This is an archived copy of the Xen.org mailing list, which we have
preserved to ensure that existing links to archives are not broken.
The live archive, which contains the latest emails, can be found at
http://lists.xen.org/
I had this same problem a one year ago. My host have one processor Icore7 and motherboard Asus , I don´t remember the name. I did same process like you , backup using DD before I make a copy img for other machine using cp for external volume NFS. The host any time freeze . I belive the problem is raid controller of motheboard because my stuff team change SO this host for Windows 2008 still problem . Windows Server 2008 and Linux aren´t compatible with this motherboard. Til today I don´t found any motheboard for Icore7 100% compatible for Linux so for use XEN production enviroment I use IBM XEON or processor XEON with motherboard INTEL for server, never motheboard Desktop .
the storage serer is a Xeon Quad 2.13GHz (E5506) with 8GB
DDR3 of RAM, a SuperMicro motherboard (X8ST3) and a LSI Logic SAS
Controller (MegaRAID SAS 1078), it also has 16 hard disks in
RAID50 and now is sharing with iSCSI 44 LVM partitions under the
CentOS 5.5 S.O.
Most of the time, the system shows 100% idle for all cores,
the load avarage is 0.09 0.04 0.00 and used 1.3GB of RAM with
772MB buffer, 160MB cache.
i have 4 debian squeeze with xen 4 and all them with 4
to 6 HVM virtual machines. All the VMs are located in a
fith machine (A SuperMicro LSI Controller as storage).
The host servers are Intel i7 with 24GB fo RAM (Debian
Squeeze + Xen4) and the storage a Xen 3.0 with 8GB of RAM
(CentOS 5.5 + TGTD) and thery are
all connect with a gigabit network dedicated for the
iSCSI.
My problem is, when working at the storage, doing
backup procedures with the command dd and using the bzip2,
the IOWait increases a litle bit, the normal state is
close to 0%, but when copying or moving the files
generated by the backup, usually bigger than 2G, the
IOWait increases too much, reaching 100%, in cases in all
4 cores, making the virtual machines stop responding and
even if i stop the procedure, the virtual machines doesnt
comes back, been necessary a reboot or shutdown.
Does anyone knows about this problem ? And there is
any configuration that i could do in the iSCSI-Target and
Initiator to increase de performance? And a configuration
in the virtual machines filesystem to prevent it from
stopping?
Thanks you all,
Regards,
Tny M.
_______________________________________________
What is the specs of your storage server?
It sounds like the storage server is underpowered or
overloaded quite a lot.