Despite being an appliance there is still some manual labour to do on the vCenter Server Appliance part of vSphere 6.5. The relative small number of hosts managed by the vCenter Server instance of my customer had a full disk twice in a small time frame. Luckily VMware decided to split the location of logfiles over several virtual disks and as a result vCenter Services were not affected.
In a serie of blogposts I will explain how to manage the various logfiles and dumpfiles.
In part 1 I will explain how to get rid of the Java performance chart files which are stored in /var/log/vmware/perfcharts/ .
This perf charts gets filled up with java dumps because it doesnt have enough memory allocated to its own service. You can dynamically allocate more memeory to the service to prevent the service from crashing and generating dumps:
- To resolve this issue, follow below steps:
Log in to vCenter Server appliance using SSH or console.
- Take a backup of the /etc/vmware/service-layout.mfx file by running this command:
cp service-layout.mfx service-layout.mfx.bak
- Stop the vmware-perfcharts service by running this command
service-control –stop vmware-perfcharts
- Open the service-layout.mfx file using a text editor by running this command:
Change CompressClassSize size from 94 to 64 for the row vmware-perfcharts.
- Save and close the service-layout.mfx file.
- Start the vmware-perfcharts service by running this command:
service-control –start vmware-perfcharts
Now increase the dynamic sizing of vmware-perfcharts by running this command:
- cloudvm-ram-size -C 1024 vmware-perfcharts