Nagios Support Forum

I have 8 stacked 3750 cisco switches.

I used the wizard to add the switch.
It detected all the ports, and looked like it added everything.

When I go to view Service Details, there is nothing regarding this switch.
No ports are monitored.

Now I'm having the problem, where any changes I made stay out of sync.
I'm in the Nagios Core Configuration Manager.
I make a change to a switch port.
After making the change the status changes to "Out of Sync"
So I hit the button: "Apply configuration"
It does it's think and says configurations applied successfully.
But then I go back and look at the port, and it still says "Out of Sync"

This are some useful debugging steps in this documentation.
http://library.nagios.com/library/produ ... es-into-xi

You should also try this:
http://go.nagios.com/forum/134/506

Those suggestions were not very helpful.

What I finally discovered is if I installed a switch using the wizard to detect all the ports.
If a 1GB port had a 100Mb device plugged into it, the port was detected as 100MBs.
Then the default settings for high and low settings would be set to something like 20,20 and 50,50.

Since I knew the port was a 1GB, changed the high and low settings to 200,200 and 500,500.

Whoola! the graphs were now visible.

I did have change these settings for several of the ports that were not correctly, but I changed these settings, I had graphs.
I could also now see the switch under Service Details.

I'm not sure I follow. By High/Low settings you are referring to the Thresholds for alerts?

Do you have an example of how this changed the prefdata output? I have a suspicion that this change was unrelated to the graph starting to work, do you have a connection other then time to go by?

That was it, I changed the high/low thresholds, and I got them to work.
It make no sense.

I am now having problems again, I added another switch today, and the switch I was originally having problems with no longer has graphs again.

If I remove the new switch, the graphs come back.

This software is very confusing, and problems I am having make no sense at all.

Why would adding a new switch cause the graphs to disappear on another switch, and the switch to longer be listed under Service Details.

both switches were added using the Monitoring Wizard.

That's what I'm saying, you are not the only factor inflicting changes on the system. Think of it this way, "Graphs are being removed at the same time of when I add a new host." VS "I added a host so that must have been the reason for the Graphing to stop."

Find another vector or dimension that links the events together, make sure you have two dimensions(time and something else or something else and something else). I'm not sure that's even possible, but in order to fix this you need to think of time as just being coincidence.

For graphs watch the performance data and record it's values, keep a record.

Ok,

So do you have any clue as to why this would happen?

I just started using Nagios a week ago.
So far it is very frustrating, and I can find very little documentation.
Are there any step instructions for adding switches, and also for deleting switches from Nagios xi?

Before that I used Zenoss to monitor, and everything worked perfectly, and I never had problem setting up a switch to be monitored.

This new switch I added was added before, but then was removed when I was troubleshooting the 8 stacked switches.

This new switch is actually 4 stacked switches.
So after I got the 8 stack switch to work, I had to add the stacked switches back.
Just as soon as a I add the switch back, the details for the other switch disappear.

Is there possibly some left over files that need to be deleted when a switch is removed from nagios?

There are some known issues with changing case. Try using different names, like add a date(month/year) suffix. This bug will be addressed in 1.2, though we know there was something corrected we don't know the extent of this change's effect.

I'm not sure this will improve further in future release, but at least we know about the problem and are looking at other side effects.

The dimension I was talking about and that you had to look for might have been non-uniq names. Hosts are added into NDOBD when a check result is sent, but they are never removed(AFAIK). The data can be found in MySQL. Perhaps it's better to de-activate hosts instead of deleting them, to keep historical data accessible.

As you are experimenting it's likely that current conditions don't reflect proper operating conditions. I believe that if you weren't wrenching hosts by deleting and recreating them that things would have worked better and hopefully this isn't something normally done in production.

Take a look at this thread for more information.
http://go.nagios.com/forum/295