...
Make sure to start the correct Kyvos Manager from the correct directory.
Data Integrity Concerns
...
Measure values aren’t matching with expected results/other BI software
Check the filter conditions on register files
Check the SQL query on register files
Check the number of records in the cube build summary
Ensure that the source table had valid records when the cube was scheduled for build
No data/ETL load was under process when the cube build was running
In the Kyvos register files, validate that the column data types are correct as per the source data
Check the summary function of measure(s) in the cube design
Are your records falling beyond the configured cube sliding window strategy?
Check the join condition/column of your dimension and fact tables in cube DRD
Check if the discrepancy is because aggregates are shown as “unknown”
NOTE:NULL or blank does not contribute to aggregations in Kyvos, is your BI tool doing otherwise?
...
Dimension records are missing in viewer filter/browsing results
In the raw data of dimension table, check the values in all the columns of dimension for the missing record, are all column values valid as per column data type?
Ensure that the missing record(s) were present in the source table when the cube was scheduled for build.
Are you missing record(s) which has NULL/blank value(s) in any column?
If it is because of NULL/blank values, then set appropriate configurations as mentioned in the valid values table.
...
How does Kyvos handle special values like NULL, invalid value, and blanks?
Refer to the following table that shows how various special values are treated with respect to the data type of source column and the cube entity which is created using the source column.
Also, see Refine > Advanced Properties > Data Cleansing on the cube designer screen.
Data Type | Example | Level | Attribute | Measure | Primary Key |
CHAR | “” (blank) | Valid | Valid | Invalid | Invalid |
CHAR | NULL | Invalid | Invalid | Invalid | Invalid |
NUMBER (long) |
| Invalid | Invalid | Invalid | Invalid |
NUMBER (double, float, decimal) |
| Invalid | Invalid | Invalid | Invalid |
Date |
| Invalid | Invalid | Invalid | Invalid |
Behaviour of kyvos.build.blankvalue.badrow
Entity | TRUE (Default) | FALSE |
Dimension | The complete record gets skipped. | Invalid value is shown as NULL. |
Dimension | The complete record gets skipped. | The complete record gets skipped. |
Measure | Measure value will not be considered in totals. | Measure value will not be considered in totals. |
Primary Key | Record will be skipped. | Record will be skipped. |
Foreign Key | The totals of this record will be shown as unknown. | The totals of this record will be shown as unknown. |
...
What are valid/invalid values for Levels on which Date Data Type is configured?
Date Data Type is not respected/supported on:
Regular/Location Type dimension
Level if Date Data Type is not configured on any of its parent Levels in the hierarchy
Attribute(s) of any dimension
Date Data Type | Valid Values |
Year | Any valid number >= 1970 |
Quarter |
|
Month |
|
Week (Of Year) |
|
Day (Of Month) | Valid number according to Year-Month to which it belongs to according to Gregorian calendar |
DATE | Any valid date >= 1-Jan-1970 as per configured date format |
Rule for week start day
Kyvos supports SUNDAY as the week start day. Hence, any custom time hierarchy having a WEEK-type level must comply with this prerequisite.
...
If the hierarchy is Y-Q-M-W-D, then the week number for 2009-Jan-03 must be 1, while the week number for 2009-Jan-04 (SUNDAY) must be 2.
| Year | Quarter | Month | Week | Day |
Record 1 | 2009 | 1 | 1 | 1 | 3 |
Record 2 | 2009 | 1 | 1 | 1* | 4 |
*Shows invalid data.
Rule for a week that spans across the year
...
For example, consider the last WEEK of the Year 2008 (it has first 4 days in December 2008 and last 3 Days in January 2009).
Valid data example:
| Year | Quarter | Month | Week | Day |
Day 1 | 2008 | 4 | 12 | 53 | 28 |
Day 2 | 2008 | 4 | 12 | 53 | 29 |
Day 3 | 2008 | 4 | 12 | 53 | 30 |
Day 4 | 2008 | 4 | 12 | 53 | 31 |
Day 5 | 2009 | 1 | 1 | 1 | 1 |
Day 6 | 2009 | 1 | 1 | 1 | 2 |
Day 7 | 2009 | 1 | 1 | 1 | 3 |
Invalid data example:
| Year | Quarter | Month | Week | Day |
Example 1 | 2008 | 4 | 12 | 1* | 29 |
Example 2 | 2009 | 1 | 1 | 53* | 2 |
*Shows invalid data.
Miscellaneous
...
How to get a count of queries that were run against the dimension/measures for Kyvos cube?
The Query Analyzer feature in Kyvos helps with your request in getting the details regarding the count of queries that were run against the dimension/measures.
Navigate to the cube > Click on the three dots button > Analyze Queries.
...
How to find the performance of Kyvos queries?
It can be seen from Live Query Analysis and Historic Query Analysis in the workbook section.
Navigate to Kyvos Menu -> Workbooks -> Query Analysis
...
Getting frequent alerts for memory shoot-up on Kyvos Manager
Check the buffer/cache memory on the node for which the alerts are triggered.
It's safe to clear the buffer/cache memory from the terminal.
Or change the alert level for these alerts in the Alerts and High Availability section from Kyvos Manager.
...
Unable to initiate new operations, such as restart cluster or restart BI server
This is generally caused by the locking of the Derby DB.
Steps to remove Derby DB lock:
Stop the KyvosManager Kyvos Manager server from the terminal by executing the stop command.
Navigate to kyvosmanagerdata/server/db/ folder and delete all the log files ending with .lck extension.
Navigate to kyvosmanagerdata/server/db/ankushdb/ folder and delete the db.lck and dbex.lck files.
Start the KyvosManager server from the terminal by executing the stop command.
...
Preview on Register file created over SQL based registered file is failing for dynamic schema
For dynamic schema where the table is created over two parquet files, one with data type as int and another as double, add a new property spark.sql.hive.convertMetastoreParquet=false in the Cube Advanced properties.
The property controls whether to use the built-in Parquet reader and writer for Hive tables with the parquet storage format (instead of Hive SerDe). The default value is set as true.
Internally, this property enables RelationConversions logical rule to convert HiveTableRelations to HadoopFsRelation
Additional questions
What if my data limits are reached after a few incremental builds?
How can I connect to Kyvos cubes from Tableau and Excel?
Does Kyvos support row/column level security?
My queries are taking more time, How can I monitor and analyze slow-running queries?
What happens if my build fails?
I accepted all recommendations, but my build failed. What can I do now?
How many fact rows can I have?
What are the data limits?
What are the entities that I can add to the cube without a full build?
What are the entities that I can modify in a cube without a full build?
Kyvos Lite shows these msgs on KM
AWS environment validations have failed
Auto Deployment will not be attempted
...
Go to the Validation URL from the Outputs tab on AWS Console. View the error or check details by downloading logs.
...