Automation: Monitoring the Monitors Isn't It

Published: April 13, 2009

by Dan Burger

A few weeks ago, I noticed that Tom Duncan was making a presentation about Management Central, a component of System i Navigator, at the The OMNI User meeting in Chicago. I was curious how the group responded to his topic because Management Central can be a little quirky. As it turned out, Duncan says, the topic was quite well received based on the amount of questions the audience had about Management Central.

Most people know about Management Central, but like a lot of things on the AS/400 and its successors, the familiarity varies from having simply heard about it to having a limited knowledge of how it works. The full functionality remains somewhat mysterious. Duncan is an experienced user. His seasoning goes back to the early 1990s, which is pretty much when Management Central was introduced. "I've been through the wars on this thing," he says.

Some of you veterans of the AS/400 probably remember the early days of Management Central, when Java was relatively new to the AS/400. Because Management Central was written in Java, it was not without its problems. Duncan recalls that when new Java PTFs were put on the box, "you had to cross your fingers and hope nothing would break. It was grim in terms of how buggy it was. That was one of the bumpier parts in the road."

It's a lot more stable now, he's quick to point out. Duncan runs the OS/400 V5R4 operating system on a configuration that includes a single-partitioned 570, a 550 with three partitions, and a single-partitioned 520. "We run multiple system monitors and a whole lot of job monitoring," Duncan says while declining to be more specific because his company prefers it that way.

"I use the product more than most people," Duncan claims. "I've had some performance issues that I've addressed with IBM, and the feedback to me has been 'we never anticipated anybody running that many monitors.' There's a central machine that runs Management Central for the various '400s in the system, and IBM's advice to me is that I break up the monitoring among various central machines--distribute the load. While there is a specific Management Central Java server that runs on all of the '400s, it only talks to the central machine. The Management Central Java server on the central machine has to do a lot of work and it sometimes gets overloaded. That's why I was advised to distribute the monitoring workload among several machines."

So Duncan is somewhat of a monitoring maniac. I don't mean that in a bad way. He's just figured out how to take monitoring to the nth degree. Some AS/400 users never touch the higher levels of the machine's or its management software's capabilities. When it comes to monitoring, Duncan has done just the opposite. He's your go-to systems monitoring guy. If your local user group is looking for a speaker (and most are), you might want to dial up Duncan. By the way, he's currently the president of OMNI, so he's not hard to find.

One person who attended the recent presentation brought up a problem Duncan has heard about before. The question dealt with monitors that occasionally fail, causing the system admin to restart them. This was not exactly what automation is supposed to provide, the attendee groused.

"I told him in V5R2, IBM added an auto restart function that can be configured so that it knows what to do when a monitor fails. The function can be programmed to know how long it should attempt to restart or how many attempts it should make. However, that functionality isn't as stable as I'd like to see it. So that is a drawback. Monitors will failover and they won't always restart. So, sadly, you have to monitor your monitors. That can be very annoying and very debilitating. If you are relying on your monitors to alert you to certain conditions and one or two failover and don't restart, you are not getting informed and you can get bit in the butt."

As far as Duncan knows there is no fix in the pipeline at IBM.

"I'm not sure IBM is looking for fixes and enhancements in this functionality. I haven't gotten IBM to buy off on the fact that this stuff doesn't always restart."

It could be that IBM has other fish to fry. With the i 6.1 operating system, Big Blue began pushing its System Director product, which assumes most of the functionality of System i Navigator and some of the functionality of Management Central. It just so happens that some functionality didn't get added to System Director. Guess what one of those functions is. Yep, it's the monitoring features. If you want that functionality, you have to stay with the Navigator product.

Go figure.

RELATED STORIES

New Web Console Debuts with i5/OS V6R1

Admin Alert: V6R1 Changes for the i5/OS Administrator, Part 1

iSeries Navigator Job Monitor

Admin Alert: When Management Central Server Won't Start

                     Post this story to del.icio.us
               Post this story to Digg
    Post this story to Slashdot