A few days ago I noticed a warning message appearing in vCenter Server pointing to some issues with the VMware Common Logging Service.
The service status was showing that the disk space had been filling in steadily reaching the 30% warning threshold.
Considering the infrastructure had not experienced significant changes, I decided to postpone disk space extension and try to find the root cause of this problem.
With the help of the du command, it has become clear some of the subfolders in /var/log/vmware were quite large in size.
It shouldn’t be a problem if the log rotation happens and old data is removed from the disk. However, the /storage/log/vmware/cloudvm/cloudvm-ram-size.log file size was 1.4G, and it seemed to be increasing without log rotation.
An attempt to find out about the cloudvm-ram-size.log file pointed me to the article which William Lam wrote in early 2015 – apparently it logs activities of a built-in dynamic memory reconfiguration process called cloudvm-ram-size.
The issue is documented in ‘Log rotation of the cloudvm-ram-size.log file is not working (2147261)‘ in VMware Knowledge Base.
As per that article, the problem is fixed in vSphere 6.5. For vSphere 6.0, you need to configure log rotation for the cloudvm-ram-size.log file and run the logrotate command manually to archive it to cloudvm-ram-size.log-xxxxxxxxx-.bz2 file.
VMware recommends to do periodic cleanup of older .bz2 files in the /storage/log/vmware/cloudvm location!!! This can be done by adding a rotate parameter to the configuration file as follows:
create 0660 root cis
It was a quick fix!