Query with group show results inconsistent when I execute several times
I’m testing Druid version 25.0.0 with ingestion realtime from Kafka.
I have a table (my-tabla) with the following columns:
<code>- __time: mapped from source data
- message: Format JSON
- path: String
- id
</code>
<code>- __time: mapped from source data
- message: Format JSON
- path: String
- id
</code>
- __time: mapped from source data
- message: Format JSON
- path: String
- id
When I execute several time in Druid query frontend (admin panel) the following query
<code>SELECT COUNT(*) AS "Count", path
FROM "my-tabla"
GROUP BY (path)
ORDER by "Count" DESC
</code>
<code>SELECT COUNT(*) AS "Count", path
FROM "my-tabla"
GROUP BY (path)
ORDER by "Count" DESC
</code>
SELECT COUNT(*) AS "Count", path
FROM "my-tabla"
GROUP BY (path)
ORDER by "Count" DESC
By count show for example the first time of execution:
11.256.000
Second time (increased as expected):
11.260.000
Third time: (unexpected because decrease)
10.300.000
I tried with different options from Frontend like:
<code>- Use cache
- Use approximate TopN
- Use approximate Count(Distinct)
</code>
<code>- Use cache
- Use approximate TopN
- Use approximate Count(Distinct)
</code>
- Use cache
- Use approximate TopN
- Use approximate Count(Distinct)
And other options of druid broker:
<code> druid.broker.cache.useCache=false
druid.broker.cache.populateCache=false
druid.broker.cache.useResultLevelCache=false
druid.broker.cache.populateResultLevelCache=false
</code>
<code> druid.broker.cache.useCache=false
druid.broker.cache.populateCache=false
druid.broker.cache.useResultLevelCache=false
druid.broker.cache.populateResultLevelCache=false
</code>
druid.broker.cache.useCache=false
druid.broker.cache.populateCache=false
druid.broker.cache.useResultLevelCache=false
druid.broker.cache.populateResultLevelCache=false
¿What configuration can produce this behavior?
I’m expecting that count by Path increase in new query executions
New contributor
Camilo Verdugo is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.