Hi there,
I am using xen hypervisor for analysing guest to guest communication throughput.
I have Quad Core Intel Xeon server with F11 installed on it.
My Dom0 and DomU specs are:
Dom0: #uname -ra
Linux HPCNL-SR-2 2.6.30-rc6-tip #1 SMP Tue Aug 25 12:26:16 PKST 2009 i686 i686
i386 GNU/Linux
DomU: # uname -ra
Linux F11-G5S2 2.6.29.4-167.fc11.i686.PAE #1 SMP Wed May 27 17:28:22 EDT 2009
i686 i686 i386 GNU/Linux
while running netperf benchmark I have got such results in different cases
DomU to Dom0 2000 to 3000Mbps
Dom0 to DomU also ok
But in case of DomU to DomU throughput is around 0.14 to 0.29 Mbps
I have found these warnings in my log files
/var/log/xen/dom1
domid: 6
Warning: vlan 0 is not connected to host network
qemu: ignoring not-understood drive
`sda2''
Watching /local/domain/0/device-model/6/logdirty/next-active
Watching /local/domain/0/device-model/6/command
char device redirected to /dev/pts/4
/usr/src/xen-3.4-testing.hg/tools/ioemu-dir/hw/xen_blktap.c:628: Init blktap
pipes
xs_read(): vncpasswd get error.
/vm/fb477f56-bb98-c144-54a4-74dd4379b094/vncpasswd.
Also dom0 file /var/log/messages contains
Sep 14 14:37:10 HPCNL-SR-2 logger: /etc/xen/scripts/vif-bridge: iptables setup
failed. This may affect guest networking.
Is there any iptable issue in this case I have tried it after flushing all
iptables but no use.
Doing all in bridge mode of communication. Kindly help me identifying and the
resolving the issue.
Also please explain me Why in the newer versions there is no vif0.0 for
dom0? As said in running xen book "Herbert Xu has proposed a patch that
removes the need for this loopback device. In the future, there may be
no more Xen loop device in bridging mode"
In Domain0, backend vif0.0
is called the Xen loopback (or netloop) device. Its function is
different from the guest domain''s vif because the guest domain frontend
vif passes the packets in the guest domain to its proxy backend in the
driver domain (often Domain0). The Xen loopback device vif0.0 exists
only to enforce a data copy on the guest domain to Domain0 path. The
copy is desired because otherwise a packet sitting on the Domain0
receiving buffer would hold the guest domain''s page containing the
packet for an unbounded amount of time. There is no guarantee of when
the user process will read the data and free the socket buffer.I would
like to add one more thing in your knowledge that in my case guest to
internet speed is ok. wget between guests shows 35K/s to the max.and
the throughput in this case is 35.1K/s eta 4h 3m
In xen 3.4 how are they managing this packet sitting on the dom0 receiving
buffer for unbounded time?
Regards,
Fasiha
Keep up with people you care about with Yahoo! India Mail. Learn how.
Yahoo! India has a new look. Take a sneak peek http://in.yahoo.com/trynew
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
