Hello everyone,
I''ve managed to configure and start Remus. However, it works only while
the protected server is not loaded. As soon as I start a
memory-intensive server application, the stream of Remus messages stops,
leaving me with an error message. I think the message is not always the
same, but here is what I''ve got from the last run:
PROF: suspending at 1289921328.328959
PROF: resumed at 1289921330.903716
xc: error: Error when flushing output buffer (32 = Broken pipe):
Internal error
tc filter del dev vif1.0 parent ffff: proto ip pref 10 u32
RTNETLINK answers: Invalid argument
We have an error talking to the kernel
Exception xen.remus.util.PipeException: PipeException(''tc failed: 2, No
such file or directory'',) in <bound method BufferedNIC.__del__ of
<xen.remus.device.BufferedNIC object at 0x7f27844df210>> ignored
I have tried to pin one physical core to Domain 0 and another one to the
protected domain, using vcpu-pin but it doesn''t help.
Currently I am running the Xen-unstable tree (last updated 3-4 days
ago), 2.6.32.25 pv-ops kernel as Dom-0 and 2.6.18.8 as Dom-U, but I had
a very similar problem with Xen 4.0.0 and 2.6.18.8 as Dom-0.
Any suggestions?
Thanks
Darko
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users
Hello, Seems that remus tries to use "tc" command. Think it''s for traffic shaping to take priority. Is the package installed on your dom0 ? Regards, JB Le 16/11/2010 16:45, Darko Petrović a écrit :> Hello everyone, > > I''ve managed to configure and start Remus. However, it works only while > the protected server is not loaded. As soon as I start a > memory-intensive server application, the stream of Remus messages stops, > leaving me with an error message. I think the message is not always the > same, but here is what I''ve got from the last run: > > PROF: suspending at 1289921328.328959 > PROF: resumed at 1289921330.903716 > xc: error: Error when flushing output buffer (32 = Broken pipe): > Internal error > tc filter del dev vif1.0 parent ffff: proto ip pref 10 u32 > RTNETLINK answers: Invalid argument > We have an error talking to the kernel > Exception xen.remus.util.PipeException: PipeException(''tc failed: 2, No > such file or directory'',) in <bound method BufferedNIC.__del__ of > <xen.remus.device.BufferedNIC object at 0x7f27844df210>> ignored > > I have tried to pin one physical core to Domain 0 and another one to the > protected domain, using vcpu-pin but it doesn''t help. > Currently I am running the Xen-unstable tree (last updated 3-4 days > ago), 2.6.32.25 pv-ops kernel as Dom-0 and 2.6.18.8 as Dom-U, but I had > a very similar problem with Xen 4.0.0 and 2.6.18.8 as Dom-0. > > Any suggestions? > > Thanks > Darko > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Thanks for your help, JB. Yes, I would say that it is installed. The "tc" command does exist. After all, everything works correctly with non-intensive workloads. Any other suggestions, anyone? On 11/16/2010 09:31 PM, Jean Baptiste FAVRE wrote:> Hello, > Seems that remus tries to use "tc" command. Think it''s for traffic > shaping to take priority. > Is the package installed on your dom0 ? > > Regards, > JB > > Le 16/11/2010 16:45, Darko Petrović a écrit : > >> Hello everyone, >> >> I''ve managed to configure and start Remus. However, it works only while >> the protected server is not loaded. As soon as I start a >> memory-intensive server application, the stream of Remus messages stops, >> leaving me with an error message. I think the message is not always the >> same, but here is what I''ve got from the last run: >> >> PROF: suspending at 1289921328.328959 >> PROF: resumed at 1289921330.903716 >> xc: error: Error when flushing output buffer (32 = Broken pipe): >> Internal error >> tc filter del dev vif1.0 parent ffff: proto ip pref 10 u32 >> RTNETLINK answers: Invalid argument >> We have an error talking to the kernel >> Exception xen.remus.util.PipeException: PipeException(''tc failed: 2, No >> such file or directory'',) in<bound method BufferedNIC.__del__ of >> <xen.remus.device.BufferedNIC object at 0x7f27844df210>> ignored >> >> I have tried to pin one physical core to Domain 0 and another one to the >> protected domain, using vcpu-pin but it doesn''t help. >> Currently I am running the Xen-unstable tree (last updated 3-4 days >> ago), 2.6.32.25 pv-ops kernel as Dom-0 and 2.6.18.8 as Dom-U, but I had >> a very similar problem with Xen 4.0.0 and 2.6.18.8 as Dom-0. >> >> Any suggestions? >> >> Thanks >> Darko >> >> _______________________________________________ >> Xen-users mailing list >> Xen-users@lists.xensource.com >> http://lists.xensource.com/xen-users >> >> > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Hello, Yes, everything works great with non-intensive load. Also, maybe remus does not try to do traffic shapin with non-intensive load :) Another suggestion, so :) Does vif1.0 exists ? As remus tries to execute tc command on vif1.0 interface, you may have an issue here. Regards, JB Le 17/11/2010 14:45, Darko Petrović a écrit :> Thanks for your help, JB. > Yes, I would say that it is installed. The "tc" command does exist. > After all, everything works correctly with