IH Summaries holds wrong counts

General
,
1

Hi

I have data set of IH summaries which is Not Materialized

for development needs I need to push Outcome Time back and forth to check how app behavior will change depending on Interaction History

I have noticed that IH Summaries are initially following my changes for Outcome Time, but after some number of updates it is just stuck and count does not change any more

Did anyone noticed this? Is there a way of bringing this back to live?

I did system reboot with cache clean up and eventually IH truncate to fix this. But this is really not a solution

2

@AndzejS3 I believe your issue was logged against Pega 8.8.2

The explanation provided by our support team:

There was some confusion here because of the way upgraded vs. new PRPC clusters work for IH summaries.

Systems which are upgraded to 8.8 from an earlier Pega release will use IH DB. Systems which are installed starting with 8.8 will use pxResponsesStream by default.

The problem here was that the customer created a new Pega cluster pointed it at an existing set of IH fact records and expected IH summaries to work. But this didn't work because the new PRPC system was expecting to use stream for IH aggregation.

The fix was to make the following changes:
(1) Set the following DSS to ensure IH records ae written to the DB:

Ruleset: Pega-DecisionEngine
Name: interactionHistory/writeToIH
Value: True

(2) Set the following prconfig to make sure the last record time for a customer is written to ih_customer_v2 and ensure ih_customer_v2 table is checked on reading. This is to ensure that topup is enabled. This is also the default value if not specified at all

RuleSet: Pega-Engine
Name: prconfig/ih/aggregates/last_customer_update_method/default
Value: CASSANDRA

(3) Copy the Interaction History Data-Admin-DataFlowScheduler-Aggregation record from an existing cluster to this cluster. This will set the source for pre-aggregation back to IH DB instead of stream

(4) Restart all nodes.

Note:
In 8.8.5, there will be a case to automate the migration from IH DB to Stream and from Stream to IH DB.