Environment
NetIQ Sentinel 7.2 Sentinel Server
Situation
- Sentinel was installed, and subsequently upgraded, in a SLES HA environment.
- /var/opt/novell/log/install.log does not have any entries for the latest upgrade.
- The rpm database on each node reports the current release.
- Executing `db.sh sql SIEM dbauser "select db_version from
esec_db_version" returns the previous version.
Resolution
Prior to following this workaround, ensure the following is done as
a precaution:
- Perform a backup of the config and data.
- Unless otherwise stated, all steps are to be performed on a
passive node.
- On the selected passive node, stop cluster services (rcopenais stop)
- Remove Sentinel.(/opt/novell/sentinel/setup/uninstall–sentinel)
- Copy the install.props that was generated on the initial node that sentinel was installed from.
- If not already present and extracted, copy the sentinel_server-7.2.0.0-1514.x86_64.tar.gz file to /tmp
- Change to the /tmp folder. Expand the Sentinel 7.2.0 gz file if not already extracted (tar -zxvf sentinel_server-7.2.0.0-1514.x86_64.tar.gz)
- CD into the sentinel_server-7.2.0.0-1514.x86_64 folder
- Run the following:
./install-sentinel --no-start --cluster-node --unattended=/tmp/install.props && insserv -r sentinel - Rsync the following files between the active node and this
re-installed node:
- /etc/opt/novell/sentinel/config/obj-component.ConnectionManager.properties
- /etc/opt/novell/sentinel/config/obj-component.JasperReportingComponent.properties
- /etc/opt/novell/sentinel/config/configuration.properties
- /etc/opt/novell/sentinel/config/user.properties - Stop the Sentinel clustered resource. Manually perform the
following:
a. Mount the shared resource.
b. (Optional) Manually assign the clustered address as a secondary address. (ifconfig eth0:1 169.172.1.2). - Manually start the sentinel daemon (as user novell, run
server.sh start) and wait 5 - 10 minutes.
If `netstat -an | grep 8443` still doesn't show the server is listening on the port, /opt/novell/sentinel/setup/configure.sh may require to be executed. If it is, don't change anything. - Perform the upgrade as user root:
a. If not already present and extracted, copy the sentinel_server-7.2.1.0-1561.x86_64.tar.gz to the /tmp of the temp node.
b. If not already extracted, extract sentinel_server-7.2.1.0-1561.x86_64.tar.gz.
c. cd sentinel_server-7.2.1.0-1561.x86_64
d. ./install-sentinel - Verify Sentinel starts successfully, is accessible, and is starting to process events.
- Verify the install was recorded in the
/var/opt/novell/sentinel/log/install.log file.
- Stop Sentinel (As user novell, run server.sh stop)
- Manually dismount the shared resource. If 9b was performed, manually unbind the secondary address (ifdown eth0:1).
- Verify sentinel is not set to start automatically. (chkconfig
sentinel stop).
- Start cluster services (rcopenais start).
- Verify Sentinel starts successfully on one of the nodes, is
accessible, and is starting to process events.
Cause
Each node, including the active node, was upgraded using the
--cluster-node parameter.
Additional Information
The --cluster-node parameter instructs the upgrade not to touch the
/var/opt/novell folder, as it tells the install routine that the
current node is a passive node within the cluster.
This workaround is not applicable for a Sentinel HA Appliance environment.
This workaround is not applicable for a Sentinel HA Appliance environment.