We have an IBM DS3512 with redundant controllers and redundant switches connected to three blades. (using MRU)
Simple design is that each blade has two hba's each connected to their own Fibre Switch and both controllers on the san are connected to both switches.
When I power on the second Fibre Switch the paths come up and are marked as active/standby active (io)/standby but this is when the trouble starts.
The disk access to vm's begins to burst ie. If I try to upload a 5GB file it goes from 70MB/s to 0MB/s backwards and forwards.
Any ideas what might be causing this? Should both HBA's have an active path...?
Check switch configuration.
And be sure to keep active/standby.
The issue may be cause by LUN tresspass across controllers, so check also if you have some warning on storage side.
Is there anything I need to set to get it to be Active/Passive?
No errors on the SAN all green lights… Events are clear.
Each HBA has an Active/Passive path but one is market Active/Passive (IO) which I assume it means it’s currently using for traffic.
There must be some sort of issue with VMWare constantly changing paths or something weird as IO becomes unuseable when I connect the second switch.
This is the Current Design:
HBA1> SWITCH1 > SAN Controller A
SAN Controller B
HBA2> SWITCH2 > SAN Controller A
> SAN Controller B
The Luns show up on both HBA’s of my ESXi with the same names etc. The path policy is set to MRU and the host type on the IBM SAN is set to VMWare.