Working on a vSphere 5.5 to 6 upgrade and when attempting to re-mediate the host to 6 using HP’s custom ESXi version, received a failure due to “Cannot Create A Ramdisk of size 385MB to store upgrade image”. After doing some investigation, appears some of the VMware default partitioning does not leave enough space for the upgrade image due to the image’s size especially in HP’s case. Environment was all HP G8 servers running HP’s custom ESXi base. A quick check of the partitions confirmed the available space was smaller than the image needed, couple of different ways to attack this, I chose the following. Easiest method is to find a datastore we can move the default scratch space to that has more room which is the method I chose.
- Need to identify the ID of the datastore you want to use, easiest way is either via GUI in Manage—>Storage or CLI on host run: esxcli storage vmfs extent list
I chose to create a Scratch directory under the datastore to keep things clean.
- Log in to vCenter Server using the vSphere Web Client/GUI, click Hosts and Clusters, then select the specific host. Click the Manage tab and then click Settings. Click System and then click Advanced System Settings.
Click Edit and add the path to the scratch directory. The path will look something like this, replace red ID with the datastore ID from step 1.
- Once you have this set, reboot the host for the changes to take effect.
- Once host is back online should be able to re-mediate the host without any further issues.
I have so far seen this on HP’s custom images but would imagine this would happen with any custom ESXi install and/or plugins that exceed VMware default sizes.