On Thu, Jan 6, 2011 at 9:16 PM, Jim Wildman <jim@xxxxxxxxxxxxx> wrote:
> On Thu, 6 Jan 2011, Nico Kadel-Garcia wrote:
>
>> On Thu, Jan 6, 2011 at 12:43 PM, Eero Volotinen <eero.volotinen@xxxxxx> wrote:
>>
>>>
>>> VirtManager + RHEL 5/6 does not require use of windows client.
>>
>> virt-manager is too stupid to be permitted to live outside of an
>> intensive care unit. It completely mishandles configuration options
>> that are easily available from the command line, such as the use of
>> mutliple disk drives at image setup time, and has very poor handling
>> of randomized NIC configurations, and it behaves extremely poorly over
>> remote X connections. It also has no clue and little capability to
>> properly handle pair-bonded connections for high reliability upstream
>> connectivity.
>>
>
> I'm also waiting (not holding my breath) for clear documentation
> from upstream of which settings are stored in which files. Or even
> documentation that is internally consistent with itself.
>
> say /etc/xen or /etc/libvirt/* or /var/lib/libvirt/*??
>
> Has that been cleaned up and documented?
I was testing it with KVM, for comparison to VMWare, and didn't get as
far as that. The network configuration, multiple disk at install time,
and dog-slow performance of KVM prevented further exploration. KVM was
being heavily advertised by RedHat so I wanted a look, and was
completely underwhelmed. The requisite "bridged" network ports have to
be set manually on the server, since the built-in network
configuration tools have no clue how to do it. This means network
pair-bonding has to be done in the guest domain, and it turned out
that PXE didn't work at all in the guests.
It was completely useless: hopefully RHEL 6 and CentOS 6 get it right.
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
