Early last year we implemented a vCenter 4.0 cluster (ESX 4.0 hosts) connected to a CX4-120 via redundant HBAs/Switches/etc. I recall configuring the Path Selection Policy for each of the orignal VMFS LUNs on each host for Round Robin and confirmed in our documentation.
Today, while trying to diagnose an obvious inbalance across the CX4's SPs, I found that some hosts were now using the NMP FIXED policy for the majority of VMFS LUNs! :smileyshocked: All hosts Fixed for first 5 (original) LUNs & RR all other LUNS (created *after* the 1st five).
- Recently all hosts were fully updated/patched (to 4.0 U3).
- LUNs were migrated from static RAID Groups to FAST Pools
Hoping someone has seen this this within their environments?
The simple answer is someone changed your pathing policies. If you are using an active/active array the default pathing policy would have been fixed.
Also, if your LUNs changed their LUN serial #'s when they moved to the new pools, they would potentially also get defaulted back to FIXED.