non-intensive workloads. > > Any other suggestions, anyone? > > On 11/16/2010 09:31 PM, Jean Baptiste FAVRE wrote: >> Hello, >> Seems that remus tries to use "tc" command. Think it''s for traffic >> shaping to take priority. >> Is the package installed on your dom0 ? >> >> Regards, >> JB >> >> Le 16/11/2010 16:45, Darko Petrović a écrit : >> >>> Hello everyone, >>> >>> I''ve managed to configure and start Remus. However, it works only while >>> the protected server is not loaded. As soon as I start a >>> memory-intensive server application, the stream of Remus messages stops, >>> leaving me with an error message. I think the message is not always the >>> same, but here is what I''ve got from the last run: >>> >>> PROF: suspending at 1289921328.328959 >>> PROF: resumed at 1289921330.903716 >>> xc: error: Error when flushing output buffer (32 = Broken pipe): >>> Internal error >>> tc filter del dev vif1.0 parent ffff: proto ip pref 10 u32 >>> RTNETLINK answers: Invalid argument >>> We have an error talking to the kernel >>> Exception xen.remus.util.PipeException: PipeException(''tc failed: 2, No >>> such file or directory'',) in<bound method BufferedNIC.__del__ of >>> <xen.remus.device.BufferedNIC object at 0x7f27844df210>> ignored >>> >>> I have tried to pin one physical core to Domain 0 and another one to the >>> protected domain, using vcpu-pin but it doesn''t help. >>> Currently I am running the Xen-unstable tree (last updated 3-4 days >>> ago), 2.6.32.25 pv-ops kernel as Dom-0 and 2.6.18.8 as Dom-U, but I had >>> a very similar problem with Xen 4.0.0 and 2.6.18.8 as Dom-0. >>> >>> Any suggestions? >>> >>> Thanks >>> Darko >>> >>> _______________________________________________ >>> Xen-users mailing list >>> Xen-users@lists.xensource.com >>> http://lists.xensource.com/xen-users >>> >>> >> >> _______________________________________________ >> Xen-users mailing list >> Xen-users@lists.xensource.com >> http://lists.xensource.com/xen-users >> > > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Here''s the catch: CPU scheduling. I''ve changed the number of Dom-0 VCPUs from 2 down to 1 (because anyway I have one spare physical core dedicated for Dom-0). Now it works like a charm... Darko On 11/17/2010 04:32 PM, Jean Baptiste FAVRE wrote:> Hello, > Yes, everything works great with non-intensive load. Also, maybe remus > does not try to do traffic shapin with non-intensive load :) > > Another suggestion, so :) > Does vif1.0 exists ? As remus tries to execute tc command on vif1.0 > interface, you may have an issue here. > > Regards, > JB > > Le 17/11/2010 14:45, Darko Petrović a écrit : > >> Thanks for your help, JB. >> Yes, I would say that it is installed. The "tc" command does exist. >> After all, everything works correctly with non-intensive workloads. >> >> Any other suggestions, anyone? >> >> On 11/16/2010 09:31 PM, Jean Baptiste FAVRE wrote: >> >>> Hello, >>> Seems that remus tries to use "tc" command. Think it''s for traffic >>> shaping to take priority. >>> Is the package installed on your dom0 ? >>> >>> Regards, >>> JB >>> >>> Le 16/11/2010 16:45, Darko Petrović a écrit : >>> >>> >>>> Hello everyone, >>>> >>>> I''ve managed to configure and start Remus. However, it works only while >>>> the protected server is not loaded. As soon as I start a >>>> memory-intensive server application, the stream of Remus messages stops, >>>> leaving me with an error message. I think the message is not always the >>>> same, but here is what I''ve got from the last run: >>>> >>>> PROF: suspending at 1289921328.328959 >>>> PROF: resumed at 1289921330.903716 >>>> xc: error: Error when flushing output buffer (32 = Broken pipe): >>>> Internal error >>>> tc filter del dev vif1.0 parent ffff: proto ip pref 10 u32 >>>> RTNETLINK answers: Invalid argument >>>> We have an error talking to the kernel >>>> Exception xen.remus.util.PipeException: PipeException(''tc failed: 2, No >>>> such file or directory'',) in<bound method BufferedNIC.__del__ of >>>> <xen.remus.device.BufferedNIC object at 0x7f27844df210>> ignored >>>> >>>> I have tried to pin one physical core to Domain 0 and another one to the >>>> protected domain, using vcpu-pin but it doesn''t help. >>>> Currently I am running the Xen-unstable tree (last updated 3-4 days >>>> ago), 2.6.32.25 pv-ops kernel as Dom-0 and 2.6.18.8 as Dom-U, but I had >>>> a very similar problem with Xen 4.0.0 and 2.6.18.8 as Dom-0. >>>> >>>> Any suggestions? >>>> >>>> Thanks >>>> Darko >>>> >>>> _______________________________________________ >>>> Xen-users mailing list >>>> Xen-users@lists.xensource.com >>>> http://lists.xensource.com/xen-users >>>> >>>> >>>> >>> _______________________________________________ >>> Xen-users mailing list >>> Xen-users@lists.xensource.com >>> http://lists.xensource.com/xen-users >>> >>> >> >> _______________________________________________ >> Xen-users mailing list >> Xen-users@lists.xensource.com >> http://lists.xensource.com/xen-users >> >> > > _______________________________________________ > Xen-users mailing list > Xen-users@lists.xensource.com > http://lists.xensource.com/xen-users >_______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users