RaSca ha scritto:
Hi everybody,I'm running a Debian Etch two node ha cluster with the topology of streamline high availability and load balancing with ldirectord. I followed the ultramonkey suggested configuration as explained here : http://www.ultramonkey.org/3/topologies/sl-ha-lb-eg.html. For each cluster node there is a loopback interface named lo:0 with the same address of the virtual balanced service (but with 32bit netmask). This interface is configured with a pre-up script that sets the ARP advertisements restrictions with sysctl.
[...]
So what the problem may be ? Note that even if the errors are logged, the service still works and balance on the two cluster node.Thanks for your help,
Hi all,finally, after some time passed trying to understand what was wrong i solved my problem by creating a script which disable the lo:0 interface on the master node. What is not explained in the document at ultramonkey site is that the master node should not have a loopback interface with the same address of the Virtual IP, that is not logical. As I said, i create a script named /etc/init.d/ldirectord_lo-set with this content :
VIP=<my virtual ip>
host=`/bin/hostname`
case "$1" in
"stop")
/sbin/ifconfig lo down
/sbin/ifconfig lo up
echo 1 > /proc/sys/net/ipv4/conf/lo/arp_ignore
echo 2 > /proc/sys/net/ipv4/conf/lo/arp_announce
echo 1 > /proc/sys/net/ipv4/conf/all/arp_ignore
echo 2 > /proc/sys/net/ipv4/conf/all/arp_announce
/sbin/ifconfig lo:0 $VIP netmask 255.255.255.255 up
/sbin/route add -host $VIP dev lo:0
exit 0
;;
"start")
/sbin/ifconfig lo:0 down
echo 0 > /proc/sys/net/ipv4/conf/lo/arp_ignore
echo 0 > /proc/sys/net/ipv4/conf/lo/arp_announce
echo 0 > /proc/sys/net/ipv4/conf/all/arp_ignore
echo 0 > /proc/sys/net/ipv4/conf/all/arp_announce
exit 0
;;
"*")
echo $0: Usage: $0 {start|status|stop}
exit 1
;;
esac
And this is the new haresources :
node-1 LVSSyncDaemonSwap::master ldirectord::/etc/ha.d/ldirectord.cf
ldirectord_lo-set IPaddr2::<VIRTUALIP>/27/eth0
The master node is at the same time also a real server, for this reason it should not have a loopback interface. So, when the service is stopped on a node the loopback interface is activated (node is NOT the master) and when it is started the interface is be disabled (node IS the master).
Everything now works without warnings. -- RaSca "Il regno dei cieli e' simile ad un mercante che va in cerca di perle preziose. Trovatane una di grande valore va, vende tutti i suoi averi e la compra" - Gesu' Cristo http://web.tiscali.it/rascasoft
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
