Reduce amount of ESXi logs

We were looking into amount of ESXi logs we were collecting and we discovered that two applications in ESXi were on verbose logging level although we had set “config.HostAgent.log.level” to info. Those applications were rhttpproxy and fdm. They we generating millions of lines per day.


To reduce rhttpproxy log level you need to edit /etc/vmware/rhttpproxy/config.xml and replace the “verbose” value in log level section with “info” for example. After this restart the rhttpproxy service.

fdm (HA agent)

To change HA agent log level you need to modify the cluster advanced settings and add option “das.config.log.level” with value “info”. After this disable High Availability on the cluster and reenable the High Availability.

Powercli lines to do this:
New-AdvancedSetting -Entity $cluster -Type ClusterHA -Name ‘das.config.log.level’ -Value info
Set-Cluster -Cluster $cluster -HAEnabled:$false
Set-Cluster -Cluster $cluster -HAEnabled:$true

NTP service not starting on ESXi 7 after restart.

We noticed that NTP service is not starting after ESXi 7 patching although it’s configured to “Start and Stop with host”. According to VMware KB article (link) there is no fix for this issue at the moment.

I wrote a small script which I run every 5 minutes to check and start NTP service if it’s stopped.

$esx_hosts = Get-VMHost -State Maintenance,Connected
$esx_hosts | Get-VMHostService | Where-Object {$_.key -eq “ntpd” -and $_.Running -ne “True”} | Start-VMHostService 

Scan entity (Check Compliance) task takes a long time.

Recently when we were solving the USB/SD-card boot issue (link) we noticed that “Scan entity” tasks on some hosts took more than 15 minutes. In VMware Community there is a thread describing similar issue – One post there mentioned that the problem only exists on hosts with FC storage.

Based on that information we did a test and results were following. Scan entity task took about 16 minutes on a host with 100+ FC LUNs. Scan entity task took about 1 minute on a same host with all FC LUNs disconnected (FC ports disabled). We also tested on host with about 10 LUNs – scan entity tasks took about 3 minutes. I have created a case to VMware to understand how is LUN count related with scanning for patches.