Hide Forgot
Description of problem: I have 6 virtual instances of different Redhat OSes on a Sun X2200 with KVM and QEMU. Until the last updates everything was really nice. Now The virtual machines are hanging for seconds several times every minute. I'm on the virtual machines via ssh and every now and then (let's say 2 - 4 times a minute) it hangs for 1 .. 5 seconds. pinging the ip address of the virtual machines shows no replies during these hangs either, so it seems, the entire instance is hanging. When stracing the qemu-kvm process it's busy all the time, so it's not the underlaying OS hanging. Furthermore it does not happen with all 6 virtual machines at the same time. The host system is a Redhat-6 with kernel-2.6.32-358.18.1.el6.x86_64 qemu-kvm-0.12.1.2-2.415.el6_5.3.x86_64 The virtual instances are: CentOS-6.4 CentOS-5.9 Fedora-17 latest available patch level CentOS-4.8 Redhat-EL 5.6 (2 instances) All these are basiclly idling, in local top programs i see 99% idle or more. On the host system i see typically: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 6334 qemu 20 0 16.6g 10g 5552 S 112.6 17.2 23:49.99 qemu-kvm 6869 qemu 20 0 16.6g 1.3g 5096 S 99.7 2.1 21:24.13 qemu-kvm 6595 qemu 20 0 16.6g 4.1g 5552 S 37.2 6.5 21:07.94 qemu-kvm 5661 qemu 20 0 39.9g 1.6g 5356 R 34.5 2.5 15:09.04 qemu-kvm 5264 qemu 20 0 30.1g 1.3g 5356 R 20.3 2.1 9:11.64 qemu-kvm 5987 qemu 20 0 10.6g 1.4g 5840 S 1.3 2.3 1:10.55 qemu-kvm though none of them is actually doing anything. The lowest one is the Fedora instance. It seems to idle "most efficiently". Version-Release number of selected component (if applicable): kernel-2.6.32-358.18.1.el6.x86_64 qemu-kvm-0.12.1.2-2.415.el6_5.3.x86_64 How reproducible: Configure KVM, with official network interfaces, i.e. no NAT, no local DHCP and DNS, no spanning tree, just a virtual bridge connecting the eth0 with the virtual interfaces. Run several virtual instances and use them Steps to Reproduce: 1. See above 2. Login to one of the virtual instances 3. Do something and may ping the instance at the same time from another host 4. Probably watch the CPU utilization on the hosting system Actual results: Connection or the entire virtual instance hangs for several seconds (typically 1 .. 5) Expected results: Nearly constant responsiveness. Ok it may vary due to real load on the host system or on the virtual instance, but not that hard. Without any load i'd want to expect nearly constant responsiveness. Additional info: It did not happen with kernel-2.6.32-279.19.1.el6.x86_64 qemu-kvm-0.12.1.2-2.209.el6_2.4
Additional Info: On a different host system (Also RHEL6 in same patch level) i have other virtual instances running and there it's just the same. It was all fine some months ago but now i see these pauses.
In the meantime i played a bit with that and found: Reducing the number of CPUs of the virtual CentOS-4 instance to 2 changed the situation significantly. Overview: Host system: Sun X2200 with 4 CPU cores 6 Virtual systems: # OS #CPUs 1 CentOS-6.4 2 1 CentOS-5.9 2 1 Fedora-17 2 2 RHEL-5.6 4 1 CentOS-4.8 2 (now, had 4 before) Now when none of the virtual hosts has notable load (what is true most of the time), any single node is very responsive. If any of the systems has a bit more load, e.g. one virtual CPU is 100% busy, the situation is like before: All of them are really slow and hardly usable. Unlike a year ago, there seems to be a quite sharp line in the total load. If the load is higher than this limit, all virtual systems show very bad responsiveness. I'd be interested in options to configure the scheduler that is managing the qemu processes, if there are any.
Albert, thanks for taking the time to enter a bug report with us. We appreciate the feedback and look to use reports such as this to guide our efforts at improving our products. That being said, we're not able to guarantee the timeliness or suitability of a resolution for issues entered here because this is not a mechanism for requesting support. If this issue is critical or in any way time sensitive, please raise a ticket through your regular Red Hat support channels to make certain it receives the proper attention and prioritization to assure a timely resolution. For information on how to contact the Red Hat production support team, please visit: https://www.redhat.com/support/process/production/#howto
Albert, are you still experiencing these hangs? Did old kernel work well with newer qemu-kvm? (kernel-2.6.32-279.19.1.el6.x86_64 and qemu-kvm-0.12.1.2-2.415.el6_5.3.x86_64) Thank you.
(In reply to Radim Krčmář from comment #7) > Albert, are you still experiencing these hangs? > > Did old kernel work well with newer qemu-kvm? > (kernel-2.6.32-279.19.1.el6.x86_64 and > qemu-kvm-0.12.1.2-2.415.el6_5.3.x86_64) Setting needinfo.
Thanks for the inquiry ! Unfortunately i have currently no possibility to test this. So please feel free to close this bug for now. As soon as i have the chance to check that again, i can reopen, if necessary. Is that ok for you ?
(In reply to Albert Flügel from comment #9) > Thanks for the inquiry ! Unfortunately > i have currently no possibility to test this. > So please feel free to close this bug for now. > As soon as i have the chance to check that again, i can reopen, if necessary. > Is that ok for you ? OK, thanks.