Installing Cloud Control 12c agent on other blades thru RPM and response file sets wrong cluster name

Question

Hi,
Hopefully someone can help me with this. We installed em Cloud Control 12c and its agents on miscellaneous machines.
We first proceeded with two target-nodes part of a Grid Infra cluster (let's call it proclu_1). OS is RH6.
Then, using the same method (1/ yum -y install /tmp/oracle-agt-12.1.0.3.0-1.0.x86_64.rpm and then
2/ /etc/init.d/oracle-agt RESPONSE_FILE=/usr/lib/oracle/agent/agent.properties) we installed the agent on 2 other
blades that are part of another cluster (proclu_2). The agent.properties file above just contained the usual key-value lines:
OMS_HOST=omsmachine.stuff.be
OMS_PORT=4889
AGENT_REGISTRATION_PASSWORD=ametemp
AGENT_USERNAME=oracle
AGENT_GROUP=oinstall
AGENT_PORT=3873
ORACLE_HOSTNAME=promachine.stuff.be
Now if I change dir to /usr/lib/oracle/agent and search for the name "proclu*" from there, I find the name of the
first install (on first cluster proclu_1):
[oracle@bladb0406 agent]$ pwd
/usr/lib/oracle
[oracle@bladb0406 oracle]$ find . -name "proclu_1*"
./agent/agent_inst/sysman/emd/state/has/proclu2/cluster/proclu
./agent/agent_inst/sysman/emd/state/configstate/cluster/proclu_1
./agent/agent_inst/sysman/emd/state/statemgmt/cluster/proclu_1
./agent/agent_inst/sysman/log/targets/cluster/proclu_1
My questions:
1/ on the first two nodes (blada0204 & blada0406) of our first cluster doing this 'find' returns the
same directory names with "proclu_1", this is what we expected: where did the oracle-agt script pick up
our cluster name ?
2/ why has it set the same -and wrong- "proclu_1" name on our two other nodes (bladb0204 & bladb0406) ??
(we would've expected "proclu_2"...)
Is there a way to correct this, or should I reinstall the agent (after configuring/modifying something somewhere
so that it picks up the correct cluster name...) ??
Thanks a lot.
Regards,
Seb

Answer 1

Unfortunately, this goes back to the RAC install procedures.   Check the cluster name for each cluster.   If they're all installed the same, then you'll need to manually discover/edit name when you promote so that your clusters are unique in EM.
To verify name set for clusterware installation:
<CRS_HOME>/bin/cemutlo -n