> Alex Balashov wrote:
> > Why would you want to do this versus sending a gratuitous ARP? > > Nagaraja, Madhukar (Madhukar) wrote: > > > Hello, > > Is there any script just like IPaddr for the MAC address takeover ?? > > If not, any other options. > > > > Please let me know. > > > > Thanks. Hi, I used a script I wrote to take macaddresses over. I did this because some devices dont accept the gratuituos arp. In fact, I take the MAC over and after that I take the virtual IP over too sending the gratuituous ARP to update the switch. I think that this works only if your network card supports mac address modification with "ip link set NETWORK_INTERFACE address DESIRED_MAC_ADDRESS" I use a script and a folder with four configuration files. For example, the name of my cluster is mycluster and there are two nodes, mycluster1 and mycluster2, then I have four files like these: |-- here-comes-the-folder-with-the-cluster-conf | |-- cluster.conf | `-- mac | |-- mycluster | |-- mycluster1 | `-- mycluster2 The content of each file is: :::::::::::::: cluster.conf :::::::::::::: CLUSTER="mycluster" :::::::::::::: mycluster :::::::::::::: eth0 01:01:01:12:34:56 :::::::::::::: mycluster1 :::::::::::::: eth0 00:14:35:7c:25:2f :::::::::::::: mycluster2 :::::::::::::: eth0 00:14:35:7c:26:b8 The cluster.conf file contains the cluster name. The files in the mac folder contain the virtual macs in the mycluster file, and the network cards real macs of each node in the mycluster1 and mycluster2 file. These files have the same name as the returned by `hostname` on each cluster node. In this example, the cluster uses the network interface eth0 with example IPs 10.0.0.1, 10.0.0.2 and 10.0.0.3. Take care too use unused mac address ranges for your virtual macs, I think the 01:01:01 vendor does not exist (http://www.coffer.com/mac_find/?string=01%3A01% 3A01), and then I set the second part of the virtual cluster mac as I like. Then I have a script that I use in haresources to take the macs over to the active node: :::::::::::::: haresources :::::::::::::: # /etc/ha.d/haresources mycluster1 cluster-mac-takeover 10.0.0.1/24 Finally I put the script cluster-mac-takeover in the init.d folder with the following contents: #!/bin/bash # Cluster MAC Takeover # # Variables CONFDIR="here-comes-the-folder-with-the-cluster-conf" DESC="Cluster MAC Takeover" HOSTNAME="`hostname`" LOCK="$DIR/var/tmp/cluster-mac-takeover" IP="/sbin/ip" CAT="/bin/cat" # # Configuration if [ -f $CONFDIR/cluster.conf ]; then source $CONFDIR/cluster.conf else echo "Cluster MAC Takeover Conf Not Found." exit 1 fi if [ ! -f $CONFDIR/mac/$CLUSTER ]; then echo "Cluster MAC Takeover Conf Not Found." exit 1 fi if [ ! -f $CONFDIR/mac/$HOSTNAME ]; then echo "Cluster MAC Takeover Conf Not Found." exit 1 fi # ## Compatibility COMPATIBILITY="yes" # # Funcions start() { $CAT $CONFDIR/mac/$CLUSTER | grep -v "^#" | grep -v "^$" | while read INTERFACE do IF="`echo $INTERFACE | awk '{ print $1 }'`" MAC="`echo $INTERFACE | awk '{ print $2 }'`" # Cluster Mac Address Takeover $IP link set $IF address $MAC done } stop() { $CAT $CONFDIR/mac/$HOSTNAME | grep -v "^#" | grep -v "^$" | while read INTERFACE do IF=`echo $INTERFACE | awk '{ print $1 }'` MAC=`echo $INTERFACE | awk '{ print $2 }'` # Host Mac Address Reset $IP link set $IF address $MAC done } # # Process case $1 in start) if [ -f $LOCK ]; then echo "$DESC is already started." else echo -n "Starting $DESC " start if [ "$?" -eq "0" ]; then touch $LOCK echo done. else echo failed. fi fi ;; stop) if [ ! -f $LOCK ]; then echo "$DESC is already stopped." else echo -n "Stopping $DESC " stop if [ "$?" -eq "0" ]; then rm -f $LOCK echo done. else echo failed. fi fi ;; status) if [ -f $LOCK ]; then if [ "$COMPATIBILITY" = "yes" ]; then echo "running" else echo "$DESC is started." fi else echo "$DESC is stopped." exit 3 fi ;; *) echo "Usage: $0 start|stop|status" exit 2 ;; esac exit 0 The COMPATIBILITY in the script flag is set so that the script behaves as expected by heartbeat (AFAIK) and the script returns also the expected exit codes. Maybe this is not be the best way to do this and maybe there exists already a builtin solution I dont know but I hope this helps, Best Regards, Eric Janz _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
