If some target agent machines are turned off when delivering a software package to a group, does this hold up the delivery to the other machines?
The configuration parameter 'Maximum number of simultaneously connected agents on Scalability Server' determines how many agents a package can be deployed to simultaneously but is a machine that is turned off also using up one of these connections.
So for example if this configuration parameter is set to 25 and 100 agents are targeted of which 50 are turned off, how many agents will actually run a jobcheck at the time of the initial delivery.
After creating a Software Delivery job for an agent machine, the Scalability Server will send a trigger message to this agent. The agent will then run a jobcheck process that connects to a DPU process on the Scalability Server.
The following configuration policy controls the maximum number of DPU processes on the Scalability Server:
DSM -> Software Delivery -> Scalability Server -> Concurrency: Maximum number of simultaneously executing agents on Scalability Server (Default = 25)
For example, if this parameter is set to 25 and 100 agents are targeted of which 50 are turned off, then 25 agents that are not turned off will initially run the jobcheck, automatically triggered by the Scalability Server. So machines that are turned off are not counted as one of the Maximum number of simultaneously connected agents on the Scalability Server.
After the job completes on one agent, and a DPU becomes available, then another agent will be sent a trigger.
Suppose a manual jobcheck is done on a separate agent machine while all DPU's are in use, then this will not result in a job starting immediately on that agent machine as there is no DPU available.
- This is the behavior without WOL (Wake On Lan) having been configured.
- For a very large number of targets with a large number of machines turned off, see technical document TEC481062.