Unsolved
This post is more than 5 years old
6 Posts
0
1026
April 24th, 2018 05:00
Issue Mapping HBA to VNX5600 with Brocade 6500's
Hi All,
Hoping someone may be able to provide some guidance here. I am having trouble getting LUN’s mapped to select HBA’s on a couple ESXi 6.5 hosts, while other HBA’s on the same host are working properly. Storage is a VNX5600, switches are Brocade 6500’s, HBA’s are QLogic 2560’s, FW 8.05 and driver 2.1.57.0.
I have a cluster of 6 hosts, each with 2 x single port QLogic 2560’s in Poweredge R730’s. The trouble I am having is that I have 3 hosts where the HBA’s are not picking up storage and/or registering with the VNX5600 from within ESXi, and not registering themselves on the initiators page in the VNX. I have checked and double checked the zoning and it appears identical to the card that is working. I have removed and recreated it using web tools.
The card that is not picking up the LUN’s does appear in the Name Server section of the switch, so I can only assume the switch is seeing it properly. However the cards do not always register the FDMI host name. The other odd bit is, if I change the zoning and rescan storage on the host so that it does its registration process, *sometimes* the HBA will register its WWN in the storage as UNKNOWN and only as a single path, not both paths. I have tried associating the UNKNOWN initiator with the correct host, assigning to a storage group, and using engineering mode to ensure all paths are registered.
I have also rebooted and powered cycled the effected hosts to ensure the HBA’s are registering. I am at a loss here as the environment is new to me. There are many (40+ hosts) that have the same setup (2 x single port QLogic 2560’s) and they are all working properly with paths from both HBA’s showing up. I do not understand what would be causing these inconsistencies. I have gone as far as to mimic the working configurations in terms of using the same zone mappings to the front-end ports on both controllers.
I have swapped cables, transceivers, switch ports. None of which I suspect would be the problem otherwise the switch would not being the HBA.
Any help would be greatly appreciated!
kelleg
4.5K Posts
0
April 24th, 2018 13:00
Generally the HBA's in a single host must be the same model. In one of the hosts where the HBA's not not connecting to the array, are there more than the two 2560 HBA's? You also say that other HBA's in a host are connecting to the array. Are these the same model/make as the 2560's?
You might also want to contact VMware to see if they have any suggestions?
glen
121xx
6 Posts
0
April 24th, 2018 16:00
Hi Glen,
Thanks for the help. The cards in the servers are the same model, same driver, firmware and flash bios. It appears the problem is actually that these problematic cards are not properly registering, since under the FC4 section of the Brocade's, the field is 'none'. So it looks like a problem between the hba and the switches.
Any additional pointers that can anyone provide is greatly appreciated.
121xx
6 Posts
0
April 26th, 2018 04:00
Seems I figured out the issue. I was changing switch ports without powering down the HBA. Somehow that information was being stored. I had to remove power to the server/HBA entirely, change the switch port, then power back on. Only then would the FDMI information properly register.
I do have another problem now in getting datastores to successfully create but that is a different thread.
Thanks for all your help.