This section covers configuration settings that are specific to SAS Model Studio. To learn how to set these properties in SAS Environment Manager, see SAS Environment Manager: Configuration Page.
This property specifies the maximum number of nodes that can be running for each execution provider and tenant. For example, if there are 3 tenants and this value is 5, there will be a maximum of 15 nodes executing for any execution provider, 5 for each tenant. This property is available under the sas.analytics.execution definition.
Consider increasing the value of this property when you have a lot of users running pipelines. However, this property should be increased only if the deployment has the resources to handle additional nodes running simultaneously. The default value is 10.
This property is applied to the entire sas-analytics-execution deployment. Changing the deployment scaling (number of pods) does not change the maximum number of concurrent node executions across the entire system.
The number of seconds to keep temporary tables in memory after they are last accessed. The OUTFOR table is the only temporary table. The default setting for this is 3600 seconds (one hour). This property is available under the sas.forecasting definition.
The maximum number of CAS sessions to maintain for each user. The default is 10 sessions.
The number of seconds to keep pooled CAS sessions after they are last used. The default is 3,600 seconds.
The maximum number of CAS sessions to maintain across all users. The default is 500.
The following properties should be changed only when directed to by SAS Technical Support or by SAS Research and Development.
These properties control the process of selecting pipeline node job requests from a queue to be sent to the Job Execution Service.
specifies the length of time a node lock is valid. If a node is selected to be executed, and it is not started within this time, it is eligible to be selected again. The default value is two minutes (2m). See Units of Measurement for Configuration Properties to specify the unit of time measurement.
specifies the period to delay after launching a node before selecting the next node to launch. The default value is ten milliseconds (10ms). The value should be no larger than the value of Node Launcher Cycle Time. See Units of Measurement for Configuration Properties to specify the unit of time measurement.
specifies how often to send a signal to all of the tenant work loops to find nodes to launch. The default value is 500 milliseconds (500ms). The value should be at least as large as the value for Node Launcher Rest Period. See Units of Measurement for Configuration Properties to specify the unit of time measurement.
The maximum number of CPU cores utilized on the CAS server by the Interactive Modeling node. The default is zero, for which the CAS server uses all available CPU cores of every worker node. Reducing this number has demonstrated better performance.
This property affects only forecasting projects that include Interactive Modeling.
These properties configure the retry logic of eventan incident that disrupts the normal flow of any process that generates the time series. Examples of events are holidays, retail promotions, and natural disasters. listeners for these specific events:
specifies the period to delay before retrying a failed event by the listener of Flow Job Cancel events. The default is two seconds (2s). See Units of Measurement for Configuration Properties to specify the unit of time measurement.
specifies the number of retry attempts for the listener of Flow Job Cancel events. The default value is 10.
specifies the period to delay before retrying a failed event by the listener of Flow Job Nodes Created events. The default value is two seconds (2s). See Units of Measurement for Configuration Properties to specify the unit of time measurement.
specifies the number of retry attempts for the listener of Flow Job Nodes Created events. The default value is 10.
specifies how long to delay before retrying the listener of Node Job Created events. The default value is two seconds (2s). See Units of Measurement for Configuration Properties to specify the unit of time measurement.
specifies the number of retry attempts for the listener of Node Job Created events.
Missing jobs and transient issues with other services can prevent the analytics execution service from handling job state transitions as intended. This can cause the state of a component (node) or pipeline (graph) to remain in a running state long after the associated job has completed. The following properties can be used to adjust the timing of fallback processes intended to transition these stuck components and pipelines to an appropriate state. Do not adjust these unless directed to do so by SAS support.
The frequency at which the hung graph cleanup task should be run. This task locates graph entries that do not have corresponding running JES jobs and transitions them to the appropriate state. The default value is 60 seconds. See Units of Measurement for Configuration Properties to specify the unit of time measurement.
The minimum time since the last update of a graph that should be checked for corresponding running JES jobs. If the job does not exist or is not running and also has not been updated in this amount of time, the graph state is updated accordingly. The default value is two minutes. See Units of Measurement for Configuration Properties to specify the unit of time measurement.
The frequency at which the hung node cleanup task should be run. This task locates node entries that do not have corresponding running JES jobs and transitions them to the appropriate state. The default value is 30 seconds. See Units of Measurement for Configuration Properties to specify the unit of time measurement.
The minimum time since the last update of a node that should be checked for corresponding running JES jobs. If the job does not exist or is not running and also has not been updated in this amount of time, the node state is updated accordingly. The default value is 60 seconds. See Units of Measurement for Configuration Properties to specify the unit of time measurement.
Use one of the following units of time measurement when updating the configuration.
ns — nanosecondsus or
µs — microsecondsms — millisecondss — secondsm — minutesh — hoursd — daysw — weeksy — years