I upgraded a server from ESXi 6.0 to 6.5. When I did this, I lost 12 out of my 17 LUNs. In going over the errors in the vmkernel.log I saw the error "Failed to add legacy uid XXXX on path XXXX Already exists" In looking at which volumes did not come back, lo and behold they had two LUN IDs based off the different paths.
Here's an example:
A LUN that comes back has H-LUN ID 00001 across Ports 0A, 0C, 1A, and 1C while a LUN that doesn't come back has H-LUN ID 0006 for Ports 0A and 1C and H-LUN ID 0005 for Ports 0C and 1A.
From all that I have read, all LUN paths need to have the same IDs. For whatever reason ESXi 6.0 was more accommodating of this issue, but 6.5 isn't having any of it.
If I un-present all LUNs to the server, then re-present them, I can get all 17 LUNs back. However, when I reboot the server, only the 5 LUNs come back and I have to un-present/re-present the LUNs to get them back.
I can't really let this situation go on. How can I fix the the existing 12 volumes that have 2 different LUN IDs?