Details
-
Bug
-
Resolution: Done
-
Major - P3
-
None
-
2.2.2
-
None
-
ALL
Description
In our cluster the balancer spends a fair amount of time trying to move around chunks for a database that doesn't exist any longer.
The database name is "aggregate":
exception: Dropping collection failed on the following hosts:
|
set1/mongo1.domain.net:27019: { errmsg: \"ns not found\", ok: 0.0 },
|
set2/mongo2.domain.net:27019: { errmsg: \"ns not found\", ok: 0.0 }
|
in the config DB:
mongos> db.databases.find({"_id": /aggregate/})
|
{ "_id" : "aggregate_2", "partitioned" : false, "primary" : "shard0001" }
|
Note that aggregate_2 is a different database - the point is that the database "aggregate" is not present
mongos> db.chunks.find({"ns": /aggregate\./}).count()
|
2271
|
Just to be sure, no chunks associated with aggregate_2:
mongos> db.chunks.find({"ns": /aggregate_2/}).count()
|
0
|
I am not sure how we could end up in this situation but I would like to find a way out, is it safe to just delete all chunks in config.chunks associated with the "aggregate" ns?
mongoDB version 2.2.2