This article describes a scenario where applied Native Multipathing Policy (NMP) doesn't remain persistent across reboot.
VMware vSphere ESXi 6.x
VMware vSphere ESXi 7.x
VMware vSphere ESXi 8.x
Issue occurs due to claim rules injected by third party. If the third party multipathing policy is having a system generated satp rule on the ESXi then we cannot change it to NMP policy.
1. Initial Multipathing Policy
Storage devices were initially configured to use a third-party multipathing policy.
2. Validate the Current Multipathing Policy
Before making any changes, verify if a third-party multipathing policy is listed in the SATP rule list by running the following command: esxcli storage nmp satp rule list | less
Note: System-generated rules cannot be removed.
3. Change Multipathing Policy to Round Robin
Manually set the multipathing policy to Round Robin and reboot the host.
4. Post-Reboot Logs
After rebooting, while reviewing the vmkernel.log , you may observe logging that indicates conflicting claim rules are set for the device.
5. Involve Hardware Vendor
Please reach out to your hardware vendor to check the issue further. If required, uninstall NCM (Nimble Connection Manager) and revalidate the multipathing policy.
Note: In the above example we worked on Nimble storage array, but the same pattern can be validated on other arrays.