Thresholds

Last modified 14 Mar 2024 12:46 +01:00
Threshold feature
This page is an introduction to Threshold midPoint feature. Please see the feature page for more details.
Since 4.0
This functionality is available since version 4.0.

MidPoint provides a powerful mechanism to synchronize data among different resources. Usually, there is a resource which is authoritative and midPoint should take appropriate actions to keep the data consistent according to the authoritative resource. In most cases, authoritative resource is represented by HR system. However, there are situations when the data in HR are not 100% correct and midPoint should not propagate such data to other resources.

For example, imagine that by the mistake, thousands of accounts was disabled in the HR system and there is a reconciliation task in midPoint scheduled to run every night. If no one noticed the mistake before the reconciliation run it would end up with disabling users across all connected systems. To prevent such situations midPoint has to know which changes are critical and what to do in the case they occur.

Setting Thresholds

For now, it is possible to set up thresholds for different tasks (reconciliation, recomputation, synchronization,…​). It is done by assigning role with defined policy rules to the task. Following is the example for setting thresholds for reconciliation task.

Policy rule settings
<role oid="f45688cc-332c-409b-a67d-5332dc65768d">
    <name>Stop on create 5th user</name>
       <inducement>
        <policyRule>
            <name>Add user</name>
            <policyConstraints>
                <modification>
                    <operation>add</operation>
                </modification>
            </policyConstraints>
            <policyThreshold>
                <lowWaterMark>
                    <count>5</count>
                </lowWaterMark>
            </policyThreshold>
            <policyActions>
                <suspendTask/>
            </policyActions>
        </policyRule>
        <order>2</order>
    </inducement>
</role>

Policy rule above specifies that midPoint should monitor creation of new users (specified by policyConstraints) and when the limit is reached (count=5), task execution is suspended (policyAction=suspendTask). The fifth user is not added.

After defining such role with policy rules, midPoint has to know that such role should be taken into account. Therefore, it is needed to assign this role to the task as in the example below.

Before 4.4

Reconciliation task (before 4.4)
<task oid="10335c7c-838f-11e8-93a6-4b1dd0ab58e4"
    xmlns="http://midpoint.evolveum.com/xml/ns/public/common/common-3">
    <name>Reconciliation: Dummy</name>
    ...
    <handlerUri>http://midpoint.evolveum.com/xml/ns/public/task/lightweigth-partitioning/handler-3</handlerUri>

    <workManagement>
        <partitions>
            <partition>
                <index>1</index>
                <handlerUri>http://midpoint.evolveum.com/xml/ns/public/model/synchronization/task/reconciliation/handler-3</handlerUri>
                <stage>simulate</stage>
                <errorCriticality>
                    <policy>fatal</policy>
                </errorCriticality>
            </partition>
            <partition>
                <index>2</index>
                <handlerUri>http://midpoint.evolveum.com/xml/ns/public/model/synchronization/task/reconciliation/handler-3</handlerUri>
                <stage>execute</stage>
                <errorCriticality>
                    <policy>fatal</policy>
                </errorCriticality>
            </partition>
        </partitions>
    </workManagement>
    ...
    <assignment>
        <targetRef oid="f45688cc-332c-409b-a67d-5332dc65768d" type="RoleType"/>
    </assignment>
</task>

Additionally, we have to add a part of configuration to the task itself regarding task partitioning. In this part we will specify that the task should fail based on a violation of the policy which we have defined above.

We are capable of executing the task in multiple stages which we define with multiple partitions. The task can be executed first in a simulation stage where the task computes the resulting actions without applying them. In this stage we are capable of evaluating the thresholds specified in the policy and in the case of a violation the respective policy action can be initiated. If there is no violation of the threshold we are capable to continue to the execution stage where the computed actions will be applied.

The configuration might as well contain only one partition, only with an execution stage. In this case the action will be applied up until the task reaches the threshold.

4.4 and After

In 4.4 we can use the activities to configure simulation (now called preview) and execution stages.

The selection of execution mode is done by setting executionMode configuration property. The value of preview means we only simulate the execution, without committing any effects. On the other hand, full (the default) means the regular execution.

If we have to run a simple activity, like e.g. import, first in preview and then in full mode, we can use activity composition. It would look like this:

Import task with "preview, then execute" mode - in 4.4 and after
<task xmlns="http://midpoint.evolveum.com/xml/ns/public/common/common-3">
    ...
    <activity>
        <composition>
            <activity>
                <order>1</order>
                <identifier>simulate</identifier>
                <work>
                    <import>
                        <resourceObjects>
                            <resourceRef oid="40f8fb21-a473-4da7-bbd0-7019d3d450a5" />
                            <kind>account</kind>
                            <intent>default</intent>
                        </resourceObjects>
                    </import>
                </work>
                <executionMode>preview</executionMode>
            </activity>
            <activity>
                <order>2</order>
                <identifier>execute</identifier>
                <work>
                    <import>
                        <resourceObjects>
                            <resourceRef oid="40f8fb21-a473-4da7-bbd0-7019d3d450a5" />
                            <kind>account</kind>
                            <intent>default</intent>
                        </resourceObjects>
                    </import>
                </work>
                <executionMode>full</executionMode>
            </activity>
        </composition>
    </activity>
</task>
In the future we plan to eliminate the need of specifying the work section twice for such tasks.

The reconciliation task, being a pre-defined composite activity, is already prepared for this preview + execution combination. It contains two sub-activities, namely resourceObjectsPreview and remainingShadowsPreview that are prepared to be run in preview execution mode. They are disabled by default. So the only thing you have to do is to enable them by appropriate tailoring:

Reconciliation task (4.4 and after)
<task
    xmlns="http://midpoint.evolveum.com/xml/ns/public/common/common-3"
    xmlns:ri="http://midpoint.evolveum.com/xml/ns/public/resource/instance-3">

    <name>Reconciliation in Preview and Execution Mode</name>

    ...

    <assignment>
        <targetRef oid="f45688cc-332c-409b-a67d-5332dc65768d" type="RoleType"/>
    </assignment>

    <activity>
        <work>
            <reconciliation>
                <resourceObjects>
                    <resourceRef oid="10000000-0000-0000-0000-000000000003" />
                    <kind>account</kind>
                    <intent>default</intent>
                    <objectclass>ri:inetOrgPerson</objectclass>
                </resourceObjects>
            </reconciliation>
        </work>
        <tailoring>
            <change>
                <reference>resourceObjectsPreview</reference>
                <controlFlow>
                    <processingOption>process</processingOption>
                </controlFlow>
            </change>
            <change>
                <reference>remainingShadowsPreview</reference>
                <controlFlow>
                    <processingOption>process</processingOption>
                </controlFlow>
            </change>
        </tailoring>
    </activity>
</task>
Was this page helpful?
YES NO
Thanks for your feedback