I've got a number of datastores setup using the round robin path selection policy in a 3 host cluster. I had been troubleshooting some throughput problems and discovered that most of my datastore's path selection policies had been changed from round robin to fixed. These all exist on an Equallogic iSCSI SAN and other targets. There's only one other administrator and he doesn't typically mess with this stuff. We did have a power outage a while back. I know you're all going to think my organization is stupid, so let me just say I agree! We're a bank and we don't have a generator that powers our datacenter. We have redundant disks, power supplies, switches, servers, etc., but no power backup. Our cluster lives on some APC UPSs that might give it half an hour of power. Well our power went out for 4 hours and everything shut down dirty. Can this cause this sort of thing? For the record, we're looking at APC software for our ESXi cluster. I'm just trying to find out how those policies could have been changed. Is this default behavior? Should I watch out for anything else?
