Loading...
Loading...
Prior to HPE Alletra Storage MP B10000 version 10.5.x, the HPE Alletra Storage MP B10000 did not specify any limits for NVMe deallocate request sizes. Consequently, hosts could send very large NVMe deallocate requests, sometimes as large as 2GB or more.However, these oversized requests can occasionally cause request timeouts. To mitigate this issue, starting with release 10.5.0, the HPE Alletra Storage MP B10000 will send a maximum NVMe deallocate request size of 32MB in the NVMe identify command to NVMe over Fibre Channel (FC-NVMe) or NVMe over TCP (NVMe/TCP) hosts.Consequently, specific ESXi advanced configuration parameters must be set and enabled on the ESXi host. Failure to set and enable these parameters will result in any deallocate requests that exceed the maximum supported size to be rejected by the HPE Alletra Storage MP B10000 array. This causes the array to not reclaim the space associated with the deallocation for general namespace (volume) usage, leaving it stranded within the current namespace.While this space can still be used for data within the same namespace (volume), it cannot be allocated to another volume. As a result, storage space is not freed, leading to increased capacity utilization.
HPE Alletra Storage MP B10000 running firmware version 10.5.x (or earlier) with one or more VMware ESXi server hosts attached with NVMe over Fibre Channel (FC-NVMe) or NVMe over TCP (NVMe/TCP).Operating Systems affected:VMware ESXi 7.xVMware ESXi 8.x
This issue will be corrected in a future firmware release for HPE Alletra Storage MP. This advisory will be updated when a solution is available.As a workaround, apply the following settings for every ESXi host connected over NVMe SSDs over FC or TCP (NVMeoFC or NVMeoTCP) before upgrading to array firmware version 10.5.0, and when adding new ESXi hosts with NVMe connections to an array running 10.5.0VMware ESXi 7.xesxcli system settings advanced set -o /Scsi/NvmeMaxUnmapLbaCount 65536esxcli system settings advanced set -o /Scsi/NvmeMaxUnmapBlockDescriptorCount 10Reboot the hostConfirm the settings have been applied:esxcfg-advcfg -g /Scsi/NvmeMaxUnmapLbaCountValue of NvmeMaxUnmapLbaCount is 65536esxcfg-advcfg -g /Scsi/NvmeMaxUnmapBlockDescriptorCountValue of NvmeMaxUnmapBlockDescriptorCount is 10VMware ESXi 8.xesxcli system settings advanced set -o /Scsi/NvmeUseDsmTp4040 1Reboot the hostConfirm the settings have been appliedesxcfg-advcfg -g /Scsi/NvmeUseDsmTp4040Value of NvmeUseDsmTp4040 is 1After the changes have been completed to all ESXi hosts, log a support ticket stating that the ESXi host changes were made and that the array is ready for version 10.5.0.Once the changes have been made on the ESXi hosts, contact HPE support to have the block for upgrade to 10.5.0 released.
Operating Systems Affected:Not Applicable
Click on a version to see all relevant bugs
Hewlett Packard Enterprise Integration
Learn more about where this data comes from
Bug Scrub Advisor
Streamline upgrades with automated vendor bug scrubs
BugZero Enterprise
Wish you caught this bug sooner? Get proactive today.