[
https://issues.apache.org/jira/browse/DRILL-7269?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dmytriy Grinchenko updated DRILL-7269:
--------------------------------------
Description:
The problem hiding in {{MongoTestSuit}} and the way how the data r bootstrapped
on distributed cluster start up. It looks like not all shard come online, while
we already starting to upload test data sets to DB.
Below is an comparison of data between sharded cluster and single-mode:
{code:title=sharded}
#: full_name
0: "Mary Pierson"
1: "John Reed"
2: "Lynn Kwiatkowski"
3: "Donald Vann"
4: "Judy Owens"
5: "Lori Lightfoot"
{code}
{code:title=single}
#: full_name
0: "Steve Eurich"
1: "Mary Pierson"
2: "John Reed"
3: "Lynn Kwiatkowski"
4: "Donald Vann"
5: "Judy Owens"
6: "Lori Lightfoot"
7: "Kumar"
8: "Kamesh"
{code}
was:
The problem hiding in {{MongoTestSuit}} and the way how the data r bootstrapped
on distributed cluster start up. It looks like not all shard come online, while
we already starting to upload test data sets to DB.
The bug start appear after fixes done in DRILL-7196, where we have deployed a
sharded cluster but have used it as a single-node.
Below is an comparison of data between sharded cluster and single-mode:
{code:title=sharded}
#: full_name
0: "Mary Pierson"
1: "John Reed"
2: "Lynn Kwiatkowski"
3: "Donald Vann"
4: "Judy Owens"
5: "Lori Lightfoot"
{code}
{code:title=single}
#: full_name
0: "Steve Eurich"
1: "Mary Pierson"
2: "John Reed"
3: "Lynn Kwiatkowski"
4: "Donald Vann"
5: "Judy Owens"
6: "Lori Lightfoot"
7: "Kumar"
8: "Kamesh"
{code}
The Mongo server startup logs attached respectively.
> Mongo Unit-tests not able to import properly the test data when running in
> sharded mode
> ---------------------------------------------------------------------------------------
>
> Key: DRILL-7269
> URL: https://issues.apache.org/jira/browse/DRILL-7269
> Project: Apache Drill
> Issue Type: Bug
> Affects Versions: 1.17.0
> Reporter: Dmytriy Grinchenko
> Assignee: Dmytriy Grinchenko
> Priority: Major
>
> The problem hiding in {{MongoTestSuit}} and the way how the data r
> bootstrapped on distributed cluster start up. It looks like not all shard
> come online, while we already starting to upload test data sets to DB.
> Below is an comparison of data between sharded cluster and single-mode:
> {code:title=sharded}
> #: full_name
> 0: "Mary Pierson"
> 1: "John Reed"
> 2: "Lynn Kwiatkowski"
> 3: "Donald Vann"
> 4: "Judy Owens"
> 5: "Lori Lightfoot"
> {code}
> {code:title=single}
> #: full_name
> 0: "Steve Eurich"
> 1: "Mary Pierson"
> 2: "John Reed"
> 3: "Lynn Kwiatkowski"
> 4: "Donald Vann"
> 5: "Judy Owens"
> 6: "Lori Lightfoot"
> 7: "Kumar"
> 8: "Kamesh"
> {code}
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
