VxRail: Node Health-check Fails for Test scratch

The 'scratch' health-check makes sure that logs in "vmkernel.log" can be accessed in the scratch partition:

/var/log/vmkernel.log

The results of this health-check can be one of the following:

Each test that passes is not listed in the summary report, for ease of reading.
An example of the health-check output is shown below:

#========================#======#=========#====================================================================#==============#
|  Hostname / Category   |Status  Dell_KB |  Warnings or Failures, unless tests Passed                         ; Product S.N. |
#========================#======#=========#====================================================================#==============#
| node02                 | Critical 43145 | scratch: /scratch/log/vmkernel.log has not been written in the last 486096 seconds|

The 'scratch' test verifies that the file '/scratch/log/vmkernel.log' can be accessed.
The timestamps on the most recent lines are checked, and an error are reported if the most recent lines are over 2 hours old.
If for any reason the vmkernel.log cannot be found in the scratch partition, the test reports a critical failure.

Check the VMkernel log:

A failure of this test indicates that the Scratch partition was not readable by the test and this partition should be checked.
Access the command line on a node and check the scratch partition:

/scratch

If the partition can be accessed, check the contents of the following log, which should contain recent lines of events:

/scratch/log/vmkernel.log

The scratch log location may have been changed, and the logging location can be checked using the command:

esxcli system syslog config get

For example:

Local Log Output: /scratch/log

If the above is not /scratch/log, check the vmkernel.log in the alternative path.

Check the VMkernel link:

There must be a link to the scratch log (for example, /scratch/log/vmkernel.log), in /var/log:

# ls -la /var/log/vmkernel.log
lrwxrwxrwx    1 root     root            25 Sep  8 12:01 /var/log/vmkernel.log -> /scratch/log/vmkernel.log

If this is not present, create the file link to match the entry above:

ln -s /scratch/log/vmkernel.log /var/log/vmkernel.log

To fix a lack of logging to vmkernel.log:

If the vmkernel.log is not being written to (all entries are over 2 hours), the host should be rebooted.

To Reboot:

1. Put node in Maintenance Mode
2. Reboot the node
3. Exit node from Maintenance Mode
4. Run VxVerify again

If the issue is still present contact Dell Support.

Another possibility is that the cluster has a custom global directory value set in advanced system settings for Syslog.global.logDir (which is outside of the VxRail standards).
This can be a network share, or a syslog server misconfiguration.

For example, a customer can customize the Syslog.global.logDir in all nodes to save the logs in a Network File System (NFS), like the example below.
 

Check the variable Syslog.global.logDir, under Advanced System Settings of the node, and confirm that the value is []/scratch/log.
If the value is different from default []/scratch/log, the health-check can report that VMkernel.log is not found.
More information about the variables above can be located on the VMware article below.
https://knowledge.broadcom.com/external/article?legacyId=2003322