Menu
▾
▴
Re: [OpenNode-users] KVM Missing timer interrupts?
|
From: <ope...@li...> - 2010-06-03 20:05:58
|
Hello, I've already rebooted with the new kernel, but i'm afraid the problem persists... The guest machine, seems to hang (I have the clock and the task manager open), and as soon as I generate network traffic, it speeds up until "apparent time" reaches "real time"...kind of weird. Test scenario: Host: Centos 5.4 ovzkernel-2.6.18-194.3.1.el5.028stab069.6 kmod-kvm-83-164.asys.8 kvm-83-164.asys.8 Guest: Windows XP SP3 guest, up to date. Network driver: e1000. 1 vCPU 1024Mb RAM Procedure: 1) Start the guest 2) Connect via RDP 3) Open the clock applet and the task manager 4) Wait a couple of minutes, until there is some time drift 5) From the HN, ping -f ip.of.guest and see the clock speed up _fast_ until it reaches the current time. As I said, it seems as if the guest wasn't receiving timer interrupts at the rate needed (or at all). It's very noticable when it's idle and has no network traffic (only RDP traffic is forwarded to the guest). Hope that helps. Thank you. Julian J. M. On Thu, Jun 3, 2010 at 13:05, <ope...@li...> wrote: > Im hoping that after reboot when kvm and kernel versions match your > problem will go away perhaps - please let us know afterwards. > > ---------------------------------------------- > Andres Toomsalu, an...@ac... > > > > > ope...@li... wrote: >> I mean that (because I first installed manually the kvm and kmod-kvm >> packages) i'm using kmod-kvm-83-164.asys.8, with ovhkernel ending in >> .5, (instead of .6). I know it was a mistake ;), more info below. >> >> While I'm waiting for reboot, I downloaded the XP VM to a local >> vanilla Centos5.4 server, and it doesn't show the slow clock problem. >> I'm hoping that the kernel upgrade resolves the issue. >> >> >> >> # uname -r >> 2.6.18-194.3.1.el5.028stab069.5 >> >> Latest ovz kernel installed is >> ovzkernel-2.6.18-194.3.1.el5.028stab069.6 (but need to reboot to use >> it) >> >> # rpm -ql kmod-kvm >> /lib/modules/2.6.18-194.3.1.el5.028stab069.6/extra/kmod-kvm >> /lib/modules/2.6.18-194.3.1.el5.028stab069.6/extra/kmod-kvm/ksm.ko >> /lib/modules/2.6.18-194.3.1.el5.028stab069.6/extra/kmod-kvm/kvm-amd.ko >> /lib/modules/2.6.18-194.3.1.el5.028stab069.6/extra/kmod-kvm/kvm-intel.ko >> /lib/modules/2.6.18-194.3.1.el5.028stab069.6/extra/kmod-kvm/kvm.ko >> >> # modprobe -l | grep kvm >> /lib/modules/2.6.18-194.3.1.el5.028stab069.5/weak-updates/kmod-kvm/ksm.ko >> /lib/modules/2.6.18-194.3.1.el5.028stab069.5/weak-updates/kmod-kvm/kvm.ko >> /lib/modules/2.6.18-194.3.1.el5.028stab069.5/weak-updates/kmod-kvm/kvm-amd.ko >> /lib/modules/2.6.18-194.3.1.el5.028stab069.5/weak-updates/kmod-kvm/kvm-intel.ko >> >> >> Thanks for your help, >> Julián. >> >> >> On Thu, Jun 3, 2010 at 12:43, <ope...@li...