Is it possible to modify the FCS database information on an MDS switch? We have a 5 switch fabric which we are monitoring with ECC 5.2 SP5. We recently changed the IP address of the mgmt0 port on some of the switches in the fabric but ECC still shows the old address in Properties and in StorageScope reports. If you rediscover the switch using the new IP address it updates, but then seems to change the properties of the other switches in the fabric - it seems to pick up the IP address of the other switches from somewhere in the MDS. I'm guessing it queries the FCS database - when I query it from the MDS CLI, it shows two Switch MGMT-Addresses - the new one and the old one. You end up going round in circles. You rediscover switch A which corrects the IP address but changes switch B to the old address. You then rediscover switch B which corrects that one but changes switch A back to its old address. It took me a day of thinking I was going mad to find out what was happening!
No I didn't remove the old IP address first. I assumed I was just changing it - I didn't realise I would need to remove the old one first.
The issue is really what is seen in ECC - I can access the switches via the new addresses OK but ECC shows the old address when you select properties or in Storagescope reports. It also occasionally issues "Statistics collection failed" messages which I was guessing may be because it is trying to use the old IP address
I think you can assign multiple IP addresses to the mgmt0 interface, but am unable to confirm right now. That is why I wanted to see the out put of the 'show int mgmt0' command. If it shows both ip addresses, the old and the new, then you can just go into config mode and enter 'no ip w.x.y.x ....'
The MGT port should really only have and display one IP address for it. I'm not sure how via CLI or the GUI you could assign multiple addresses to it but you never can tell there could be a way. An output of the MGT port configuration as previously noted is the way to confirm that.
I talked to one of the guys in ECC support and he stated that it is very possible you have multiple entries for the switch in the ECC database and this could be causing the issue. He recommended in deleting all the associated switches and rediscovering them. This should resolve the issue.
Here is a post of the sh int mgmt0 followed by sh fcs data
Each VSAN shows two Switch MGMT-Addresses under the FCS data - I think this is what ECC is discovering (unless the old IP address is stored somewhere else on the switch as well)
I have been round the loop of deleting all the switches out of ECC and rediscovering them but it discovers the first switch in the fabric OK, then changes the discovered IP of the first switch when it moves on to the next
sh int mgmt0
mgmt0 is up Hardware is FastEthernet Address is 0013.1957.e080 Internet address is 188.8.131.52/27 MTU 1500 bytes, BW 100 Mbps full Duplex 453967754 packets input, 1518768615 bytes 0 multicast frames, 0 compressed 1 input errors, 13688 frame, 0 overrun 0 fifo 300492652 packets output, 2101371907 bytes, 0 underruns 0 output errors, 63821 collisions, 432 fifo 0 carrier errors
Have tried editing the FCS but it says Platform Management Address not available, I think because there is no pre defined platform name. If you define a new platform name, it doesn't get linked to the old data