I am having fits trying to add a DELL MD3220i to our existing infrastructure. We purchased the MD3220i to offload lab and non-production environments on cost effective storage.
Environment:
3x Dell M610 blades with vSphere 4.0
- 4x Broadcom 5709
- - 2 pNICs in vSwitch0 with vMotion and Management
- - 2 pNICs in vSwitch1 with VLAN tranked VM traffic
- 2x Qlogic 8Gb FC HBA
1x Dell MD3220i
- 2x Storage controllers w/4 iSCSI ports each
The goal was to add the iSCSI interfaces to vSwitch1 and trunk the data to the MD3220i. I created 2 new vmkernel ports (vmk1 & vmk2) with 2 seperate subnets and VLANs (500 & 501). I configured the VLAN span the appropriate switches and the MD3220i. I then used vmkping to verify connectivity between all endpoints. I add vmk1 and vmk2 to vmhba32 using esxcli. I added the first iSCSI address to the software initiator as a dynamic disocvery target.
When I scan the software intiator, all LUNs are found and all paths are listed as active. But I cannot access the LUN unless I go through the paths, one by one, and eventually mark a path as preferred. This is with the FIXED path selection. MRU and Round Robin make it worse.
The vmkernel logs show constant reports that the LUN is "not available on this device" as it searches for the proper path.
I have never seen anything like this. Any ideas why the connectivity is good, the LUNs are discovered, but the pathing is completely wrong? It behaves like the MD3220i is failing to tell VMware were to connect for an active path.
