HACMP: source address problem

[ogniemi]

Hello,

I have a cascading cluster built of 2 LPARs each one having 2 service addresses one for internal communication and second one for customer access.
So, 2 of service addresses are in network 10.1.1.0/24 and 2 in network 10.1.2.0/24.


Node1:
10.1.1.7
10.1.2.7
default gateway: 10.1.1.2


Node2:
10.1.1.17
10.1.2.17
default gateway: 10.1.1.2



I need to keep source address 10.1.1.7 after takeover services of node1 to node2, but source is 10.1.1.17 when both application servers are on node2.

Status of node2 after takeover (node1 crashed)

# netstat -in
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
en0 1500 link#2 0.9.6c.44.11.8 216269 0 17525 0 0
en0 1500 10.1.2 10.1.2.17 216269 0 17525 0 0
en1 1500 link#3 0.9.6c.44.11.9 566097 0 1223549 0 0
en1 1500 10.1.1 10.1.1.17 566097 0 1223549 0 0
en4 1500 link#4 0.9.6c.44.16.10 201255 0 35664 0 0
en4 1500 10.1.2 10.1.2.7 201255 0 35664 0 0
en5 1500 link#5 0.9.6c.44.16.11 200458 0 37938 0 0
en5 1500 10.1.1 10.1.1.7 200458 0 37938 0 0
en7 1500 link#6 0.9.6c.44.16.13 26648 0 24048 0 0
en7 1500 10.0 10.2.1.17 26648 0 24048 0 0
lo0 16896 link#1 2580909 0 2583174 0 0
lo0 16896 127 loopback 2580909 0 2583174 0 0
lo0 16896 ::1 2580909 0 2583174 0 0



# netstat -rn
Routing tables
Destination Gateway Flags Refs Use If PMTU Exp Groups

Route Tree for Protocol Family 2 (Internet):
default 10.1.1.2 UGc 0 73 en1 - -
.........................................
..........CUT HERE.......................
.........................................



Has anyone idea how to handle with this problem? I just need after takeover the following (source should be address of en5 instead of en1):

# netstat -rn
Routing tables
Destination Gateway Flags Refs Use If PMTU Exp Groups

Route Tree for Protocol Family 2 (Internet):
default 10.1.1.2 UGc 0 73 en5 - -
.........................................


I don't expect slutions with static routes.



best regards, m.

 

[DSMARWAY]

hi ,

Don't fully understand what you require ( its probably me)
can you give a listing of cllsif ?

 

[ogniemi]

Hi,

this is not a cluster configuration problem. Cluster works perfectly.

Some of remote hosts permit access only from 10.1.1.7. Having after takeover:

this:

en1: 10.1.1.17
en5: 10.1.1.7

and this:

default 10.1.1.2 UGc 0 73 en1 - -

connection to remotes will always fail because source address will be 10.1.1.17 (and should be 10.1.1.7).

 

[ogniemi]

please tell me if it will work (just replace en1 with en5 - both are in the same network:

route change default -interface en5

????

should I flush routing table before with "route -f" ???

 

[ogniemi]

SOLVED!!!!!!!!!!!!!


route change 0 -if en5


regards,, m.