Manually Tagging SSD or NVMe Devices as Flash in vSphere ESXi
search cancel

Manually Tagging SSD or NVMe Devices as Flash in vSphere ESXi

book

Article ID: 341618

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

Symptom:

  • Storage devices that are physically NVMe or SSD are listed with a "Drive Type" of HDD in vCenter.

  • This misidentification can lead to performance degradation, high latency at the Virtual Machine (VM) level, and may block application or OS upgrades (e.g., Red Hat OpenShift Data Foundation) that require SSD/Flash-backed storage.

Environment

  • VMware vSphere ESXi 8.x
  • Storage Hardware: HPE XP8 (OPEN-V), HPE 3PAR, or other NVMe/SSD architectures requiring manual flash tagging.

Cause

ESXi utilizes heuristic identification to determine storage media types. If a storage vendor does not support automatic flash disk detection, or if there is no explicit Storage Array Type Plugin (SATP) claim rule for the specific array framework (e.g., HPE OPEN-V), ESXi defaults to identifying the device as standard HDD. Without a declarative rule instructing the Native Multipathing Plugin (NMP) to append the enable_ssd flag, the hypervisor fails to correctly enumerate the architecture as flash media.

Resolution

To enable the SSD option on SSD based disks/LUNs that are not detected as SSD by default on vCenter:
 

  1. Verify that the device is not in use. Ensure the virtual machine(s) accessing the device are powered off and if it contains a mounted datastore, ensure that the datastore is unmounted. For more information on the unmounting operation, see How to unmount a LUN or detach a datastore device from ESXi hosts.

  2. Browse to the host in the vSphere Client object navigator.

  3. Click the Configure tab.

  4. Under Storage, click Storage Devices.

  5. From the list of storage devices, select one or several HDD devices to mark as flash devices and click the Mark as Flash Disks icon.

    In case you get an error - "Cannot change the host configuration. Cannot mark disk naa.######################### as "Flash". "Unable to reconfigure the disk claim rules. The disk is in use." ", unmount the datastore from all the hosts and then use "Mark as Flash Disks".

  6. Click Yes to save your changes.
 
 
 
To enable the SSD option on SSD based disks/LUNs that are not detected as SSD by default on CLI:
  1. Log in to the host as root, using either the Console DCUI or SSH.
  2. If the device is currently in use, ensure the virtual machine(s) accessing the device are powered off and if it contains a mounted datastore, ensure that the datastore is unmounted. For more information on the unmounting operation, see How to unmount a LUN or detach a datastore device from ESXi hosts.

  3. Run this command and identify the device to be tagged and its SATP:

    # esxcli storage nmp device list

    You see output similar to:

    naa.#########################
    Device Display Name: DGC Fibre Channel Disk (naa.#########################)
    Storage Array Type: VMW_SATP_CX
    Storage Array Type Device Config: {navireg ipfilter}
    Path Selection Policy: VMW_PSP_MRU
    Path Selection Policy Device Config: Current Path=vmhba4:C0:T0:L25
    Working Paths: vmhba4:C0:T0:L25

  4. In this sample output, the SATP is VMW_SATP_CX. This may vary depending on your environment. For example, the SATP may be VMW_SATP_LOCAL. Ensure to use the correct SATP in these steps.

    Notes:
    • In this procedure, the device ID naa.######################### is used. Use the device ID returned for your own device in these steps.
    • If you have the SSD drives in the RAID 1 setup, and the RAID controller you are using does not support NON-RAID mode or Pass-Through Mode, then you must re-configure your SSD drives in a RAID 0 setup.

  5. Run this command to add a PSA claim rule to mark the device as SSD:

    # esxcli storage nmp satp rule add --satp=SATP_TYPE --device <naa device ID> --option "enable_ssd"

    For example:

    # esxcli storage nmp satp rule add --satp=VMW_SATP_CX --device naa.######################### --option "enable_ssd"

    Note: Run this command to enable local and SSD at the same time:

    # esxcli storage nmp satp rule add –-satp=VMW_SATP_LOCAL –-device <naa device ID> --option "enable_local enable_ssd"

    For example:

    # esxcli storage nmp satp rule add --satp=VMW_SATP_LOCAL --device naa.######################### --option "enable_local enable_ssd"

    You must restart the ESXi host for the changes to take effect.

  6. Run this command to unclaim the device:

    # esxcli storage core claiming unclaim --type=device --device <naa device ID>

    Note: If you have a local SSD device, the device might not be detected after you apply the rule to unclaim the device. After you restart the ESXi host, you can verify that the device has the rule applied.

  7. Run these commands to reclaim the device:

    # esxcli storage core claimrule load
    # esxcli storage core claimrule run
    # esxcli storage core claiming reclaim -d <naa device ID>

  8. Run this command to check the SSD status:

    # esxcli storage core device list -d <naa device ID> | grep SSD

    You see output similar to:

    <naa device ID>
    Display Name: DGC Fibre Channel Disk (<naa device ID>)
    Size: 204800
    Device Type: Direct-Access
    Multipath Plugin: NMP
    Devfs Path: /vmfs/devices/disks/<naa device ID>
    Vendor: DGC
    Model: RAID 5
    Revision: 0220
    SCSI Level: 4
    Is Pseudo: false
    Status: on
    Is RDM Capable: true
    Is Local: false
    Is Removable: false
    Is SSD: true
    Is Thin Provisioned: unsupported
    Attached Filters:
    VAAI Status: unknown
    Other UIDs:
    <vml device ID>

Notes:

    • You can use a generic vendor model notation as an alternative to device/LUN. For example:

      # esxcli storage nmp satp rule add --satp VMW_SATP_CX --vendor "foo" --model "bar" --option "enable_ssd"

    • The SATP change is not required for the device to function.
    • To untag the device, you must perform the entire process in reverse; you must remove the claimrule, unclaim the device, load the claimrules again, and then reclaim.
    • You must restart the ESXi host to tag the local datastore (if ESXi is installed locally) as the SSD drive.


Additional Information

  • When adding a new rule, if you receive an error stating that there was a duplicate rule found, you may need to remove the rule and add it again with the correct SSD flag. For example:
# esxcli storage nmp satp rule add --satp VMW_SATP_LOCAL --device <naa device ID>
Error adding SATP user rule: Duplicate user rule found for SATP VMW_SATP_LOCAL matching device <naa device ID> PSP and PSP Options
  • Remove the existing rule, and then re-add:
# esxcli storage nmp satp rule remove --satp VMW_SATP_LOCAL --device <naa device ID>

# esxcli storage nmp satp rule add --satp VMW_SATP_LOCAL --device <naa device ID> --option "enable_ssd"
  • Run this command to verify the added VMW_SATP_LOCAL satp rules:

    # esxcli storage nmp satp rule list --satp=VMW_SATP_LOCAL
  • When a disk is marked as a flash disk, vSphere primarily tries to use it as a cache for the datastore. Since SSDs are much faster than hard disks it improves the performance of the datastore. If you mark a hard disk as a flash disk, the behaviour would be the same, but since a hard disk can't deliver the same performance, you'll notice degraded performance on the datastore. So make sure to mark the devices only if you are certain that they are flash devices.
Powered by Wolken Software