CC 6.1 and SMI agent

Personally I found the Installation Guide to be very helpful in this respect. Are you running into specific problems with the discovery, or are you just trying to get more info to plan out your deployment?

If there are specific questions I may be able to help a bit. I've got multiple FCC agents with multiple EOS and FOS SMI providers monitoring switches on several different networks in multiple data centers.

Allen,

We are having some strange discover issues when we have a ED48K director isl'd to a number of blade switches. The ED48K is on one Subnet/network and the Blades are on another subnet/network. We are using the SMI agent and have discovered the ED48K as our Proxy switches in the SMI config, we then discover the ED48k's in CC but what normally happens is it says it needs to discover the blades by SNMP. I am trying to figure out how CC discovers the Blades through the 48K's when all we have put into the SMI agent config is the ED48k switches.

How many SMI's have you got deployed and FCC's?? have you also edited the providers.xml etc when you have a host with dual nic's and what to force the smi agent to use a specific nic??

Thanks

Andy

I have 5 SMI providers deployed through our Canadian data centers. These are broken down to three MEOS and two FOS SMI providers. One of each are deployed at two separate data centers and the third MEOS SMI provider is deployed at a third site (with no B series switches at that site yet).

I have not had to do any special modifications for dual home hosts as all the providers are deployed on VMs and only have a single virtual network port each.

As for how the other switches are being discovered, you only need to provide a single proxy connection (hence the term proxy switch) into a Brocade fabric to discover all the switches in that fabric. However you will need to provide the additional information for SNMP discovery as both protocols are used for communication between the switch and the ControlCenter infrastructure.

If you add the additional switches to the SMI provider you can enable the functionality to use a different switch as proxy automatically if a fabric segmentation event occurs (or you lose IP connectivity to your primary proxy). Instructions for this are also included in the SMI provider documentation. You don't HAVE to do this, but it could allow you to maintain monitoring and management functionality in the event of some kind of IP connectivity issue.

Can you please tell me if you have to SNMP trap coorectly to all the switches in a fabric before you will get a full discovery on a single switch? This is what we have been told by EMC PS people. We currently have a total of 46 swithes partially discovered. it seems we can not get a full discovery. We have  a DCX as the proxy , and that works fine. We have 3900's, 4100's, 4900's and other never switches. Do you know any special SNMP settings for snmpv1 that need to be set besides the EW community string on each host? Thanks

Hi,

please check brocade SMI-S provider release notes if your current switches / FOS combination is supported for SMI-S discover... we have four 24000 with 5.3.x and there are not supported because they have FC2-16 blades installed.

We have 25 switches (200e / 7500 / 24000 / 48000 and 4016 [FTS blade switches]) and all could be discovered by SMI-S only our 24000 had problems. But now I use other switches for that fabric as proxy...