Defining network Paramus using virt-install and kickstart for VM install
by JWH
I want to create a fedora VM client on a previously installed fedora bare metal host.
The network definition on the host is 192.168.1.x/24. I want the guest to also be on 192.168.1.x/24.
Dnsmasq has been disabled and stopped on the host.
1) Do I need to do anything else on the host?
2) What do I use for network arguments to virtual-install?
3) What do I use for network arguments to kickstart?
Older docs say I need to modify config files in /etc/sysconfig, is that really necessary?
Thanks.
Sent from my iPad
5 years, 10 months
Re: NFS mount from VM -
by Rick Stevens
On 08/13/2018 11:26 AM, Bob Goodwin wrote:
> On 08/13/18 13:53, Rick Stevens wrote:
>> On 08/13/2018 10:08 AM, Bob Goodwin wrote:
>>> I have an Fedora 28 computer with Fedora 27 running in virtual manager.
>>> My NFS server refuses its connection although the Fedora28 computer
>>> itself connects routinely. Ping and ssh can connect to the NFS server
>>> from the VM. Stopping Firewalld and setenforce 0 in the VM make no
>>> difference.
>>>
>>> NFS exports is:
>>> [bobg@ASRock-J3455M ~]$ cat /etc/exports
>>> /home/exports
>>> 192.168.1.0/24(rw,sync,insecure,no_root_squash,no_subtree_check)
>>>
>>> Can anyone suggest what I might have wrong?
>> The VM is probably on the 192.168.122.0/24 network (that's the default
>> virbr0 network set up by libvirt--check via "ip addr show" on the VM)
>> and your export is only to the 192.168.1.0/24 network so NFS mount
>> requests from 192.168.122.0/24 are rejected by the server.
>> ----------------------------------------------------------------------
>> - Rick Stevens,
> +
>
> That's the problem, everything in the VM shows up as 192.168.122.whatever.
>
> Normally my router assigns the addresses. How do I fix this? will it
> need a fixed address? The computer it is living in is 192.168.1.83 and I
> assumed, out of ignorance of this default, it would be the same.
Using the default NAT networking used in libvirt causes libvirt (well,
actually dnsmasq) to act as the DNS server for the VMs.
You can change to using a different network model for the VM (choose the
macvtap tied to your host's NIC). My host machine has two NICs, so I'd
have to choose macvtap on eno1 or enp4s0, depending on which network I
want it on. Have a look at this wiki page:
https://wiki.libvirt.org/page/Networking#Bridged_networking_.28aka_.22sha...
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
- -
- Vegetarian: Old Indian word for "lousy hunter" -
----------------------------------------------------------------------
5 years, 9 months
Re: After an update, VM's no longer have Internet access
by Ed Greshko
On 1/27/19 7:48 AM, Patrick O'Callaghan wrote:
> Same here. To eliminate some variables, I turned off my dnsmasq
> service, disabled it and rebooted. The problem is still there: for a
> few moments the guests are network-reachable, then they aren't. They
> may come back, they may not. Or one does and the other doesn't. It's
> completely unpredictable. If I could even figure out which component is
> causing the problem I could BZ it, but nothing stands out.
>
> I'll keep looking but I'm seriously considering a complete system
> reinstall, something I haven't done in about 5 years, in case some
> cruft from earlier iterations of Fedora is somehow lurking in the
> shadows.
Well, I can't say that I've ever seen "intermittent" problems like that caused by SW. But
since the host and guest are on the same HW it seems to be the only thing that makes sense.
The only thing that comes to mind is that communication on a LAN with IPv4 takes place
based on the MAC address and ARP request/response. If somehow guest obtained the same MAC
address for their interfaces one may see odd behavior.
--
Right: I dislike the default color scheme Wrong: What idiot picked the default color scheme
5 years, 4 months
Re: After an update, VM's no longer have Internet access
by Patrick O'Callaghan
On Mon, 2019-01-28 at 06:18 +0800, Ed Greshko wrote:
> If you use wireshark to monitor just vnet0 and do an ssh to the guest do you see an ARP
> request/response happen first? Is it correct?
