-
Type:
Question
-
Resolution: Done
-
Priority:
Major - P3
-
None
-
Affects Version/s: 3.6.5
-
Component/s: Sharding
-
None
-
None
-
3
-
None
-
None
-
None
-
None
-
None
-
None
We have upgraded MongoDB sharded cluster from 2.0 to 3.6 in the past 6 months.
And then, we enabled balancer, but no chunk migration occurred for 5 hours.
After we enabled blancer, primary config server logged,
2018-XX-XXT09:36:39.007+0900 I SHARDING [Balancer] Refreshing chunks for collection db.sharded-collection based on version 0|0||000000000000000000000000 2018-XX-XXT09:36:39.103+0900 I SHARDING [ConfigServerCatalogCacheLoader-1] Refresh for collection db.sharded-collection took 95 ms aand found version 7|27819||5a66b9b7eb81557948d33b6f 2018-XX-XXT13:50:29.159+0900 I SHARDING [Balancer] Refreshing chunks for collection db.sharded-collection based on version 7|27819||5a66b9b7eb81557948d33b6f 2018-XX-XXT13:50:29.164+0900 I SHARDING [ConfigServerCatalogCacheLoader-2] Refresh for collection db.sharded-collection took 4 ms and found version 7|27819||5a66b9b7eb81557948d33b6f
Our dev MongoDB cluster's primary config server logged "version 22" and it migrates chunks well.
Currently, balancer is disabled. We have never used balancer before. So, chunks are quite unevenly distributed.
I attached our server's stats. No jumbo chunk exists. Any guesses on what blocks balancer's chunk migration and possible solution?