[
https://issues.apache.org/jira/browse/DRILL-5948?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16246292#comment-16246292
]
Paul Rogers commented on DRILL-5948:
------------------------------------
As it turns out, Drill will generate two batches even for a single record:
* The first batch is empty and carries just the schema. (Used for JDBC/ODBC to
report schema up front.)
* The second batch carries the first (or only) set of records.
>From a metric perspective, we could change Drill to not count the initial,
>empty batch.
> The wrong number of batches is displayed
> ----------------------------------------
>
> Key: DRILL-5948
> URL: https://issues.apache.org/jira/browse/DRILL-5948
> Project: Apache Drill
> Issue Type: Bug
> Affects Versions: 1.11.0
> Reporter: Vlad
> Attachments: json_profile.json
>
>
> I suppose, when you execute a query with a small amount of data drill must
> create 1 batch, but here you can see that drill created 2 batches. I think
> it's a wrong behaviour for the drill. Full JSON file will be in the
> attachment.
> {code:html}
> "fragmentProfile": [
> {
> "majorFragmentId": 0,
> "minorFragmentProfile": [
> {
> "state": 3,
> "minorFragmentId": 0,
> "operatorProfile": [
> {
> "inputProfile": [
> {
> "records": 1,
> "batches": 2,
> "schemas": 1
> }
> ],
> "operatorId": 2,
> "operatorType": 29,
> "setupNanos": 0,
> "processNanos": 1767363740,
> "peakLocalMemoryAllocated": 639120,
> "waitNanos": 25787
> },
> {code}
> Step to reproduce:
> # Create JSON file with 1 row
> # Execute star query whith this file, for example
> {code:sql}
> select * from dfs.`/path/to/your/file/example.json`
> {code}
> # Go to the Profile page on the UI, and open info about your query
> # Open JSON profile
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
