HA4.5 on Aix5.2

[fjw999]

Hi,

I have just installed HA4.5 on Aix5.2(ML02) and am experiencing a problem. I have setup all the tcpip/IPAT etc but when I start HA I find that the new service ip address only reply's to every second ping and will only let me telnet to the new address every second attempt. Has anyone got any ideas why this could be happening.


Thanks

 

[petertp6]

Do you setup hardware address takeover?
If not, you ping/telnet with your ARP table (right IP/ wrong MAC address),
network device needs time to find out a new adapter(different MAC address) take this IP.

 

[fjw999]

I've set it up for HWAT which works fine when i bring HA up. I can see a message in the HA logs which says "ClearTooLong: In function clearTooLong" I'm not sure if this is connected to the problem i am seeing though.

Thanks

 

[DSMARWAY]

Hi ,

Have you got the latest HACMP 4.5 update loaded ?
i.e. lslpp -l | grep cluster

4.5.0.x I think 9/10 is out , check fix database for this.

 

[aixTim]

The telnet/ping every second time is classic symptom of having 2 default gateways (a feature of aix5), where one does not not have a route back (ie private network).

look for the "=>" on the right in netstat -r

 

[petertp6]

If you have 2 default G/W,
I suggest you use static route to make sure your request go to right G/W.
In my case,
a server(AIX 5.1) missed for a while in network----but not for all,
someone can still see it in network.
In the last, I found it has more than one G/W.
netstat -nr show it send packets to wrong G/W.
So I use Static route to force it use right one.

 

[fjw999]

After I start the cluster I do see the extra "=>" but i haven't set up any default gateways. After I start HA I see a new route appearing for my destination network with a GW of 127.0.0.1 which has the "=>" beside it, I presume this is what's cusing my problem. Is there any way of stopping this happening in Aix5.2

 

[petertp6]

As IBM told me, You have only way to solve it for now.
Setup a static route .