hacmp 4.3.2 topology sysc error

[nmk0429]

hi! everyone.
i am with aix se in ibm rs6000 in korea.
Help me of my problem.
thank you for anyone.
1)System : 7015 R50, 7017 S70 HACMP, OS:AIX V4.3.2.
Cluster version : hacmp 4.3.1
2)Problem : topology sync time error occured
3)Message : could not open a socket connection to Node oracle1. check root passwords, permissions, and /.rhosts file

 

[DSMARWAY]

hi,

Is this a new set-up?

check your .rhosts file under root , and make sure you can
rlogin to each machine both ways ?

what type of cluster is it i.e. cascading,rotating

in your topolog section , can you list what you have got configured.

 

[nmk0429]

thank you for answer by DSMARWAY user.

1)My system is new setup hacmp.
2)Hacmp type is mutual takeover of cascading in hacmp.
3)i can rlogin to each machine both way.

i want your fast response because continue complaining
our's customer!

thank you so much everyon with this site.

 

[DSMARWAY]

hi,

what version of hacmp are you running i.e. HACMP classic
or HACMP cluster ?

Can you ping all network cards ?

Are there any errors showing up in errpt regarding ethernet?

How have you got your network cards plugged i.e. into a switch or a hub ?

If a switch , what speed are the ports set to ?
what speed are your network cards set to ?

i.e. do lsattr -El ent? ?=number of card

make sure the switch port and network card are set to the same speed , if set to auto negotiate , force the speed
i.e. if your network card is capable of 100MB then
set it to 100MB Full duplex .

the other thing is post your configuration :-
type command

cllsif ( lists details of network cards)

clshowres ( lists resource groups)

one other thing is can you do an rsh and run a command
on both servers ?

Also what oerating system maintenance level are your running

i.e. type instfix -i | grep -i ml
( note last line)

hth

 

[aixat]

WOW
YOU are with IBM System expert at IBM site in korea.

check

http://www.redboooks.ibm.com

then search on hacmp.

you can do search on techsupport.service.ibm.com then search on vmm and memory - the problem is the timer process gets pagged out to paging space and when it is back in process the timeout occurs.

Search on this forum under the name avadh you will find hits on how to do the tuning for oracle.

 

[nmk0429]

hi everyone!
thank your for DSMARWAY's answer.
this is answer for your(DSMARWAY) question.
1)HACMP version:HACMP classic 4.3.1.
2)i can ping all network card.
3)i was not message of errpt of ethernet.
4)network card connected to hub.
5)rsh,rlogin possible both setvers.
6)oslevel 4.3.2
7)ML is 2

 

[DSMARWAY]

hi,

if you are using mutual takeover policy , have you got 2 resource groups and in each topology you have , 2 boot/ service addresses , 2 standby and non-ip heartbeat connectionfor each resource group and are they uniquely named in /etc/hosts file
Check in each resource group your ip addresses are in the same subnet i.e. 255.255.255.0 ( for service/boot & standby)


what i am thinking , may be wrong is that it is failing to sync your topology , either you have some mismatch in here
with naming or in /etc/hosts file . I guess you've probably checked this ?

you can run the cldiag utility under /usr/sbin/cluster/diag to see if you can debug the error
if set to debug , this enables clstrmgr debugging output
will be logged under hacmp.out
note be careful when using this option , if unsure don't run .

You can try running following commands , ( these may be for HACMP 4.4 only but you can try it ,

clvinval - checks if any PTFS installed will break HA
clvbos - checks all necessary BOS s/w are installed
clvreq - checks pre-reqs are installed
clvhacmp - checks hacmps files, processes
clverify - runs sync on topolog& resource

one other thing you can try is refreshing inetd daemon
refresh -s inetd

then retry sync topology