Details
-
Bug
-
Resolution: Fixed
-
Major
-
1.0.0-dp.12
-
None
-
None
-
0
Description
Creating this ticket to address these scenarios/questions:
- Currently, during the bootstrap process, the C++ client does not wait for a select_bucket response prior sending a get_cluster_config. That would appear to violate the RFC:
This must occur prior to the cluster configuration request to ensure that a configuration containing bucket-specific context is returned.
- Some recent logs show the a new session that has not completed the bootstrap process was able to receive a response to a get_cluster_config even though it had not yet sent the get_cluster_config request w/in the bootstrap process
- How does a cluster session config update a bucket session config?
- I assume this is correct behavior, but it would appear unclear in code how this is possible
- Some recent logs show that a session can send a KV operation that uses a bucket config that is different than the session’s config.
- I don't think this should be happening.
Attachments
Issue Links
- relates to
-
CXXCBC-472 Handle config updates in scenario where bucket is not found
- Open
-
CXXCBC-143 Segfault in bootstrap_handler when mcbp_session stopped while handling message
- Closed