KVM High LoadAverage Understanding



  • Hello,

    I installed pfsense on a kvm/ubuntu_server_14.04 machine, with intel celeron 2 core and 4GB RAM.
    I have allocated 2 CPU and 2GB RAM to the 'virtual pfsense firewall'.
    I also checked this 2 box on pfsense web_gui :

    Hardware TCP Segmentation Offloading
    Hardware Large Receive Offloading
    

    And on the kvm/ubuntu_server_14.04 machine this from command line :

    
    #That's is 4 find the virtual interface used by pfsense
    for vm in $(virsh list | grep running | awk '{print $2}'); do echo -n "$vm:"; virsh dumpxml $vm| grep -oP "vnet\d+" ; done
    #That's the Output
    PfSenseFireWall:vnet1
    vnet2
    #That's the command used 4 turn off tx offloading 
    sudo ethtool -K vnet1 tx off
    sudo ethtool -K vnet2 tx off
    

    I add this in rc.local :

    # Turn tx OffLoading @ Boot
    sudo ethtool -K vnet1 tx off
    sudo ethtool -K vnet2 tx off
    

    I've installed this package on pfsense :

    freeradius2
    iperf
    ntopng
    

    I set 3 interface :

    em0 wan
    em1 lan
    ovpnc1 vpn_client
    

    On LAN I've enabled a Captive Portal with a custom page (login..error..ecc..) and I autenticate users using freeradius.
    No shaper is enabled . In the future I'll add some other wan with VLAN on the actual wan interface.
    But now I need to understand this command line 's output :

    top -aSH
     PID USERNAME   PRI NICE   SIZE    RES STATE   C   TIME    WCPU COMMAND
       11 root       155 ki31     0K    32K CPU0    0  59:24  97.75% [idle{idle: cp
       11 root       155 ki31     0K    32K RUN     1  59:45  96.97% [idle{idle: cp
    51557 root        20    0   158M 37280K nanslp  1   0:46   2.88% /usr/local/bin
        4 root       -16    -     0K    32K -       0   0:42   0.98% [cam{doneq0}]
       12 root       -60    -     0K   288K WAIT    1   0:21   0.20% [intr{swi4: cl
       12 root       -88    -     0K   288K WAIT    1   0:21   0.20% [intr{irq15: a
        0 root       -92    -     0K   192K -       0   0:07   0.10% [kernel{dummyn
       20 root        16    -     0K    16K syncer  1   0:04   0.10% [syncer]
        0 root       -16    -     0K   192K swapin  1   0:38   0.00% [kernel{swappe
       12 root       -88    -     0K   288K WAIT    1   0:14   0.00% [intr{irq14: a
       12 root       -56    -     0K   288K WAIT    1   0:12   0.00% [intr{swi5: fa
    50894 root        20    0 24028K  4268K kqread  1   0:07   0.00% redis-server:
      310 root        42    0   278M 54536K select  0   0:04   0.00% /usr/local/bin
    51557 root        20    0   158M 37280K nanslp  1   0:04   0.00% /usr/local/bin
       12 root       -92    -     0K   288K WAIT    1   0:02   0.00% [intr{irq11: r
       14 root        -8    -     0K    48K -       1   0:02   0.00% [geom{g_down}]
       18 root        20    -     0K    32K sdflus  1   0:02   0.00% [bufdaemon{/ w[/code]
    [code]5:09PM  up  1:31, 1 user, load averages: 1.33, 1.43, 1.49[/code]
    Who's idle ? Is this process using 97.75% of CPU ?
    [code]   11 root       155 ki31     0K    32K CPU0    0  59:24  97.75% [idle{idle: cp
       11 root       155 ki31     0K    32K RUN     1  59:45  96.97% [idle{idle: cp[/code]
    There is a way to reduce the load average ?
    [code]1 user, load averages: 1.33, 1.43, 1.49[/code]
    I try to install kvm using virt type = qemu and in a second test using virt type = kvm
    But with the same results .
    
    Can someone write what are the performance using more wan tagged with VLAN using switch and interface 1000GB and if I can have problem using kvm ?
    
    Thx in advance 4 any reply.
    Best regards
    Christian Arminio 
    
    [/code]
    

Log in to reply