>
> [...]
Even without trying the ssh there is a constant traffic of ARP requests
with no replies:
52:54:00:b0:20:88 ff:ff:ff:ff:ff:ff ARP 42 Who has 192.168.122.1? Tell 192.168.122.167
52:54:00:8b:88:60 is the vnet0 interface. 192.168.122.1 is the gateway,
192.168.122.167 is the guest.
Nothing ever comes back. IOW the guest is trying to do ARP resolution
but nothing is answering it (avahi-daemon is running, as is the libvirt
copy of dnsmasq). Also:
$ ip neigh|grep 122
192.168.122.167 dev virbr0 lladdr 52:54:00:b0:20:88 STALE
192.168.122.193 dev virbr0 lladdr 52:54:00:1d:55:89 STALE
Those are the two guest addresses.
> [egreshko@meimei .ssh]$ sudo firewall-cmd --info-zone=public
> public (active)
> target: default
> icmp-block-inversion: no
> interfaces: enp2s0 vnet0 wlp4s0
> sources:
> services: dhcpv6-client dns kde-connect mdns ssh
> ports:
> protocols:
> masquerade: no
> forward-ports:
> source-ports:
> icmp-blocks:
> rich rules:
Nothing to remark on there I think. I have some extra ports and
services enabled but that's to be expected.
poc
5 years, 4 months
Re: After an update, VM's no longer have Internet access
by Ed Greshko
On 1/28/19 7:12 AM, Patrick O'Callaghan wrote:
> On Mon, 2019-01-28 at 06:18 +0800, Ed Greshko wrote:
>> If you use wireshark to monitor just vnet0 and do an ssh to the guest do you see an ARP
>> request/response happen first? Is it correct?
>>
>> [...]
> Even without trying the ssh there is a constant traffic of ARP requests
> with no replies:
>
> 52:54:00:b0:20:88 ff:ff:ff:ff:ff:ff ARP 42 Who has 192.168.122.1? Tell 192.168.122.167
>
> 52:54:00:8b:88:60 is the vnet0 interface. 192.168.122.1 is the gateway,
> 192.168.122.167 is the guest.
>
> Nothing ever comes back. IOW the guest is trying to do ARP resolution
> but nothing is answering it (avahi-daemon is running, as is the libvirt
> copy of dnsmasq). Also:
>
> $ ip neigh|grep 122
> 192.168.122.167 dev virbr0 lladdr 52:54:00:b0:20:88 STALE
> 192.168.122.193 dev virbr0 lladdr 52:54:00:1d:55:89 STALE
>
> Those are the two guest addresses.
Humm.... I see
37 67.694929326 RealtekU_f3:3f:02 RealtekU_9a:e8:49 ARP 42 Who has
192.168.122.1? Tell 192.168.122.86
38 67.694969398 RealtekU_9a:e8:49 RealtekU_f3:3f:02 ARP 42 192.168.122.1
is at 52:54:00:9a:e8:49
[egreshko@meimei ~]$ ip neigh|grep 122
192.168.122.86 dev virbr0 lladdr 52:54:00:f3:3f:02 REACHABLE
(Prior to an ssh it was STALE even with ARP traffic)
>> [egreshko@meimei .ssh]$ sudo firewall-cmd --info-zone=public
>> public (active)
>> target: default
>> icmp-block-inversion: no
>> interfaces: enp2s0 vnet0 wlp4s0
>> sources:
>> services: dhcpv6-client dns kde-connect mdns ssh
>> ports:
>> protocols:
>> masquerade: no
>> forward-ports:
>> source-ports:
>> icmp-blocks:
>> rich rules:
> Nothing to remark on there I think. I have some extra ports and
> services enabled but that's to be expected.
