The other day when I was building a vSphere 6.0 environment up in my lab for testing I ran into an issue where performance was extremely slow in the web client and I was continually receiving an error that the VMware-dataservice-sca and vsphere-client status would change from green to yellow. When I deployed the VCSA/PSC appliance I choose “Tiny” as the size option. Even though my implementation is going to be under the 10 hosts and 100 VMs, I think this build was just not enough, and performance in the web client was just really lacking. Searching the VMware KB I came across 2144950. I found out this is a known issue affecting vCenter Server 6.0. Here are the steps that I used to work around the error and gain performance back in the web client.
First I added additional RAM to the appliance. Pretty straight forward, no magic there. Then I used SSH to connect to the appliance and ran the follow command:
cloudvm-ram-size -C XXX vsphere-client
Replace the XXX with the size in MB that you want to increase the heap size.
If you are running a Windows vCenter Server, find C:\ProgramFiles\VMware\vCenter Server\visl-integration\usr\sbin\cloudvm-ram-size.bat and run this command:
cloudvm-ram-size.bat -C XXX vspherewebclientsvc
Again swap out the XXX with the size in MB that you want to increase the heap size. Don’t forget to restart the vSphere client service.
The other day I’m bring up another vSphere 6.0 environment for our VDI team in our engineering test lab and for some reason im having all sorts of issues. I’m installing a VCSA with embedded PSC and connecting it to and existing SSO domain. I have no idea whats going on, its going horrible. One time the install will fail, then the next it will complete, but enhanced linked mode is just acting weird….Well unbeknownst to me the QIP team decided to cut over DNS to new appliances and that was reeking havoc across the environment. So now that I’ve killed (I kid) the guy who was doing this I’m left with a mess to clean up. Finally DNS is working properly so I’m going to re-deploy the PSC/VCSA again but before I do that, I have to clean up the one that I don’t want anymore. Lucky for us its a pretty easy job.
The first step I had to do was make sure that my appliance was powered down. I knew that there was no other VCSA pointing to this PSC. If you are unsure if any other vCenter is connected to the PSC you are removing, you can check by logging into the vSphere web client and go to the advanced vCenter server settings and look for a property called config.vpxd.sso.admin.url and the value of this setting is the PSC the vCenter server is using. If you find any other vCenters VMware has KB 2113917 to help you re point your vCenter to a different PSC.
Once that is all sorted out, next we need to connect to another PSC is the same SSO domain via SSH and run the following command:
cmsso-util unregister --node-pnid Platform_Services_Controller_FQDN
--username administrator@your_domain_name --passwd vCenter_Single_Sign_On_password
After that completes, delete the appliance from your inventory and check in Administration -> System Configration -> Nodes to make sure that its not listed there.
Removing a VCSA is just about the same as above just have to make one change in the command:
cmsso-util unregister --node-pnid vCenterServer_System_Name --username
administrator@your_domain_name --passwd vCenter_Single_Sign_On_password
If you need some additional info on these steps, check out KB 2106736