Fresh ODI installations sometimes end up with unreachable ODI agents like this example:
oracle(AFIS)@phl21240-db03 (QA):~$ /u01/Middleware/user_projects/domains/base_domain/bin/agent.sh -NAME=OracleDIAgent1 2019-01-28 16:47:18.932 NOTIFICATION New data source: [ODIQA_ODI_REPO/*******@jdbc:oracle:thin:@//phl21240-db03.qa.dc.XXX.hh:1521/AFIS] 2019-01-28 16:47:21.618 NOTIFICATION ODI-1128 Agent OracleDIAgent1 is starting. Container: STANDALONE. Agent Version: 12.2.1. Port: 20910. JMX Port: 21910. Allowing : 127.0.0.1,10.120.40.108,127.0.0.1 2019-01-28 16:47:25.931 NOTIFICATION New data source: [ODIQA_ODI_REPO/*******@jdbc:oracle:thin:@(description=(address=(host=phl21240-db03.qa.dc.XXX.hh)(protocol=tcp)(port=1521))(connect_data=(service_name=AFIS)(server=dedicated)))] 2019-01-28 16:47:26.464 NOTIFICATION ODI-1111 Agent OracleDIAgent1 started. Agent version: 12.2.1. Port: 20910. JMX Port: 21910. 2019-01-28 16:47:26.465 NOTIFICATION ODI-1136 Starting Schedulers on Agent OracleDIAgent1. 2019-01-28 16:47:26.657 NOTIFICATION Ignoring Exception in method isSupportsHeaderAuth() : Connection refused (Connection refused) 2019-01-28 16:47:26.677 WARNING ODI-2013 Ping to agent OracleDIAgent1 failed with "oracle.odi.runtime.agent.invocation.InvocationException: ODI-1424: Agent host or port cannot be reached using http://phl21240-db03.qa.dc.XXX.hh:20910/oraclediagent.". 2019-01-28 16:47:26.678 WARNING ODI-2010 Agent OracleDIAgent1 is not reachable. The operation will be repeated after 30 seconds interval until success 2019-01-28 16:47:56.684 NOTIFICATION Ignoring Exception in method isSupportsHeaderAuth() : Connection refused (Connection refused) Options
On analysis, you see that the agent indeed refuses any connections:
[E1JSchreiber@phl21240-db03 ~]$ telnet phl21240-db03.qa.dc.XXX.hh 20910 Trying 10.120.40.108... telnet: connect to address 10.120.40.108: Connection refused [E1JSchreiber@phl21240-db03 ~]$ telnet 127.0.0.1 20910 Trying 127.0.0.1... Connected to 127.0.0.1.
Reason for this might be that the Agents network interface is configured to (only) use localhost. This is caused by selecting localhost, which is default, when creating the agend using config.sh. Correct is selecting „All local addresses“ for „Server Listen Address“.
To correct the settings afterwards, you can modify <…>/Middleware/user_projects/domains/base_domain/config/fmwconfig/components/system-comp-config.xml and configure the agents hosts fqdn name as <listen-address> instead of „localhost“ like below:
<listen-address>fmw1.int.loopback.org</listen-address>