| Article Number: 530888 || Article Version: 3 || Article Type: Break Fix |
XtremIO Family,XtremIO X1,XtremIO X2
Post upgrade of ESXi running on Cisco UCS blade server customer notes loss of access to LUNs. Specifically LUNs with LUN ID greater than 255.
Within UCS Manager FIbre Channel Adapter Policies there is a configuration for "Max LUNs per Target". By default this value is set to 256.
Cisco recommended configuration for this field is 1024 for ESX or Linux Host Operating Systems.
The application of this configuration appears to vary between the fnic used with ESXi 6.5 ( and prior) and the nfnic used with ESXi 6.7. The nfnic does appear to apply this configuration whereas the fnic used with ESXi 6.5 (and prior) does not.
If you should suspect that this issue has been encountered please confirm the following.
- Are the missing LUNs using LUN ID greater than 255?
- Were the host upgraded to ESXi 6.7 from a prior release?
- Are the host running on a UCS Blade server?
- Review the host vmkernel logs for messages with following signature.
2019-02-06T18:56:48.222Z cpu6:2097702)WARNING: ScsiScan: 2384: Skipping out-of-range LUN vmhba2:C0:T2:L314 Max LUN ID supported by driver is=256
- Confirm in UCS Manager if the FIbre Channel Adapter Policy applied to the host has value for "Max LUNs Per Target" set to 256 ( the default value)
- If all the symptoms within the cause section are confirmed, please adjust the value of "Max LUNs per Target" to 1024 and push the updated policy out to the host. The host will need to reboot to apply the policy and restore access to the LUNs with LUN ID greater than 255.
- It is recommended to open a case with Cisco to track this issue on their side.