hacmp problem

[aixnag]

i have two B80 servers in hacmp 4.4.1 (AIX 4.3.3-10) cluster in 'cascading' group. One day suddenly both servers node1, node2 came to boot_ip address.


I checked up all the hacmp log files,ERRLOG,SYS LOGS. none of the cluster events logged in hacmp log files.Cluster deamons are working fine.

Finally we had rebooted the servers in bring up cluster.

I wonder, any of you guys faced this kind of problem.


Thanks in advance
cng

 

[DSMARWAY]

hi,

i don't understand your problem in full , however

your primary node will have a boot / service address
and standby address
the secondary node will have a service address and fail standby address

The service is currently running on the primary node , right?


So did both servers have the same boot address , which meant you had duplicate ip addresses ?

Did your primary server fail ?

Did you do an arp -a ( to check which ip address was mapped to which MAC address,)

Could there have been a duplicate ip address on your network
causing this ?



just a few thoughts

 

[aixnag]

thank for the reply. after a week long stuggle, sorted out the problem.

in brief, primary server came to boot_ip address from service_ip address. only one error log detected saying TE_ADAPTER_FAIL, second error TE_ADA_REJOIN, TE_STABLE with interval of 2-3 seconds each.

but no cluster event notification in cluster logs. while we were troubleshotting this problem , second server came to boot_ip address. i checked second sever no cluster logs.

finally we took decision of rebooting of servers. after rebooting cluster starting working normally.

After this i started investing to find the reason for strange behaviour of the cluster. Finally found there is a mismatch in network switch end, its configured as HALF duplex, where as my n/w interface full duplex. Due to this exchange of heart beats were abnormal to log the event as adapter/network failure.

cng