Error 28: No space on device, While upgrading ESXi host

Products

• VMware vSphere ESXi

Issue/Introduction

During ESXi upgrade via the console or SSH, the following error is displayed:

[InstallationError]
[Errno 28] No space left on device
vibs = VMware_bootbank_esx-base_version
Please refer to the log file for more details.

However, running the df -h command shows that there is sufficient free space on the visible volumes, contradicting the error message.

Environment

VMware vSphere ESXi

Cause

There may not be enough space to accommodate all the VIBs required for the upgrade to the target ESXi version. In some cases, the issue may be due to insufficient inodes rather than a lack of visible disk space.

Resolution

Solution 1: Remove Large Log Files and Core Dumps

Log files and core dumps can occupy hundreds of megabytes on VMFS volumes. You can locate large files (over 50 MB) with the following command and remove unnecessary ones:

# find / -path "/vmfs" -prune -o -type f -size +50000k -exec ls -l '{}' \;

Solution 2: Check for Inode Exhaustion

The error might be caused by a lack of available inodes (file objects) on the VMFS volume. A single VMFS volume can support up to 640,000 inodes.

• To check inode usage, run:

df -i

Example output:

Filesystem   Inodes   IUsed    IFree   IUse% Mounted on
/dev/sda1    942080   507361   434719    54%  /

• Alternatively, you can use:

stat -f /

If inode usage is at or near 100%, you need to free up files to reduce inode consumption.

Solution 3: Increase Available RAM or Configure Swap (Most Probable Cause)

Insufficient free RAM on the ESXi host can cause this error.

You can configure a temporary swap partition on a connected datastore to mitigate this:

1. In the vSphere Client, select the ESXi host and navigate to Configure > System > Swap.
2. Click Edit.
3. Select Can use datastore and specify the datastore where the swap file will reside.
4. Retry the ESXi upgrade/update.