docs don't
> indicate any way to signal it to re-read its configuration file.
> 
> How can I fix the name of the system?
> 
> Thanks for any assistance.
> Ross
> 
> 
> 
> 


Reply to: