[SERVER-6610] all nodes in the seed string of DBClientRS are considered unhidden until proven otherwise Created: 26/Jul/12 Updated: 06/Dec/22 Resolved: 28/Sep/15 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Sharding |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Greg Studer | Assignee: | Backlog - Service Architecture |
| Resolution: | Done | Votes: | 0 |
| Labels: | cxxcopy | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Assigned Teams: |
Service Arch
|
| Operating System: | ALL |
| Participants: |
| Description |
|
... this means when we first connect to the hosts, and potentially on reconnect, we'll potentially start reading from hidden secondaries until the monitoring thread forces a check. Solution is probably to pass the information from the isMaster call in _checkConnection to the creation of the new Node. |
| Comments |
| Comment by Andy Schwerin [ 28/Sep/15 ] |
|
Various improvements to the replica set monitor and to how DBClientRS and the sharding code do targeting have caused this problem to go away. The monitor now only returns nodes that it has affirmatively contacted. |
| Comment by Eric Milkie [ 24/Feb/15 ] |
|
The only use of DBClientReplicaSet in the server is for mongos; I'm not sure if it uses a seed list to initialize it. |