A note on #vCenter 5.5 #upgrade and the size of the DB partition on the target #VCSA appliance #vexpert

Ran into an interesting issue during a 5.5 to 6.5 migration. During phase 2 of the installation (final configuration of the target VCSA), this would fail with the error:

“Insufficient space on the appliance partition to accommodate all the data for upgrade data option ‘core'”

Digging into the VCSA logbundle, various logs reported an issue with partition sizing

/var/log/vmware/upgrade/upgrade-requirements.log and bootstrap.log noted the following:

2018-05-01T09:07:19.506Z WARNING upgrade_commands The following disks are too small to fit the source disk requirements (vmkdId: required total GB, available GB): {‘db’: (27.412000000000003, 25.0)}
2018-05-01T09:07:19.506Z ERROR upgrade_commands Current deployment size is too small for the existing inventory.

An odd thing to note, the GUI error references the word ‘core’ , but I assume it is not referring to the mount point /storage/core, because as you can see from the above logs, it was very clearly the DB import requirement that failed.

In one of the upgrade logs, you can also find a summery of what the upgrade process has found it needs in the form of disk sizing, based on your source vCenter.

You can find this in \var\log\vmware\upgrade\UpgradeRunner.log

An overview of our appliance mounts after it had failed the upgrade:

As you can see from the sizing, we had chosen the ‘medium’ deployment size during Phase1

First of all, it may be wise to make sure your source DB is as small as it can be, for options on this, see:
https://www.rudimartinsen.com/2017/06/27/vsphere-6-5-upgrade-vcenter-server-check-your-database-size/
https://blog.mwpreston.net/2017/03/21/vcsa-migration-only-allows-for-large-deployment/

Tips on how to clean up the VCSA 6.0 appliance prior t o upgrade are here: https://www.virtualizationhowto.com/2017/08/vmware-vcsa-6-5-upgrade-no-tiny-small-option/ and here: https://kb.vmware.com/s/article/2148587

But regardless, note that most of the partition sizes for your target VCSA are not always effected by the ‘sizing’ choice you can make at the start of Phase1.
For a great overview of this, see: https://www.virtual-allan.com/vcsa-6-5-disks-sizes/

In this case, the sizing of ‘large’  was not enough to accommodate our database, which was just barely too large.

Up to ‘large’ deployment, this disk is sized to 25gb, while ‘x-large’ make it 50gb. Which, in my opinion, is a bit conservative,

But we didn’t want to deploy the entire the VCSA as ‘large’ or ‘x-large’ either.

As William points out on his blog ( https://www.virtuallyghetto.com/2016/11/updates-to-vmdk-partitions-disk-resizing-in-vcsa-6-5.html ) , its quite easy to resize individual disks to suit your needs.

In our case, after Phase1 was completed with the deployment of the OVA, we then resized just the disk for the database, VMDK6, to 50GB, and then we saw Phase2 would complete successfully.

Btw, Another great way to see the relationship between the appliance VMDK disks and the mount points is wuth the ‘lsblk’ command, as described here: http://www.vmwarearena.com/understanding-vcsa-6-5-vmdk-partitions-mount-points/

 

 


One Comments “A note on #vCenter 5.5 #upgrade and the size of the DB partition on the target #VCSA appliance #vexpert

Leave a Reply

Your email address will not be published. Required fields are marked *