+ Reply to Thread
Results 1 to 8 of 8

Thread: computer node install: unable to get NII

  1. #1
    deian is offline Junior Member
    Join Date
    September 23rd, 2009
    Posts
    1
    Downloads
    0
    Uploads
    0

    Default computer node install: unable to get NII

    Hi,
    I installed kusu+centos 5.3. The master/isntaller node is working, but I'm trying to boot a compute node with PXE. The boot starts fine, but it fails with:

    udhcp SIOCGIFINDEX failed!: No such device

    URL: https://172.20.0.1/repos/nodeboot.cg...nstalled&boot=
    ERROR: Unable to get NII from: 172.20.0.1
    Will retry in 30 seconds.


    Any idea/more detailed posts on getting a compute node up.

    TIA

  2. #2
    eriou is offline Junior Member
    Join Date
    September 28th, 2009
    Posts
    1
    Downloads
    0
    Uploads
    0

    Default NII issue

    I am having the exact same problem. My two imaged client installed and run fine but when i try to add a diskless client it boots off PXE fine but gets stuck trying to download the NII. I'm running CentOS 5.3 as well. Though the url of the NII on the diskless client is my public interfaces ip (192.168.1.2) not my clusters interface (10.0.0.1).

    Any help would be appreciated as well.

  3. #3
    osobox is offline Junior Member
    Join Date
    December 14th, 2009
    Posts
    9
    Downloads
    6
    Uploads
    0

    Default ...unable to get NII

    Same here. What sort of animal is NII? Is it something we still have to configure or should this be automatically provisioned?
    Please help us debug this...! Thank you in advance

  4. #4
    dryrye is offline Junior Member
    Join Date
    July 11th, 2009
    Posts
    3
    Downloads
    0
    Uploads
    0

    Default

    check /var/log/messages on the master node for dhcpd leasing errors

    if you see any dhcpd errors:
    0. shut down slave node
    1. quit addhost
    2. restart dhcpd service
    3. start addhost
    4. boot slave node

    (if there are dhcpd-leasing related errors and you want something more permanent, google the error message, look for suggestions which require altering /etc/dhcpd.conf. I didn't go that far, I just restart dhcpd, which accounts for 90% of the installation problems)

  5. #5
    osobox is offline Junior Member
    Join Date
    December 14th, 2009
    Posts
    9
    Downloads
    6
    Uploads
    0

    Default

    dryrye,
    Thank you for your hints. I can't see any dhcpd errors. When trying wget from another station in the network I get:

    $ wget -O - "http://172.20.0.1/repos/nodeboot.cgi?dump=1&state=Installed&boot="
    ... snipped wget stuff ...
    <nii>
    <debug>
    Dump NII: 1
    State: Installed
    Dump CFM: 0
    Cannot find host for IP: 172.20.0.254
    IP: 172.20.0.254
    Node:
    </debug>
    </nii>

    However, boothost -l for this client's MAC gives another IP (172.20.0.3). Who is 172.20.0.254?

    Thank you in advance

  6. #6
    osobox is offline Junior Member
    Join Date
    December 14th, 2009
    Posts
    9
    Downloads
    6
    Uploads
    0

    Default

    AND, when requesting same URL from within 172.20.0.1 (kusu server) it goes along this way:

    [root@capfe ~]# wget -O - "http://172.20.0.1/repos/nodeboot.cgi?dump=1&state=Installed&boot="
    --2010-02-01 19:44:16-- http://172.20.0.1/repos/nodeboot.cgi...nstalled&boot=
    Connecting to 172.20.0.1:80... connected.
    HTTP request sent, awaiting response... 200 OK
    Length: unspecified [text/html]
    Saving to: `STDOUT'

