Monitors load of Server OS machines that are delivering desktops and applications to users.
This monitor checks the load of server machines delivering desktops and/or applications to the end users.
Load index can be in the range of 0 to 10000, where 0 means that there is no load and 10000 means that the server is fully loaded. Load index is calculated based on the load management policy settings and is used by the delivery controller when end-users are making new connections.
When users try to launch an application or desktop while all active servers are reporting a full load, a new server will be started and made available for new connections. If there are no servers to start or the machines are put in maintenance mode, then the connection will fail.
The following is a list of default load management policies settings:
CPU usage: Report full load if CPU is 90%, disabled by default
Disk usage: Report 75% load if disk queue length is 8, disabled by default
Memory usage: Report full load if Memory usage is 90%, disabled by default
Memory usage base load: Report zero load if the base operating system's memory usage is below 768 MB, enabled by default
Maximum number of sessions: Limit 100, enabled by default
Concurrent logons tolerance: Target value 2, enabled by default
CPU usage excluded process priority: Exclude processes that have a priority of Below Normal or Low, enabled by default
The following configuration parameters are customizable:
IntervalSeconds | The monitoring period (the default interval is set to 300 seconds). |
SyncTime | Time at which to start time interval (not set by default). |
NumSamples | The number of monitoring periods in which the load index has to be over threshold to trigger a health state change (the default is 1 sample). |
Threshold | The load index threshold (the default is set to 8000). |
Load index on one machine may be high for various reasons, such as:
Load management policy settings
Too few machines in the delivery group
Failure of one or more machines in the delivery group
Abnormal high resource usage of one or more users/applications on the machine
As a consequence of the above root causes, increased load can be observed on all machines in the delivery group. In some cases, high load can be observed also on only one or a few machines (e.g. end-users running defective application).
Identify first if load index is high only on the machine where the issues is reported or also on one or more other machines in the delivery group.
It is also important to find out if load is high on only one or more than one machine. Please check also if the load is balanced evenly throughout all the machines in the delivery group.
To view and compare load index between machines click on the following link:
View Server OS machine performance
If high load index is isolated on one machine only, try to check if the load management policy settings fit your needs and also check OS performance parameters on the machines, such as, cpu usage, mem usage, disk usage and disk iops, number of concurrent user sessions, etc.
If high load index is not isolated and exists on all registered servers, check if load balancing works correctly and if new machines, that can handle new users connections, are being powered on and are registering themselves to the delivery controller. Checking if there are machine registration issues can be done by clicking on the following link:
View machine failed registrations
If unused machines exist and are pending in "Unregistered" state or are being regularly restarted there may be registration issues with you machines that need attention.
For more information, see the "Citrix XenApp and XenDesktop User Guide" and especially the following information:
Target | ComTrade.Citrix.XenDesktop.ServerOSMachine | ||
Parent Monitor | System.Health.PerformanceState | ||
Category | Custom | ||
Enabled | True | ||
Alert Generate | True | ||
Alert Severity | Warning | ||
Alert Priority | Normal | ||
Alert Auto Resolve | True | ||
Monitor Type | Citrix.XenDesktop.ServerOSMachine.GenericPerformanceConsecutiveSamplesTwoState.UnitMonitorType | ||
Remotable | True | ||
Accessibility | Public | ||
Alert Message |
| ||
RunAs | Default |