-
Type: Bug
-
Resolution: Done
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: Cluster Management
-
None
There are a number of issues with DB.requestStart/requestDone:
- If client reads from a primary or secondary, subsequent reads may use a different connection. Every read goes back to the pool for a new connection (which could potentially block or fail, depending on MongoOptions.maxWaitTime)
- If client reads from a secondary (any, or tagged), there is no guarantee that subsequent reads will go to the same secondary
- If client reads from a secondary (any, or tagged), and then writes, subsequent reads will go the primary regardless of the read preference