hostsystem-config

HPC and Data for Lattice QCD

hostsystem-config

Changes to the host and rack database

The configuration for both APEmille and apeNEXT 'master' and 'unit' PCs is stored in hostsystem.xml under /afs/ifh.de/group/ape/etc/SRC
make install.apeconf
will update data in /afs/ifh.de/group/ape/apeboot/, from where then the masters can update all information that are needed for booting the unit-pcs (apehost.conf, hosts, dhcpd.conf, pxe configuration,, ...) via
amsync -a apeboot.conf -n
(-n is the dryrun mode, type -h for all comandline options). The /etc/hosts file on the apemasters is under control of the cenral IT staff. Our addon for the private network has to be copied to /afs/.ifh.de/common/installation/test/archive/hosts/AddOn/apemaster/etc and then gets automatically included. Currently it is only writable by people in group:ape_adm. Although the xml file gives many possibilities how to configure the network, we should try to stick to the following rules (at least for apenext).
apemaster:

use apemaster{n} for external network (141...)
master{n}[extension] for internal network
192.168.0.{200+n} administrative network (for example booting, etc)
192.168.{i}.{200+n}, where i is either 1,2 or 3 (see below)

blade pcs:

use an{k}{j} as a position name, reflecting which rack it is serving
host{n}[extension] is fixed to a certain hardware (not the position)
192.168.0.{n} extension '-a' for extra NIC (usually eth0) via which the host gets booted
192.168.1.{n} first gigbit onboard (usually eth1)
192.168.2.{n} second gigbit onboard (usually eth2)
192.168.3.{n} bonding of onboard gigbit (usually bond0)

In the following example master4 will be the bootserver of host101, which has three interfaces, where eth0 ist used for booting, bond0 is a bonding of eth1 and eth2. master5 will serve the nroot, apeshare and so on.


<apeConfiguration>
    <rackConfiguration>
        <rack>
            <type>apeNEXT</type>
            <number>3</number>

            <serverName>master5</serverName>
            <bootServerName>master4</bootServerName>
            <bootFile>pxelinux.0</bootFile>
            <nfsRoot>/apeboot/sl3.0.5_ro_v0.6.3</nfsRoot>

            <hostList>
                <host>
                    <name>host101</name>
                    <positionName>an30</positionName>
                </host>

            </hostList>
        </rack>
        </rackConfiguration>
        <hostConfiguration>
        <host>
            <name>host101</name>

            <interfaceList>
                <elem>
                    <nameExtension>-a</nameExtension>
                    <interface>eth0</interface>
                    <MAC>00:10:F3:08:F0:D7</MAC>

                    <IP>192.168.0.101</IP>
                    <netboot>yes</netboot>
                </elem>
                <elem>
                    <nameExtension>-g1</nameExtension>

                    <interface>eth1</interface>
                    <MAC>00:10:F3:08:F0:D5</MAC>
                    <IP>192.168.1.101</IP>
                </elem>

                <elem>
                    <nameExtension>-g2</nameExtension>
                    <interface>eth2</interface>
                    <MAC>00:10:F3:08:F0:D6</MAC>

                    <IP>192.168.2.101</IP>
                </elem>
                <elem>
                    <enable>yes</enable>
                    <interface>bond0</interface>

                    <IP>192.168.3.101</IP>
                    <netboot>yes</netboot>
                    <MACofInterface>eth1</MACofInterface>
                </elem>

            </interfaceList>
        </host>
    </hostConfiguration>
</apeConfiguration>