Server cluster - Parallelization
Last updated
Last updated
In order to access this section in LIBSAFE, the menu option is as follows: Advanced Options >> System Configuration (Cluster Processing Configuration).
LIBSAFE has an advanced functionality for the management and monitoring of servers when they work in parallel to perform the activities of the system, seeking to optimize time and performance. This functionality is called Processing Cluster and allows the system to add nodes (equivalent to servers) and services in each node in different states and amounts, so as to perform actions even in a parallel (or multiple) way. With this, LIBSAFE increases its performance power and with its integrated panel within the platform it is possible to monitor the status of each node and service in a simple, intuitive and practical way.
Each node can store a number of services defined in the system's own configuration, even allowing for one or more services of the same type in each node. In addition, each service can have different activity states, independent of the node state (services will have states whenever the node is On). Nodes can only have two states (On or Off), but they can have an exceptional state which is the Deleted state. This state, which represents that the node was created and later deleted, is kept in LIBSAFE with the aim of presenting all the activity carried out by the user since its beginning until the current date. In the node’s filter options you can visualize All at once, Only the Active or Inactive or Deleted.
From the same monitoring panel, new nodes can be created using the "New processing node" option on the upper left of the node list header.
Each node has useful information for its identification, current status and relative or user-defined data for its categorization. It also has direct access to its edition, visualization of operation details and actions that may affect the services. These accesses or links can change independently and be available or not depending on the case at a given time; this is how a node cannot allow its edition if any of its services is under a change request or can change its color depending on the response or status indicators of each one. There is a color and status legend in the window (lower left part of the node panel) that indicates the node status and its color.
From the monitoring panel, status change actions can be performed on each of the services that each node contains; this is how the system can modify its potential in the performance of a service in multitasking or multi-service format. The option to add new services to a node is only made from the node's edit link. From the general panel the services can be changed to the following states: READY, STOP, SLEEP or DELETE:
READY refers to the fact that the service will be active and can execute work at any time
STOP indicates that the service will be stopped
SLEEP indicates that the service will be in an active state but will not execute any task
Since the node monitoring panel is the virtual representation of a physical infrastructure, there may be a time lapse between the invocation of the action and the actual performance in the hardware. Therefore, the panel manages a state of transit between any two given states, which limits or conditions the availability of other actions at a given time.
When the user wants to change the state of a service, this will be in a state of pending change, which will then be confirmed by the hardware so that the user interface can continue managing the panel of nodes and services.
Only in this case, in the action options available for this service, a cancellation of the service can be made, which will reverse the request made, as long as the hardware has not finished executing the action. As shown in the previous image, the node edit option has disappeared in a logical response to the action of changing the state of a service belonging to it.