>
I was noting the difference between yours...
interfaces: enp3s0 p3p1 virbr0 virbr0-nic
and mine
interfaces: enp2s0 vnet0 wlp4s0
--
Right: I dislike the default color scheme Wrong: What idiot picked the default color scheme
5 years, 4 months
Re: After an update, VM's no longer have Internet access
by Patrick O'Callaghan
On Wed, 2019-01-30 at 13:01 +0000, Patrick O'Callaghan wrote:
> I want to try one more thing: leaving the Fedora guest on
> NAT and changing the Windows guest to macvtap (since I don't need to
> connect into it).
Interesting. I changed the Windows guest to macvtap and didn't touch
the Fedora guest. Starting with both guests shut down, the Windows
guest comes up (though it doesn't have a local IPv4 address from the
host's viewpoint, as expected). However the Fedora guest - still on NAT
- doesn't have an IPv4 address either and is completely disconnected.
dnsmasq (libvirts's version) is running normally. Both guests have IPv6
addresses and respond to ping6's.
I'm getting a growing feeling that something is really screwed up with
my installation of libvirt. I hesitate to wimp out and reinstall it,
but I'm running out of ideas.
poc
5 years, 4 months
Re: Using bind for a local caching name server, is this
configuration correct?
by Tim
On Mon, 2019-07-01 at 18:41 -0700, stan via users wrote:
> I think the failure might have something to do with NetworkManager.
> It seems that it has no way to set it to use a local bind / named
> instance as its nameserver. It always uses DNS servers set by the
> router (etc/resolv.conf), or systemd-resolved, or dnsmasq.
Do you mean you can't find anywhere to set such options, or it ignores
what you do?
Using MATE, here, but I can "Edit Connections" from the network icon in
the title bar. Make my way through to the IPv4 settings, and choose
options where everything is left up to DHCP, I can override various
aspects, or go fully manual. Likewise with IPv6.
I haven't done it for a while, and don't want to try in the middle of
doing my mail, but it's always worked in the past. I could set up one
profile to use my LAN DNS server, another to use my router's.
4 years, 11 months
Re: open ssh
by Tom Horsley
On Fri, 4 Oct 2019 12:40:58 +0300
Angelo Moreschini wrote:
> Do I have to Installase BIND? And then what else is needed? ...
My router can run a DNS name server, you might want to check
if yours has that feature. Otherwise running bind or dnsmasq
somewhere on the local network would be the way to go.
4 years, 8 months
Re: open ssh
by Fred Smith
On Fri, Oct 04, 2019 at 09:15:06AM -0400, Tom Horsley wrote:
> On Fri, 4 Oct 2019 12:40:58 +0300
> Angelo Moreschini wrote:
>
> > Do I have to Installase BIND? And then what else is needed? ...
>
> My router can run a DNS name server, you might want to check
> if yours has that feature. Otherwise running bind or dnsmasq
> somewhere on the local network would be the way to go.
I set up DNS/Bind9 and NTP on a raspberry pi somewhat over a year ago.
cheap, doesn't tie up a bigger machine someone has on desktop (like, e.g.,
me) and uses minimal electricity.
--
---- Fred Smith -- fredex(a)fcshome.stoneham.ma.us -----------------------------
But God demonstrates his own love for us in this:
While we were still sinners,
Christ died for us.
------------------------------- Romans 5:8 (niv) ------------------------------
4 years, 8 months
where is the nasty cache?
by Tom Horsley
We shuffled some systems around at work. An NFS server now has
a new IP address. On my fedora 30 desktop I see:
nslookup <server> shows the new IP address
host <server> shows the new IP address
ping <server> still pings the OLD address!!!
mount /mountpoint apparent still uses the old IP because it times out.
I'm not running nscd. I'm not running systemd-resolv. I'm
not running dnsmasq or bind as a local cache. I don't have
entries in /etc/hosts for the old IP.
So where is this "helpful" cache that has the old
IP stashed?
4 years, 7 months
