ESXi host rolls back to the previous version on the second reboot after upgrading to ESXi 7.0 u2 or newer
search cancel

ESXi host rolls back to the previous version on the second reboot after upgrading to ESXi 7.0 u2 or newer

book

Article ID: 426659

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

  • During an upgrade from ESXi 6.7 to 7.0, or from an older 7.0 release to a newer version of 7.0 or 8, the following behavior may occur:
    • The host boots successfully as the new version immediately after the upgrade.
    • Upon the second reboot, the system automatically reverts to the previous version.
  • When this issue occurs, the /var/log/jumpstart-native-stdout.log file contains the following error sequence recorded during the first boot after the upgrade:

ValueError: eui.########: Not enough disk space available to create ESX boot partitions (available=####MiB < required=3814MiB)

  • Running /usr/lib/vmware/vm-support/bin/partedUtil.sh to check the system storage partition configuration yields the following results:

Device: /vmfs/devices/disks/eui.########
Partition table:
gpt
#### ### ## ########
1 64 8191 ############################### systemPartition 128
5 8224 520191 ############################### linuxNative 0
6 520224 1032191 ############################## linuxNative 0
7 1032224 1257471 ############################### vmkDiagnostic 0
8 1257504 1843199 ############################## linuxNative 0
9 1843200 7086079 ################################## linuxNative 0
2 7086080 ######## ################################## vmfs 0 
3 ######## ######## ################################## vmfs 0

Of the partitions listed, 1, 5, 6, 7, 8, and 9 are partitions created automatically during ESXi installation.
Partitions 2 and 3 are partitions corresponding to manually created vmfs datastores.
The starting sector of partition 2 is 7086080, which means the contiguous free space preceding partition 2 is less than the capacity required for partition reconfiguration during the upgrade.
Additionally, a configuration with two vmfs partitions on a single device is not supported.

 

Environment

ESXi 7.0 and 8.0

Cause

ESXi 7.0 introduces a significant change to the system storage layout. During the upgrade process, the installer attempts to reconfigure existing partitions to accommodate this new layout.

However, if there is insufficient contiguous free space to create the new system partitions, the repartitioning process fails. Consequently, the system rolls back to the previous version upon the next reboot. 

Resolution

Perform a clean installation of ESXi to allow the installer to correctly reconfigure the system storage partitions.

Using the default partition configuration ensures that sufficient space is available for the repartitioning to succeed. Additionally, when reinstalling, avoid creating a datastore on the system storage device.

Note:
Reinstalling will erase the existing host configuration and data on the local datastore. Be sure to back up your data before proceeding.
For instructions on how to back up your ESXi host, see the following KB article:

How to back up and restore the ESXi host configuration

Additional Information

This behavior can also occur when the system BIOS on the host is set to legacy boot mode rather than EFI, which is a requirement in newer ESXi releases. This is less common and may need to be addressed by collaborating with the hardware vendor to ensure that the boot mode is correct and the boot media is properly configured and accessible.