Among the various features released for Horizon 8, the most interesting one is Agent Auto Upgrade:
“The agent auto upgrade feature allows customers to automatically initiate upgrades without manual intervention. To utilize this feature, on-premises systems must have access to CDS servers. Customers without CDS access can establish their webserver, host the agent components, and then register the agent build with the connection server to upgrade agents in VDI/RDSH desktops. This feature requires Horizon Plus or Horizon Universal License, and is available for Full Clone Desktops and RDSH Servers only. To upgrade Horizon Agent in Instant Clone Desktop Pools or RDS Farms, upgrade Horizon Agent on the Golden Image and schedule maintenance to push the new image.”
During the maintenance and updating of the Horizon Connection server components, one aspect is the necessary wait for the connection servers to correctly resume responding on TCP port 443. During one of the many activities on Horizon my customer created a simple and effective door test script. Even though it is very simple and intuitive, I want to share the code with you:
$check = netstat -ano | findstr 0.0.0.0:443
"Waiting 5 seconds and retry"
} while (!$check)
In recent days, a customer reported an anomaly on an HPE SimpliVity cluster hosting instant clone Horizon VDIs. In detail:
vSphere with seven hosts present, two were always at 98% CPU utilization and 90% RAM utilization.
Continuous vMotion generated by the VM DRS to and from those two HOSTS.
After a careful analysis, we identified that there were no problems at the vSphere infrastructure level. The issue was due to a Simplivity feature called IWO.
By disabling IWO and keeping DRS active (Full automatic) I have an optimal balance of CPU and RAM load between hosts at the expense of a slight increase in I/O trip times
Scenario – Even VM Load Distribution
I want even VM load across my cluster in terms of CPU and memory. Data locality and I/O performance are not top priorities. Most applications are CPU and memory intensive, and adding 1ms to 2ms to I/O trip times will not impact application performance.
In this scenario, IWO can be disabled thus ensuring no DRS affinity rules are populated into vCenter server. Suppressing DRS affinity rules will allow VMware DRS or allow you to directly distribute VMs across the cluster as desired to ensure all VMs are adequately resourced in terms of CPU and memory. The ‘Data Access Not Optimized’ alarm can be suppressed within vCenter server.
Many times I found myself having to demonstrate that the communication between the Unified Access Gateway and the Connection Servers was not working due to problems with poorly configured firewall rules. A very useful test is to connect to the UAG console and launch the classic CURL command:
curl -v -k https://<FQDN or IP ADDRESS CS>:443/
the outcome of which is as follows if the connection is ok (HTML output)
or the following if the connection is not enabled on the firewall