> wrote: >> >>> What do you mean by: >>> "kvm is up to date, although not build specifically for the running >>> kernel" ? >>> >>> You must run correct version of opennode ovzkernel with specially build >>> kvm module for it - otherwise problems are quaranteed (we build kvm >>> module specially for every ovzkernel release). >>> So please enable opennode-test and do "yum update opennode" - as >>> opennode rpm has pre-set kernel and kvm module dependencys - so it >>> ensures that matching ovzkernel and opennode kvm module are both >>> installed. We provide ovzkernel through opennode repos - not loading it >>> directly from official openvz repo - this is intentional for delaying >>> kernel upgrades if we have not yet build kvm module for a new openvz kernel. >>> >>> Cheers, >>> >>> ---------------------------------------------- >>> Andres Toomsalu, an...@ac... >>> >>> >>> >>> >>> ope...@li... wrote: >>> >>>> Hello Andres, >>>> >>>> I already use opennode-test, but it's true that I have to reboot to >>>> get latest ovzkernel (kvm is up to date, although not build >>>> specifically for the running kernel). I have 3 openvz ve's in >>>> production, so I'll have to wait until tonight to reboot the server. >>>> >>>> Thanks, >>>> Julián J. M. >>>> >>>> On Thu, Jun 3, 2010 at 11:38, <ope...@li...> wrote: >>>> >>>> >>>>> Hi Juilán, >>>>> >>>>> Please disable in /etc/yum.repos.d/opennode opennode repo and enable >>>>> opennode-test repo - then do yum update opennode and let it update to >>>>> newer ovzkernel and kvm module (and reboot system afterwords to load new >>>>> kernel) - see if it fixes your problems. But keep in mind that there can >>>>> be other problems with opennode-test repo package versions - so avoid it >>>>> doing on production servers. >>>>> >>>>> All the best, >>>>> >>>>> ---------------------------------------------- >>>>> Andres Toomsalu, an...@ac... >>>>> >>>>> >>>>> >>>>> >>>>> ope...@li... wrote: >>>>> >>>>> >>>>>> Hello, >>>>>> >>>>>> I have a working OpenVZ installation, and recently installed the kvm >>>>>> packages from opennode. I managed to install 2 VMs, one with WinXP and >>>>>> another with Win2003, both 32bits. >>>>>> >>>>>> The problem is that the guests slows down when there is no network >>>>>> traffic, and recovers when I, for example, ping flood the guest IP >>>>>> address. >>>>>> >>>>>> I don't know the cause, but it would seem that the guest is not >>>>>> getting timer interrupts, and can only keep up when there is another >>>>>> source of interrupts (network traffic). >>>>>> >>>>>> Hardware Node: >>>>>> >>>>>> # cat /proc/cpuinfo >>>>>> processor : 0 >>>>>> vendor_id : GenuineIntel >>>>>> cpu family : 6 >>>>>> model : 23 >>>>>> model name : Intel(R) Xeon(R) CPU E5405 @ 2.00GHz >>>>>> stepping : 10 >>>>>> cpu MHz : 2000.028 >>>>>> cache size : 6144 KB >>>>>> physical id : 0 >>>>>> siblings : 4 >>>>>> core id : 0 >>>>>> cpu cores : 4 >>>>>> apicid : 0 >>>>>> fpu : yes >>>>>> fpu_exception : yes >>>>>> cpuid level : 13 >>>>>> wp : yes >>>>>> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov >>>>>> pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall nx lm >>>>>> constant_tsc pni monitor ds_cpl vmx tm2 ssse3 cx16 xtpr sse4_1 lahf_lm >>>>>> bogomips : 4000.05 >>>>>> clflush size : 64 >>>>>> cache_alignment : 64 >>>>>> address sizes : 38 bits physical, 48 bits virtual >>>>>> power management: >>>>>> >>>>>> (There are 8 cores in total.) >>>>>> >>>>>> #uname -a >>>>>> Linux ns21xxxx.ovh.net 2.6.18-194.3.1.el5.028stab069.5 #1 SMP Tue May >>>>>> 18 17:48:31 MSD 2010 x86_64 x86_64 x86_64 GNU/Linux >>>>>> >>>>>> # rpm -qa | grep kvm >>>>>> etherboot-zroms-kvm-5.4.4-13.el5.centos >>>>>> kmod-kvm-83-164.asys.8 >>>>>> kvm-83-164.asys.8 >>>>>> kvm-qemu-img-83-164.asys.8 >>>>>> kvm-tools-83-105.asys.22.4 >>>>>> >>>>>> Guest xml config: >>>>>> <domain type='kvm'> >>>>>> <name>WinXP</name> >>>>>> <uuid>7f119fd1-bbf3-64f6-17d5-e5263b6a1502</uuid> >>>>>> <memory>1048576</memory> >>>>>> <currentMemory>1048576</currentMemory> >>>>>> <vcpu>1</vcpu> >>>>>> <os> >>>>>> <type arch='x86_64' machine='pc'>hvm</type> >>>>>> <boot dev='hd'/> >>>>>> </os> >>>>>> <features> >>>>>> <acpi/> >>>>>> <pae/> >>>>>> </features> >>>>>> <clock offset='localtime'/> >>>>>> <on_poweroff>destroy</on_poweroff> >>>>>> <on_reboot>restart</on_reboot> >>>>>> <on_crash>restart</on_crash> >>>>>> <devices> >>>>>> <emulator>/usr/libexec/qemu-kvm</emulator> >>>>>> [...] The rest is trivial, ide HD (raw image), ide cdrom, and e1000 network. >>>>>> >>>>>> I initially installed XP with 2 vCPU. Changed it to 1 after reading >>>>>> some posts regarding clock drifts... Didn't fix it, anyway. >>>>>> >>>>>> >>>>>> This is the cmd line: >>>>>> /usr/libexec/qemu-kvm -S -M pc -m 1024 -smp 1 -name WinXP -uuid >>>>>> 7f119fd1-bbf3-64f6-17d5-e5263b6a1502 -no-kvm-pit-reinjection -monitor >>>>>> pty -pidfile /var/run/libvirt/qemu//WinXP.pid -localtime -boot c >>>>>> -drive file=/home/KVM/WinXPVM.img,if=ide,index=0,boot=on,format=raw >>>>>> -drive file=/home/KVM/NetKVM-and-viostor.iso,if=ide,media=cdrom,index=2,format=raw >>>>>> -net nic,macaddr=52:54:00:6a:d4:af,vlan=0,model=e1000 -net >>>>>> tap,fd=15,script=,vlan=0,ifname=vnet0 -serial pty -parallel none -usb >>>>>> -vnc 127.0.0.1:0 >>>>>> >>>>>> >>>>>> >>>>>> Any idea on why it is happening? >>>>>> >>>>>> Thanks in advance, >>>>>> Juilán J. M. >>>>>> >>>>>> >>>>>> >>>>>> >>>>> ------------------------------------------------------------------------------ >>>>> ThinkGeek and WIRED's GeekDad team up for the Ultimate >>>>> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the >>>>> lucky parental unit. See the prize list and enter to win: >>>>> http://p.sf.net/sfu/thinkgeek-promo >>>>> _______________________________________________ >>>>> OpenNode-users mailing list >>>>> Ope...@li... >>>>> https://lists.sourceforge.net/lists/listinfo/opennode-users >>>>> >>>>> >>>>> >>>> >>>> >>>> >>> ------------------------------------------------------------------------------ >>> ThinkGeek and WIRED's GeekDad team up for the Ultimate >>> GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the >>> lucky parental unit. See the prize list and enter to win: >>> http://p.sf.net/sfu/thinkgeek-promo >>> _______________________________________________ >>> OpenNode-users mailing list >>> Ope...@li... >>> https://lists.sourceforge.net/lists/listinfo/opennode-users >>> >>> >> >> >> >> > > ------------------------------------------------------------------------------ > ThinkGeek and WIRED's GeekDad team up for the Ultimate > GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the > lucky parental unit. See the prize list and enter to win: > http://p.sf.net/sfu/thinkgeek-promo > _______________________________________________ > OpenNode-users mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/opennode-users > -- http://www.julianmenendez.es |
