Re: [opensuse-factory] wicked dhcp & virtualization
Michael Melcher wrote:
Hi Per,
Am 24.11.2015 09:23 schrieb "Per Jessen" <per@computer.org <mailto:per@computer.org>>:
Michael Melcher wrote:
Hi,
I'm setting up a virtualization host based on Leap.
We have a Windows Server that acts as AD and DHCP Server. Unfortunately wicked seems to be unable to obtain an IP Address from the DHCP Server.
Wicked reports an error: option 121 has zero length error: unable to parse dhcp4 response.
according to bugzilla.novell.com <http://bugzilla.novell.com>
id=918069 that bug has been fixed in
wicked 0.6.19. Leap uses 0.6.27
This leaves me to the following setup: eth0: Static IP vnet0: Static IP dhcpd listening only on vnet0 to prevent conflicts with the main DHCP Server.
The issue here is, that vnet0 is not available until the first VM has been started. Thus the procedure currently is: 1: boot first VM 2: assign IP to vnet0 (ifconfig vnet0 xxx.xxx.xxx.xxx) 3: start dhcpd
Surely I could script something to do this automatically for me, but the goal should be to fix wicked to get an IP address from the Windows DHCP Server.
Hi Michael
I have a similar setup, based on xen and 13.2 - no problems with guests acquiring addresses via dhcp (from a dhcp server external to the xen host).
It's not only the guests that are unable to acquire an IP from the windows server, but also the host itself.
Okay. My first thought would be - network issue. Have you checked on the host and the guest with tcpdump? Just to see what's going on - i.e. the client sending out a DHCPDISCOVER, the Windows server responding with a DHCPOFFER, followed by DHCPREQUEST and DHCPACK. It ought to give you some clues.
I can't quite work out what you're trying to do - do you want your virtual hosts to use the DHCP server on your Windows machine or do you want to run dhcpd on a virtual host?
I'm currently forced to run dhcpd on the vm host (listening on vnet0 only!) As all guests using wicked/dhclient rather than dhcpcd can't get an IP either.
I could probably try a live dvd of a distro that uses dhcpcd in a vm to see if that works. But so far only suse (sle included) fails to get an IP. Even on real hardware.
So any virtualization-related network issue can be ignored.
An Ubuntu live dvd we had to use today on real hardware got an IP. Same as acronis boot dvd (afaik it uses linux as base system too. But not 100%sure)
Also our gentoo systems (both on real hw and in vms in virtualbox) can obtain IPs from the DHCP Server.
That is weird. What does "systemctl status wickedd-dhcp4" say? On any system that doesn't work.
As far as it looks like the issue is with wicked/dhclient producing the above mentioned error - which should be fixed according to the mentioned bug report.
Right - I assume you're no longer seeing that particular error message? That the bugreport says it's fixed does not mean the fix has actually made it to your systems :-)
On a side note: I'm not the admin of the windows server and thus can't check the configuration or adjust it.
That's fine, we know it works for your other Gentoo and *buntu systems. /Per -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
-- Kind regards Michael Melcher Sent from my phone Am 24. November 2015 18:36:14 MEZ, schrieb Per Jessen <per@computer.org>:
Michael Melcher wrote:
Hi Per,
Am 24.11.2015 09:23 schrieb "Per Jessen" <per@computer.org <mailto:per@computer.org>>:
Michael Melcher wrote:
Hi,
I'm setting up a virtualization host based on Leap.
We have a Windows Server that acts as AD and DHCP Server. Unfortunately wicked seems to be unable to obtain an IP Address
from
the DHCP Server.
Wicked reports an error: option 121 has zero length error: unable to parse dhcp4 response.
according to bugzilla.novell.com <http://bugzilla.novell.com> id=918069 that bug has been fixed in wicked 0.6.19. Leap uses 0.6.27
This leaves me to the following setup: eth0: Static IP vnet0: Static IP dhcpd listening only on vnet0 to prevent conflicts with the main DHCP Server.
The issue here is, that vnet0 is not available until the first VM has been started. Thus the procedure currently is: 1: boot first VM 2: assign IP to vnet0 (ifconfig vnet0 xxx.xxx.xxx.xxx) 3: start dhcpd
Surely I could script something to do this automatically for me, but the goal should be to fix wicked to get an IP address from the Windows DHCP Server.
Hi Michael
I have a similar setup, based on xen and 13.2 - no problems with guests acquiring addresses via dhcp (from a dhcp server external to the xen host).
It's not only the guests that are unable to acquire an IP from the windows server, but also the host itself.
Okay. My first thought would be - network issue. Have you checked on the host and the guest with tcpdump? Just to see what's going on - i.e. the client sending out a DHCPDISCOVER, the Windows server responding with a DHCPOFFER, followed by DHCPREQUEST and DHCPACK. It ought to give you some clues.
No, I haven't checked that (yet). However, I checked on the option 121 thingy and figured it's an actual configuration in the dhcp server for classless static routes. Our server seems to send an empty value for it.
I can't quite work out what you're trying to do - do you want your virtual hosts to use the DHCP server on your Windows machine or do you want to run dhcpd on a virtual host?
I'm currently forced to run dhcpd on the vm host (listening on vnet0 only!) As all guests using wicked/dhclient rather than dhcpcd can't get an IP either.
I could probably try a live dvd of a distro that uses dhcpcd in a vm to see if that works. But so far only suse (sle included) fails to get an IP. Even on real hardware.
So any virtualization-related network issue can be ignored.
For now. The goal is to get both, the host and the vms to get an IP from the main dhcp server instead of from my work around dhcpd.
An Ubuntu live dvd we had to use today on real hardware got an IP. Same as acronis boot dvd (afaik it uses linux as base system too. But not 100%sure)
Also our gentoo systems (both on real hw and in vms in virtualbox) can obtain IPs from the DHCP Server.
That is weird.
Not so much considering they use a different dhcp client that obviously acts different than wicked/dhclient
What does "systemctl status wickedd-dhcp4" say? On any system that doesn't work.
I'll check that (and the tcpdump) tomorrow at work and let you know.
As far as it looks like the issue is with wicked/dhclient producing the above mentioned error - which should be fixed according to the mentioned bug report.
Right - I assume you're no longer seeing that particular error message?
Running wicked-dhcp-test (or whatever the binary is called,can't check right now) gives that error still.
That the bugreport says it's fixed does not mean the fix has actually made it to your systems :-)
It should mean exactly that, considering the bug was fixed in 0.6.19 and I'm using 0.6.27 :-)
On a side note: I'm not the admin of the windows server and thus can't check the configuration or adjust it.
That's fine, we know it works for your other Gentoo and *buntu systems.
/Per
-- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
participants (2)
-
Michael Melcher -
Per Jessen