[DRILL-4064] Querying a sharded MongoDB database sometimes returns zero rows

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.3.0
• Component/s: Storage - MongoDB
• Labels: cluster database mongodb sharding

Environment

Linux cluster

• Assignee: Unassigned
• Reporter: Nathan Griffith
• Created: 10/Nov/15 22:19
• Updated: 11/Nov/15 17:43

Description

Querying works as intended if the database is unsharded.

In my scenarios the database's shard key is not hashed, and is set to mongo's "_id" key.

Queries work correctly on a sharded database imported from 268M of tweet JSON, but fail to return rows on a sharded database imported from 218M of data structured like:

{
  "data0":"an entry",
  "data1":"another entry",
  "data2":"whoah another"
}
{
  "data0":"an entry",
  "data1":"another entry",
  "data2":"whoah another"
}
{
  "data0":"an entry",
  "data1":"another entry",
  "data2":"whoah another"
}
{
  "data0":"an entry",
  "data1":"another entry",
  "data2":"whoah another"
}