Artur Linhart - Linux communication
2007-Jul-23 14:57 UTC
[Xen-users] ping in hvm domain not working correctly
Hello,
I have found a strange behavior if pinging from the HVM domain
(windows). I reproduced this behavior on different physical machines once in
DomU Windows 2003 Server Enterprise x64 R2 and once in Windows 2000 advanced
server (both with latest patches).
On both machines the ping response times seem to be generated randomly if
running the OS instance as DomU (also negative response times are generated
or received within the ICMP) - this could mean the ICMP packets are not
delivered correctly by xen to DomU?
If the remote host cannot beaccessed, then the timeout comes correctlz. Onlz
in the case the pinged host exists and answers such strange behavior occurs.
Does anybody met this problem and what can be the resolution for
this?
With best regards
Archie
P.S. Output of ping from the DomU:
C:\Documents and Settings\Administrator>ping www.linux.org
Pinging www.linux.org [198.182.196.56] with 32 bytes of data:
Reply from 198.182.196.56: bytes=32 time=5131ms TTL=51
Reply from 198.182.196.56: bytes=32 time=-8870ms TTL=51
Reply from 198.182.196.56: bytes=32 time=9185ms TTL=51
Reply from 198.182.196.56: bytes=32 time=-8995ms TTL=51
Ping statistics for 198.182.196.56:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 5131ms, Maximum = -8870ms, Average = 1073740936ms
C:\Documents and Settings\Administrator>ping www.linux.org
Pinging www.linux.org [198.182.196.56] with 32 bytes of data:
Reply from 198.182.196.56: bytes=32 time=14239ms TTL=51
Reply from 198.182.196.56: bytes=32 time=-9624ms TTL=51
Reply from 198.182.196.56: bytes=32 time=-14019ms TTL=51
Reply from 198.182.196.56: bytes=32 time=-4163ms TTL=51
Ping statistics for 198.182.196.56:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 14239ms, Maximum = -4163ms, Average = 1073738432ms
C:\Documents and Settings\Administrator>ping www.linux.org
Pinging www.linux.org [198.182.196.56] with 32 bytes of data:
Reply from 198.182.196.56: bytes=32 time=127ms TTL=51
Reply from 198.182.196.56: bytes=32 time=11026ms TTL=51
Reply from 198.182.196.56: bytes=32 time=-10796ms TTL=51
Reply from 198.182.196.56: bytes=32 time=-15530ms TTL=51
Ping statistics for 198.182.196.56:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 127ms, Maximum = -10796ms, Average = 1073738030ms
C:\Documents and Settings\Administrator>
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users
Martin Goldstone
2007-Jul-23 15:05 UTC
Re: [Xen-users] ping in hvm domain not working correctly
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I''m guessing your using more than 1 VCPU in your HVM domains? I encountered the same problem. I''m not entirely sure what causes the problem (it may have something to do with timing figures from different VCPUs), but I''m not experiencing any problems caused by this, just the inconvenience of an odd time. The ICMP packets are still being delivered properly. Sadly, I''ve not been able to find a solution, aside from using only 1 VCPU. Of course if you are using only 1 VCPU, then this may not be the cause of your problem. You could try pinning the VCPU to one physical CPU. I''m not sure it''ll have an effect, but it''s worth a try. Mart Artur Linhart - Linux communication wrote:> Hello, > > > > I have found a strange behavior if pinging from the HVM domain > (windows). I reproduced this behavior on different physical machines once in > DomU Windows 2003 Server Enterprise x64 R2 and once in Windows 2000 advanced > server (both with latest patches). > > On both machines the ping response times seem to be generated randomly if > running the OS instance as DomU (also negative response times are generated > or received within the ICMP) - this could mean the ICMP packets are not > delivered correctly by xen to DomU? > > > > If the remote host cannot beaccessed, then the timeout comes correctlz. Onlz > in the case the pinged host exists and answers such strange behavior occurs. > > > > Does anybody met this problem and what can be the resolution for > this? > > > > With best regards > > > > Archie > > > > P.S. Output of ping from the DomU: > > > > C:\Documents and Settings\Administrator>ping www.linux.org > > > > Pinging www.linux.org [198.182.196.56] with 32 bytes of data: > > > > Reply from 198.182.196.56: bytes=32 time=5131ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=-8870ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=9185ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=-8995ms TTL=51 > > > > Ping statistics for 198.182.196.56: > > Packets: Sent = 4, Received = 4, Lost = 0 (0% loss), > > Approximate round trip times in milli-seconds: > > Minimum = 5131ms, Maximum = -8870ms, Average = 1073740936ms > > > > C:\Documents and Settings\Administrator>ping www.linux.org > > > > Pinging www.linux.org [198.182.196.56] with 32 bytes of data: > > > > Reply from 198.182.196.56: bytes=32 time=14239ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=-9624ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=-14019ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=-4163ms TTL=51 > > > > Ping statistics for 198.182.196.56: > > Packets: Sent = 4, Received = 4, Lost = 0 (0% loss), > > Approximate round trip times in milli-seconds: > > Minimum = 14239ms, Maximum = -4163ms, Average = 1073738432ms > > > > C:\Documents and Settings\Administrator>ping www.linux.org > > > > Pinging www.linux.org [198.182.196.56] with 32 bytes of data: > > > > Reply from 198.182.196.56: bytes=32 time=127ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=11026ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=-10796ms TTL=51 > > Reply from 198.182.196.56: bytes=32 time=-15530ms TTL=51 > > > > Ping statistics for 198.182.196.56: > > Packets: Sent = 4, Received = 4, Lost = 0 (0% loss), > > Approximate round trip times in milli-seconds: > > Minimum = 127ms, Maximum = -10796ms, Average = 1073738030ms > > > > C:\Documents and Settings\Administrator> > > > > ------------------------------------------------------------------------ > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFGpMOsRnwIDhcMR9MRAolEAJ9iEcM2VBbKV+HH85FneKHtWVcvAACdHt58 btgg/8AaqaM2Gk9MEOdh+tQ=+O3f -----END PGP SIGNATURE----- _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Stephan Seitz
2007-Jul-23 15:20 UTC
Re: [Xen-users] ping in hvm domain not working correctly
If you encounter this problem only by using more than one VCPUs, you might try to pin the HVM domU to cores located at one particular CPU. (e.g. cpu=''2,3'') This is just a guess, but it might solve time shift problems. Greetings Stephan Martin Goldstone schrieb:> I''m guessing your using more than 1 VCPU in your HVM domains? I > encountered the same problem. I''m not entirely sure what causes the > problem (it may have something to do with timing figures from different > VCPUs), but I''m not experiencing any problems caused by this, just the > inconvenience of an odd time. The ICMP packets are still being > delivered properly. Sadly, I''ve not been able to find a solution, aside > from using only 1 VCPU. > > Of course if you are using only 1 VCPU, then this may not be the cause > of your problem. You could try pinning the VCPU to one physical CPU. > I''m not sure it''ll have an effect, but it''s worth a try. > > Mart > > Artur Linhart - Linux communication wrote: >> Hello, > > > >> I have found a strange behavior if pinging from the HVM domain >> (windows). I reproduced this behavior on different physical machines once in >> DomU Windows 2003 Server Enterprise x64 R2 and once in Windows 2000 advanced >> server (both with latest patches). > >> On both machines the ping response times seem to be generated randomly if >> running the OS instance as DomU (also negative response times are generated >> or received within the ICMP) - this could mean the ICMP packets are not >> delivered correctly by xen to DomU?_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users