Hi Luf,
Let me have a think about it and I'll get back to you on whether there's
a workaround or something needs to be enhanced in the engine logic.
Thanks for bringing this up!
Yours, Lior.
On 20/11/14 15:11, Finstrle, Ludek wrote:
Lior Vernia píše v St 19. 11. 2014 v 17:22 +0200:
On
htmlbodyLior Vernia píše v St 19. 11. 2014 v 17:22 +0200:
On 19/11/14 14:42, Finstrle, Ludek wrote:
Engine has both networks (ovirtmgmt as eth0 and data as eth1).
Internally the cluster should communicate via ovirtmgmt (installed using
internal name for ovirtmgmt IP and this network is
Hey,
Thanks, so how come node2 moves to non-operational state rather than
non-responsive (i.e. how come it's reachable from the engine)? The IP
address via which it was installed, belongs to which network interface?
bond1 which is used for data?
Lior.
On 18/11/14 12:17, Finstrle, Ludek wrote:
htmlbodyHi Lior,
Engine has both networks (ovirtmgmt as eth0 and data as eth1). Internally the
cluster should communicate via ovirtmgmt (installed using internal name for
ovirtmgmt IP and this network is strictly local to the ovirt cluster). I'm
reaching web of engine via data IP (public
On 19/11/14 14:42, Finstrle, Ludek wrote:
Hi Lior,
Engine has both networks (ovirtmgmt as eth0 and data as eth1).
Internally the cluster should communicate via ovirtmgmt (installed using
internal name for ovirtmgmt IP and this network is strictly local to the
ovirt cluster). I'm
htmlbodyHi,
yes that's exactly what I tried to describe. Thanks Ekin.
There is no switch between 2 hosts/nodes (for ovirtmgmt network). It's
similar to situation when I tried just one host/node and put the
ovirtmgmt network to not connected network interface (will connect it
once we buy second
Hi Luf,
Apologies, I probably don't understand the details of your L2 toplogy;
how come when you turn off one host it affects the link state of of the
other one's interfaces?...
Yours, Lior.
On 14/11/14 11:43, Finstrle, Ludek wrote:
Hi,
I have 2-node ovirt cluster. Both machines has 4
Hi Lior,
If I'm not mistaken, on Luf's topology, there is no switch on the
management bond (bond0) - just a cross or regular eth cable connecting
physical interfaces on both nodes. So when one node is powered down, there
is no link detected on the other node.
Regards,
On Mon, Nov 17, 2014 at
htmlbodyHi,
I have 2-node ovirt cluster. Both machines has 4 interfaces.
2 interfaces in bond1 for data
2 interfaces in bond0 for management (short connected directly between
servers)
switch0 -\ /-\ /- switch0
node1 node2
switch1 -/ \-/ \-
Hi,
Did you set your bond devices and IP addresses via ovirt's network management,
or by hand on the nodes?
In my experiences you need to set up ip addresses on the node's network
interface via ovirt' management, because it won't working.
Tibor
- Eredeti üzenet -
Hi,
I have
htmlbodyHi,
I setup it via oVirt. The problem is that there is no link detected on
the real interfaces for management (ovirtmgmt is UP but NO-CARRIER). If
I add another interface (with link detected) to the bond (temporarily as
I need it for backup) it's working properly.
Cheers,
Luf
Demeter
11 matches
Mail list logo