Originally posted by: PontiacGeronimo
I met similiar issue and did the following trick to have service IP as the desired source IP:
Created following script on each cluster node (note: PERSISTENT variable in the script contains different persistent IP(s) on each node)
PERSISTENT=
"XXX.XXX.XXX.XXX"
for i in $PERSISTENT
do
if [ `netstat -in|grep -c -w $i` -eq 1 ] then EN=$(netstat -in|awk -va=
"$i"
'$4 == a {print $1}') MK=$(lsattr -El $EN -a netmask -F value)
if [ `netstat -in|grep -w -c ^$
{EN
}` -gt 3 -a x`netstat -in|grep -w ^$
{EN
}|sed
'1,2d'|tail -1|awk
'{print $4}'` = x$i ] then echo persistent $i being moved to be not the first alias IP on $EN ifconfig $EN delete $i ifconfig $EN alias $i netmask $MK up echo persistent $i moved to be not the first alias IP on $EN fi fi done
In "smitty cm_extended_event_config_menu_dmn" configured "Pre/Post-Event Command" using the mentioned script.
In the same HACMP menu, in "Change/Show Pre-Defined HACMP Events" I have added the configured "Pre/Post-Event Command" to be executed as "Post-event Command" for following cluster events:
acquire_service_addr
acquire_takeover_addr
swap_adapter_complete
So, everytime those events occures (eg. during cluster startup), the persistent IP alias is unconfigured and configured again and service IP becames source IP for the outgoing connections.
The other solution would be you request the same firewall rules for persistent IPs you have now for service ones.