The following steps help to troubleshoot problems with nodes that use the HTTPS agent. This assuming that agent is correctly installed. If this is not the case, see Deployment troubleshooting or Manual installation of HTTPS agents for more information.
Alternatively, to force the agent installation to replace the
values of any existing configuration parameters, preinstall the
agent without starting it (see Preinstall
an HTTPS agent), and then activate the agent with the
-force_config_mode
option as follows:
cscript opcactivate.vbs -srv
<management_server_host_name> -cert_srv
<certificate_server_host_name>
-force_config_mode
./opcactivate -srv
<management_server_host_name> -cert_srv
<certificate_server_host_name>
-force_config_mode
Check the agent installation log file. After you install or
upgrade an HTTPS agent, a log file is in available in:
<data_dir>\log\opc_inst.log.
After you deinstall an HTTPS agent, a log file is available in the following location:
%SYSTEMROOT%\temp\opc_inst.log
/var/tmp/opc_inst.log
.Check whether the management
server can resolve the node's host
name to an IP address, and whether the node can also resolve the
management server's host name. The operating system normally
provides a suitable command such as nslookup
or
dig
.
On nodes that run the Solaris operating system, the HPOM
provides the command ovgethostbyname
for this
purpose.
Check whether the management server can connect to the node, and whether the node can connect to the management server. The following command enables you to do this:
bbcutil -ping [<hostname>|<ip>][:<port>]] [count]
For example, to check whether a node can connect to manager1.example.com by sending 10 packets, open a command prompt on the node and type the following command:
bbcutil -ping manager1.example.com 10
If the connection is successful, the command returns
status=eServiceOK
.
To run the command on a management server that is part of a
cluster, you must also specify the resource group name by adding
the -ovrg
option.
bbcutil -ovrg <resource> -ping [<hostname>|<ip>][:<port>]] [count]
Check the status of the communication process, using the following command:
ovbbccb -status
Alternatively, to check the status of communication processes on a remote system, use the following command:
bbcutil -status [<hostname>|<ip>][:<port>]]
If the processes are not running, restart communication processes using the following command:
ovc -restart ovbbccb
If the communication process does not start successfully:
<data_dir>\log\System.txt
for error
messages.ovbbccb -nodaemon -verbose
and then press
Enter. This attempts to start the communication process, and
displays details of any errors.Check that the HTTP communication is possible between the node and management server. Open the following location in a web browser on a node or management server:
http://<hostname>:<port>/Hewlett-Packard/OpenView/BBC/
By default, the communication process listens for connections on port 383. You can check which port the communication process is listening on using the following command:
bbcutil -getcbport <hostname>
If HTTP communication is possible, the HTTP Communication Information Modules page opens.
Check that the node communicates with the correct management server. Check the management server using the following command on the node:
ovconfget sec.core.auth MANAGER
The command must return the correct management server hostname.
Check that the node has the correct management server core ID. Get the core ID using the following command on the management server:
ovcoreid [-ovrg server]
The -ovrg
option is only necessary if the
management server is part of a cluster. The command returns a core
ID, which must match the core ID that the following command returns
on the node:
ovconfget sec.core.auth MANAGER_ID
Check that the node communicates with the correct certificate server using the following on the node command:
ovconfget sec.cm.client CERTIFICATE_SERVER
The command must return the host name of the management server
that acts as the certificate authority. If the command returns an
incorrect value, the command ovcert -list
should
confirm that the node has no certificates. If this is the case, you
can set the correct value using the following command:
ovconfchg -ns sec.cm.client -set CERTIFICATE_SERVER <hostname>
You can then restart the agent using the following command
ovc -restart
Check that the node has received the certificates that it requires using the following command:
ovcert -check
If the certificates are missing, ensure that the node receives them. For more information, see Configuring Certificates.
ovcoreid
The core ID should exactly match the agent ID on the management server. To check this, right-click the node in the console tree, and then click Properties.... The node properties dialog appears. In the General tab, click Advanced Configuration. The Advanced Configuration dialog appears, which enables you to check that the IDs match, and change the agent ID on the management server if necessary.
The management server does not start to automatically deploy policies until the node's package inventory contains the agent package. Normally, this happens automatically. To view the node's package inventory, right-click the node in the console tree, and then click ViewPackage Inventory.
If the agent package is missing from the node's package inventory, but the agent is installed on the node, synchronize the node inventory manually. Right-click the node, and then click All TasksSynchronize inventory Packages. The management server adds the HTTPS agent package to its node inventory, and starts automatic policy deployment (if enabled).