That document doesn't clearly say how to add the UCS fabric interconnects...
Do we add the add UCS in this order?
- VIP IP address with snmp and UCS manager credentials
- click on primary FI in UCS Overview resource and add with snmp and UCS manager credentials
- click on subordinate FI in UCS Overview resource and add with just snmp (cannot use UCS manager creds unless you fail over to enter them)
- repeat and rinse for blades or discover your CMIC IP range to add blades
- also add all the blades by OS IP's so you can monitor by WMI or SNMP through production interfaces
I didn't get any of that out of the document and am not sure if even this is right....
let me try to answer your question and provide some info which can be useful.
UCS manager credentials are used only once when you are adding the UCS. If you add the UCS, blades and FI should be discovered automaticaly and you should see them on the UCS overview resource. If you already monitored any of the blade or FI they should be automaticaly linked (you can click on the blade or FI on the UCS overview and it will bring you to the Node details page). If they are not monitored and you click on them you will be asked if you want to add them (Add node wizard), When you are adding FI or Blade into NPM you dont need UCS credentials and you are adding those as "common" nodes. Just provide SNMP credentials or add them as ICMP nodes. FI and blades must be properly configured on the UCS to support this.
Hope it helps you a bit.
so VIP with UCS creds, the rest just snmp, what about adding the blades by mgmt and production interface?
with HP we would not have monitored by iLO since thats out of band and I feel adding UCS blades by CMIC is the same- I want to monitor by the production Interface.
I am not exactly sure what you mean by "mgmt and production interface".
Anyway if yours blades are not clickable on the UCS overview and you know the IP you can still add it via add note wizard and it will be recognized and linked as blade.
"mgmt and production interface"meaning adding once by the UCS blade mgmt IP address then again by the OS IP address to get WMI and/or production interface up/down etc.
I have found my best results now to be;
- Add the FI VIP address as ICMP node but with UCS creds
- Add each FI as snmp node WITHOUT UCS creds (when I added more then one device with UCS creds it duplicated database records)
- Add node CMIC ip-mgmt address nodes by ICMP
- Add node OS ip addresses by SNMP
- Create a group for the FI's and a group for the blades
- Create a dependency with FI group as parent and blade group as child
This way from any blade page I can click back up to the FI's quickly without needing to set a custom property and also suppress alerts.
Looking at adding CMIC ip-mgmt pool address range to a nightly discovery and excluding it from node-down alerts so as not to duplicate the OS alerts.