In the blog post "Solution Manager - #sapteched09 Carry Out Item 1", I showed a few of the remote operating system monitor screens we can see in Solution Manager. I also showed how some systems should not be displayed:
The stuff up there on the top left, saying "old", with the "X" in the console icon, are systems that were gone before we installed our production Solution Manager instance.
I stumbled around the Solution Manager system, trying to figure out where these bad data were coming from. Below you can see one of the tables I found, with context and segment columns showing the old system. I even deleted these rows to try to purge the views, but the old stuff just kept showing through.
After too much time wasted on this, I put it aside for a nice lunch and, like many logic problems that aren't solved on the first approach, my inner brain rearranged the problem and told me I was looking in the wrong place. The display system didn't exist when those records of the old systems were created, so of course it was pulling them from the remote system that did exist.
Even though our development system is on newer hardware, the records inside it date back almost 15 years. When CCMS was set up sometime in the last century, data records were created for every system name change. And of course, these persisted, as there is no built-in logic that would assume that after a specific amount of time, a host is dead and never coming back
In RZ20 [Sorry, make that RZ21 !], I finally found the trashcan icon for "segment names" (data records for an application server). After highlighting that row, I got the "Are you sure, punk?" message:
And then there were none.
It's easy, once you find the trash can. However much I looked on online documentation, all I could find was how to add nodes, and how to drop whole systems. I didn't find how to clean up the trash. Maybe because no one else changes systems?
[ UPDATE 05-Mar-2010 - See the wiki page for more details - Removing a defunct node from OS06N ]
2 Comments
