When an ESXi host crashes, the generated core dump can be sent over the network to a remote core dump server. The vCenter Server Appliance (VCSA) can be configured as a destination by enabling the built-in vSphere Dump Collector service. By default, the storage repository for the vSphere Dump Collector service is configured to 2GB and can be increased up to 10GB, which is the supported maximum.
In earlier vSphere releases, configuring the vSphere Dump Collector repository size could be done using the vSphere Web (Flash) Client as shown in the screenshot below.
However, as part of the vSphere (HTML5) UI Client transition, it looks like all general VCSA service configurations including the vSphere Dump Collector was not ported over. I can only assume the reasoning for this was due to the infrequency configuring or updating these settings. This has been a question that has come up a few times across both VCSA 6.7 and 7.0 environments and I was recently reminded of the solution and I realized I had not blogged about the solution.
Step 1 - SSH to your VCSA and then edit /etc/sysconfig/netdumper and update the NETDUMPER_DIR_MAX_GB parameter to the desired value (10GB Max)
Step 2 - Restart the vSphere Dump Collector service for changes to go into effect
service-control --restart vmware-netdumper
service-control --stop vmware-netdumper
service-control --start vmware-netdumper
mr rosh says
Richard Hughes says
Is there a down side to using VCSA as a dump collector? Can the drive fill up? Is there an alternative to the VCSA dump collector like a third party logging service?
Matt Mancini says
Great write up as deploying VCSA on Fusion/Workstation can be a bit tricky. Another option is to deployed VCSA on an ESXi host or nested host first then just migrate it to fusion or Workstation. I know not all users can support this option as it can be a bit resource intense but I've done this a few times when I'm rebuilding my home lab and it worked fine.