I also confirm the same problem, but changing the disk.usereportlun in advanced settings to 0 works for me!
i have multiple supermicro x7dvl boards w/ dual quad core l5320 and 24gb ram each
vsphere 5.1 installed on USB enterprise thumb drive
vcenter 5.1 linux appliance
2 mr sata 300 raid cards
2x 2tb WD RE4 raid 1 in host 1 using vmks version 3
2x 160gb WD enterprise (OLD disks) raid 1 in host 2 using vmks version 5
initially when i change it, the response time of datastore on the lsi mr sata and all datastore is painfully slow, however, i figured out how to make the datastore reasonable:
1. create raid volume on mr sata
2. in vshpere 5 change disk.usereportlun to 0
3. restart host
4. host will take longer too boot when enumerating data store (adds an extra 10 minutes or so)
5. add new volume from your mr sata card (this takes 10 minutes or so)
6. restart host
the volume should now be fully accessible and all additional volumes i've tried work the same as before as well. i've attached an external sata drive and used a NFS and an ISCSI share successfully. there is no drop in performance for the additional drives and browsing the datastores is just as quick as before the change to disk.usereportlunk
speeds of mr sata when uploading/downloading 3 large ISO files totaling ~12gb to datastore over single gigabit w/ small frames:
160gb WD: 27 MB/s writes, 37 MB/s reads, <12ms latency
2tb RE4: 75MB/s writes, 63 MB/s read, <11ms latency
before and after speeds of the other datastores remained almost identical and browsing datastores has no noticeable drop in speed. the only negative for the change in configuration is the additional time added to the host startup, and i think due to the increased startup time, i have to restart the host again using local login on the host machine after bootup about once very 10 bootups or so to get access to it.