I am new in AIX and please forgive my poor english.
I know that AIX allow same subnet IPs for different interfaces, which result in multipath routing / route striping.
My question is,
is there any best practice for the persistent and service IP with same subnet to stay on same interface, or it is okay to lie on different interfaces? What negative effect will occur for these same subnet IPs to stay on different interfaces?
Depends on how many physical network connections do you have, which hacmp version and do you want to use IP aliasing or not.
If you only have one adapter (or a pair of adapters setup with etherchannel,) and you are going to use aliasing and have HACMP 6.x than yes all your IP adresses (physical and service) need to be in the same subnet. Obviously this is the easiest way to go. If you want to configure takeover than you need two different networks / subnets - a routable one for service- and boot IP and a non-rootable for the standby adapter.
Regards
zxmaus
I use HACMP version 6.1 SP5
and use IPAT through IP alias for persistent and service IPs.
Let me briefly describe my environment:
en13: (5 active etherchannel with no backup adapter)
1.2.3.4 <-boot IP
192.168.1.1 <-persistent IP
en14: (5 active etherchannel with no backup adapter)
2.3.4.5 <-boot IP
192.168.1.10 <-service IP
en13 and en14 have same subnet IPs 192.168.1.1 and 192.168.1.10.
Is it allowed for HACMP?
I heard from someone that the best practice should allocate the service to stick with persistent, so that same subnet IPs will not appear in different interfaces. But I cant find the documentation for this.
---------- Post updated at 06:24 PM ---------- Previous update was at 10:04 AM ----------
anyone could help please~?
thank you very much!!!
The redbook says:
For IPAT via aliases,
All base IP addresses on a node must be on separate subnets (if heartbeat
monitoring over IP aliases is not used).
All Service IP addresses must be on a separate subnet from any of the base
subnets.
The service IP addresses can all be in the same or different subnets.
The persistent IP address can be in the same or different subnet from the
service IP address.
If you choose to use heartbeat monitoring over IP aliases, then the base IP
addresses can be on the same or different subnets as they are not monitored
by HACMP, only the HACMP supplied aliases are monitored.
Anyways, your setup does only make sense if you have non-hacmp'ed applications running in the same cluster - why would you need 5 link aggregated adapters to simply access the node (as you talk to the applications only via the service IPs). In addition - even when you configure link aggregation with multiple adapters, you still should have ONE backup adapter ....
You might want to consult the redbook
Regards
zxmaus
I have fulfilled as what redbook said:
Lets dont discuss on how I form the etherchannel, you can just assume it is single interface en0 and en1, with no etherchannel behind.
The redbook did not mention what will happen if persistent and service IP with same subnet lies on different interface, and I would like to know if it puts negative effect on hacmp or on AIX.
Lets make the environment looks like this:
en0
1.2.3.4 <-boot IP
192.168.1.1 <-persistent IP
en1
2.3.4.5 <-boot IP
192.168.1.10 <-service IP
What will happen if persistent IP got the same subnet as service IP, while they lies on different interfaces (en0 and en1) ?
I assume the only thing is that you cannot guarantee that the traffic that is meant to come in via the service IP really comes through the 'service' adapter and vice versa - it will rather be random. Anything else should work just fine.
Regards
zxmaus