ESXi hosts can be updated by manually downloading the patch ZIP file from the VMware download page and installing the VIB by using the esxcli software vib command. Additionally, the system can be updated using the image profile and the esxcli software profile command.
If a virtual machine has an SEsparse snapshot and if the base VMDK file size is not 4K multiple, when you query for the physical layout of the VMDK from the guest operating system or third-party applications, a physical CPU lockup might occur. As a result, the ESXi fails with a purple diagnostic screen.
The vSphere Web Client or vSphere Client might display a host profile as compliant even when it has a non-compliant value such as the size of the maximum transmission unit (MTU). Remediation of the profile works as expected, but it is not reflected in the vSphere Web Client or vSphere Client.
This issue is specific to vSphere Virtual Volumes datastores when a VMDK file is assigned to different SCSI targets across snapshots. The lock file of the VMDK is reassigned across different snapshots and might be incorrectly deleted when you revert the virtual machine to a snapshot. Due to the missing lock file, the disk does not open, and the virtual machine fails to power on.
This issue is resolved in this release. The fix makes these ESXCLI configurations persistent across reboots by writing to the ESXi configuration file. Ring-size configuration must be allowed on physical NICs before using the CLI.
If you modify the NetworkNicTeamingPolicy or SecurityPolicy of a host profile and change it to the default settings, you might fail to apply the host profile to an ESXi host and receive an error such as: Error: A specified parameter was not correct.
Virtual machines configured with a maximum number of virtual devices along with PCI passthrough might fail to power on if the limit is exceeded. The issue is identified with a panic log in the vmware.log file similar to: vmx| E105: PANIC: VERIFY bora/devices/pci/pci.c:1529.
The hostd service sometimes fails while recomposing VDI Pools, with the error message VmfsExtentCommonOpen: VmfsExtentParseExtentLine failed. You can see hostd-worker-zdump files generated in the /var/core/ directory of the ESXi host.
Due to a memory problem in vSphere Storage I/O Control, an ESXi host might stop responding and the /var/log/vmkernel.log file contains the following: MemSchedAdmit: 470: Admission failure in path: sioc/storageRM.13011002/uw.13011002 MemSchedAdmit: 477: uw.13011002 (68246725) extraMin/extraFromParent: 256/256, sioc (840) childEmin/eMinLimit: 14059/14080
In certain cases, host profiles on ESXi 6.7 might not support PEERDNS from vmknics that are attached to an NSX-T based opaque network. The issue is specific to a scenario when you navigate to Host > Configure > TCP/IP configuration > Default > DNS configuration > Obtain settings automatically from a VMkernel network adapter. If you select a vmknic that is connected to a NSX-T based opaque network, host profiles extracted from this host do not work. Such host profiles lack the Portgroup backing the vmkernel network adapter to be used for DNS policy in the DNS subprofile. Editing the host profile gives a validation error.
When a source Windows vCenter Server 6.0.x or 6.5.x contains vCenter Server 5.5.x host profiles named with non-ASCII or high-ASCII characters, UpgradeRunner fails to start during the upgrade pre-check process.
Workaround: Before upgrading Windows vCenter Server 6.0.x or 6.5.x to vCenter Server 6.7, upgrade the ESXi 5.5.x with the non-ASCII or high-ASCII named host profiles to ESXi 6.0.x or 6.5.x, then update the host profile from the upgraded host by clicking Copy setting from the hosts.
When you run the camregister command with the -x file option, for example, to register the vSphere Authentication Proxy, the process fails with an access denied error when the vCenter Single Sign-On password contains non-ASCII characters.
vCenter Server 6.7 does not support host profiles with version less than 6.0. To upgrade to vCenter Server 6.7, you must first upgrade the host profiles to version 6.0 or later, if you have any of the following components:
Due to changes in the default values in the /etc/ssh/sshd_config file, the vCenter Server 6.7 upgrade replaces any manual edits to this configuration file with the default configuration. This change was necessary as some prior settings (for example, permitted ciphers) are no longer compatible with current ESXi behavior, and prevented SSHD (SSH daemon) from starting correctly.
Workaround: If edits to /etc/ssh/sshd_config are needed, you can apply them after successfully completing the vCenter Server 6.7 upgrade. The default configuration file now contains a version number. Preserve the version number to avoid overwriting the file.
If PeerDNS for IPv4 is enabled for a vmknic on a stateless host that has an associated host profile, the iPv6PeerDNS might appear with a different state in the extracted host profile after the host reboots.
The vSphere Client does not support selecting vService extensions in the Deploy OVF Template wizard. As a result, if an OVF virtual appliance uses vService extensions and you use the vSphere Client to deploy the OVF file, the deployment succeeds, but the virtual appliance fails to start.
When an ESXi host provisioned with vSphere Auto Deploy reboots, it loses the previously set numRxQueue value. The Host Profiles feature does not support saving the numRxQueue value after the host reboots.
ESXi provides the batch QueryUnresolvedVmfsVolume API, so that you can query and list unresolved VMFS volumes or LUN snapshots. You can then use other batch APIs to perform operations, such as resignaturing specific unresolved VMFS volumes. By default, when the API QueryUnresolvedVmfsVolume is invoked on a host, the system performs an additional filesystem liveness check for all unresolved volumes that are found. The liveness check detects whether the specified LUN is mounted on other hosts, whether an active VMFS heartbeat is in progress, or if there is any filesystem activity. This operation is time consuming and requires at least 16 seconds per LUN. As a result, when your environment has a large number of snapshot LUNs, the query and listing operation might take significant time. 2b1af7f3a8