Hello,
I have four Xen servers running Xen 4.0.1 hypervisor and OpenSuSE 11.4
for Dom0. I have a NFS server hosting all DomU images exported to all
Xen servers.
Most of my DomU's are OpenSuSE but various versions. Most DomU's are
paravirtualised, though there are some less important old (2.4 kernel)
DomU's that are fully virtualised.
I noticed that if I reduce the number of VPCU's available to an OpenSuSE
DomU, live migration of that DomU doesn't succeed - it migrates, but the
resulting DomU on the target server is not responsive and is using up
100% of CPU (doing 'nothing'). The same happens if I use 'vcpu_avail' or
'maxvcpus' to start the DomU with a lower number of VCPU's, while still
being able to add some later.
No errors are logged in /var/log/xen/* and live migration works for
Centos DomU's, which suggests that the problem is in the DomU
(kernel)... Unfortunately I have no idea how to 'catch' what's going on
there.
Another issue is that even when live migration succeeds, the 'xm
console' on the target Xen server is not working anymore for the
migrated DomU.
Then there's HVM and PV drivers I created an OpenSuSE 11.4 HVM DomU,
which worked OK. But after installing xen-kmp-default, the DomU doesn't
boot anymore. It loads the initrd, it also mounts the root filesystem,
but then it 'hangs' at 'Loading drivers, configuring devices'. After a
while, there is a shower of I/O errors:
end_request: I/O error, dev sda, sector XXXXXXXXX
Other then installing the 'xen-kmp' packages, should I do something else?
Danilo
--
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
