EIF Probe Swings and Roundabouts

Nice to see some steps are being made in making the installation and configuration of components such as probes much more easier. Trying to get something as simple as getting a probe installed and and running could sometimes turn into a 2-hour long swearing session, only for you to spot the “Oh My God Am I Dumb?” typo in the props file, which invariably resulted in the screen being switched off in disgust, a swift exit to the pub. The  nco_install_integration for the revised probes couldn’t work any better and drops the correct files in the correct locations, but it would be nice if it could be made to update the nco_pa.conf file at installation too. Gone is the annoyance of trying to track down the nco_native prerequsite of the earlier versions of the EIF probe, as now everything is self-contained in the package, with the exception of Java. The tivoli_eif.rules file does now explicitly state you need Java 1.5 installed, which I don’t believe it did before, but you must make sure the netcool user, and root, can both run a java -version command successfully. If you’re testing the EIF probe from the command line, then the netcool user needs to have the path to java set in it’s profile ( e.g. PATH=$PATH:$HOME/bin:/opt/IBM/tivoli/netcool/bin:/opt/IBM/tivoli/netcool/omnibus/bin:/usr/java/jre1.5.0_21/bin). Similarly, if you’re using Process Control to start the EIF probe, and nco_pad runs as root, it too needs to be able to source the java binary from it’s PATH.

The EIF probe rules have been amended to allow it to use include files for processing events from other Tivoli products. These rules are included under the extensions directory of Omnibus 7.3.1 and cover products such as TSM,  TEC and TADDM, as well as incorporating events from ITM for Virtual Servers as well as the predicitive events from ITM Performance Analyser. To have the supplied rules picked up, you simply uncomment out the section you want from tivoli_eif.rules file and they will be picked up when the probe starts. In the tivoli_eif.rules file, there is an includes file reference called itm_event.rules which aren’t included with the EIF probe but these have been available in the past with the ITM integration piece. Perhaps they’re being revised too, but curiously the Omnibus documentation points you at a IBM URL that says these rules files will show up around April 2011. Odd.

I wanted to have an EIF probe installed on AGG_P and AGG_B running in peer-to-peer failover mode so amended the props files on each server and amended the rules files to include the ones for virtualisation and predictive eventing. I plan to take a look at these in more depth later as I have an ESXi host that runs all these VMs, which includes an ITM 6.2.2 FP3 box with Performance Analyser running. I didn’t expect the probes to start first time and I was correct. A very familiar exercise of staring at the props file looking for the mistake ensued. With the probe in debug mode, it was complaining it couldn’t connect to the NCOMS server which is always good for a laugh. The probe categorically listed AGG_P as the target, and not NCOMS. I tried explicitly starting the probe with the propsfile flag directing it at the propsfile that only mentioned AGG_P. Still complaining it couldn’t connect to NCOMS. With the lights of the pub in the distance, I almost called it a day, but thought I’d take a quick look at these new rules I was ‘include’-ing. Oh joy.

# NOTE: Assumes ObjectServer name is NCOMS please edit if different.
# ——————————————————————
alerts_target = registertarget( “NCOMS”, “”, “alerts.status”, “alerts.details” )
vmstatus_target = registertarget( “NCOMS”, “”, “custom.vmstatus”, “” )

Maybe that will save someone an hour of teeth-gnashing. Pub o’clock.

 

Mark Delaney SYSTEMSMANAGED Ltd