Admin: Operational management
Maintenance mode
It is possible to put the server into maintenance mode in advance of performing scheduled maintenance activities. Endpoints are provided for both entering and exiting maintenance mode (please see the Configuration Service’s Swagger documentation for more details).
When in maintenance mode, users (other than cluster admin users) are unable to log into the system. When the server enters maintenance mode, any logged-in users (other than cluster admin users) are logged out of the system automatically.
When entering maintenance mode, one or more message(s) can be supplied as the endpoint call’s body. These are displayed when attempting to log into the client.
Node Manager application setting Default Maintenance Mode Buffer Time defines the time prior to scheduled maintenance within which any interaction workflow activity will be placed into a Queued status (regardless of priority). For more information, please see the Interaction Designer Reference Guide.
Service control endpoints
The following service control endpoints, exposed at the Node Manager and Execution Services, facilitate management of the same:
/api/operations/status
Example return body:
{
"currentStatus": "Running",
"message": "Execution Service has restarted",
"executionServiceID": "4a9b3399-52f4-4604-a761-31a035348ef1"
}
/api/operations/start
Example return body:
{
"currentStatus": "Running",
"message": "Node Manager has restarted",
"nodeManagerID": "06343b7e-750b-4140-bef0-c4ba95c7dc52"
}
/api/operations/stop?waitTimeout=500
Example return body:
{
"currentStatus": "Stopped",
"message": "Node Manager has stopped",
"nodeManagerID": "06343b7e-750b-4140-bef0-c4ba95c7dc52"
}
If the Execution Service is stopped when a long-running activity is executing, the service will wait for a duration up to the supplied waitTimeout
(in seconds) for the activity to stop gracefully.
/api/operations/sleep?waitForQuietTimeout=500
Example return body:
{
"CurrentTaskCount": 0,
"CurrentStatus": "Sleeping",
"Message": "Execution Service is sleeping",
"ExecutionServiceID": "f04d33dd-ad74-4475-be57-faff91dc7658"
}
If the Execution Service is put to sleep, no new work will commence. Any long-running activities will continue to execute.
Prometheus integration
All RPI containers now expose support a /metrics endpoint, which can be utilized by the Prometheus monitoring tool. The endpoints provide default .NET metrics, as well as additional information at the following containers:
Execution service
Number of activities running
Total number of activities run since startup
Max number activities configured to run
Node Manager
Number of activities allocated to Execution Services
Number of triggers fired
Queue Reader
Number of items processed