    [<=> ] 0 --.-K/s <?xml version="1.0"?>
    <nii>
    <debug>
    Dump NII: 1
    State: Installed
    Dump CFM: 0
    IP: 172.20.0.1
    Node: capfe
    </debug>
    <nodeinfo name="capfe" installers="172.20.0.1" repo="/repos/1000" ostype="centos-5-i386" installtype="package" nodegrpid="1" ngtype="installer" repoid="1000" dbpasswd="jCb8GaN0" cfmsecret="b5iaLdmY1eJN5GctL2G69t66wkVCODAhU88XMlg poIA=">
    <nicinfo device="eth1" ip="10.0.2.152" subnet="255.255.255.0" network="10.0.2.0" suffix="-eth1" gateway="10.0.2.1" dhcp="0" options="" boot="1"></nicinfo>
    <nicinfo device="eth0" ip="172.20.0.1" subnet="255.255.0.0" network="172.20.0.0" suffix="-eth0" gateway="172.20.0.1" dhcp="0" options="" boot="1"></nicinfo>
    <partition device="" partition="" mntpnt="" fstype="" size="None" options="partitionID=*" preserve="0"></partition>
    <partition device="" partition="" mntpnt="" fstype="" size="None" options="partitionID=Dell Utility" preserve="1"></partition>
    <partition device="VAR" partition="" mntpnt="/var" fstype="ext3" size="2000" options="lv;vg=KusuVolGroup00" preserve="0"></partition>
    <partition device="DEPOT" partition="" mntpnt="/depot" fstype="ext3" size="10000" options="lv;vg=KusuVolGroup00" preserve="0"></partition>
    <partition device="ROOT" partition="" mntpnt="/" fstype="ext3" size="12000" options="lv;vg=KusuVolGroup00" preserve="0"></partition>
    <partition device="KusuVolGroup00" partition="" mntpnt="" fstype="" size="None" options="vg;extent=32M" preserve="0"></partition>
    <partition device="N" partition="0" mntpnt="" fstype="physical volume" size="28000" options="fill;pv;vg=KusuVolGroup00" preserve="0"></partition>
    <partition device="1" partition="2" mntpnt="" fstype="linux-swap" size="2000" options="" preserve="0"></partition>
    <partition device="1" partition="1" mntpnt="/boot" fstype="ext3" size="100" options="" preserve="0"></partition>
    <component>component-base-installer</component>
    <component>component-base-node</component>
    <component>component-gnome-desktop</component>
    <appglobals name="CFMBaseDir" value="/opt/kusu/cfm"></appglobals>
    <appglobals name="InstallerServeDNS" value="1"></appglobals>
    <appglobals name="InstallerServeNIS" value="0"></appglobals>
    <appglobals name="InstallerServeNTP" value="1"></appglobals>
    <appglobals name="InstallerServeNFS" value="1"></appglobals>
    <appglobals name="DEPOT_KITS_ROOT" value="/depot/kits"></appglobals>
    <appglobals name="DEPOT_IMAGES_ROOT" value="/depot/images"></appglobals>
    <appglobals name="DEPOT_REPOS_ROOT" value="/depot/repos"></appglobals>
    <appglobals name="DEPOT_REPOS_POST" value="/depot/repos/post_scripts"></appglobals>
    <appglobals name="DEPOT_CONTRIB_ROOT" value="/depot/contrib"></appglobals>
    <appglobals name="DEPOT_UPDATES_ROOT" value="/depot/updates"></appglobals>
    <appglobals name="DEPOT_AUTOINST_ROOT" value="/depot/repos/instconf"></appglobals>
    <appglobals name="PIXIE_ROOT" value="/tftpboot/kusu"></appglobals>
    <appglobals name="PROVISION" value="KUSU"></appglobals>
    <appglobals name="DNSZone" value="kusu"></appglobals>
    <appglobals name="PublicDNSZone" value="dcc.uncoma.edu.ar"></appglobals>
    <appglobals name="Language" value="en"></appglobals>
    <appglobals name="Keyboard" value="es"></appglobals>
    <appglobals name="PrimaryInstaller" value="capfe"></appglobals>
    <appglobals name="dns1" value="10.0.2.1"></appglobals>
    <appglobals name="Timezone_zone" value="America/Argentina/Buenos_Aires"></appglobals>
    <appglobals name="Timezone_utc" value="0"></appglobals>
    <appglobals name="Timezone_ntp_server" value="pool.ntp.org"></appglobals>
    <appglobals name="DEPOT_REPOS_SCRIPTS" value="/depot/repos/custom_scripts"></appglobals>
    </nodeinfo>
    </nii>
    [ <=> ] 3,743 --.-K/s in 0s

    2010-02-01 19:44:17 (50.3 MB/s) - `-' saved [3743]

    Please any hints?

  7. #7
    osobox is offline Junior Member
    Join Date
    December 14th, 2009
    Posts
    9
    Downloads
    6
    Uploads
    0

    Default

    Quote Originally Posted by deian View Post
    Hi,
    udhcp SIOCGIFINDEX failed!: No such device
    Now that I come to think of it, what network drivers are provided in kusu 1.1 diskless kernels? I can't seem to find a HCL.
    TIA

  8. #8
    osobox is offline Junior Member
    Join Date
    December 14th, 2009
    Posts
    9
    Downloads
    6
    Uploads
    0

    Default Solved!

    Just FYI

    Because of lack of hardware, I was trying to create a completely virtual cluster over KVM on a Quad computer.
    Seems to be, qemu-kvm emulates only a couple of network boards, none of them supported by Kusu 1.1 initrd.
    I opened up initrd, added the modules, packed it down again, problem solved.
    So, I think a good addition to the knowledge base could be:

    IF client fails boot with
    udhcp SIOCGIFINDEX failed!: No such device

    THEN check whether your network card is supported. Supported NICs in Kusu X.X are...

    Thank you again for your help!

+ Reply to Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts