Details
-
Bug
-
Resolution: Duplicate
-
Major
-
7.2.0
-
7.2.0-5154
-
Untriaged
-
Centos 64-bit
-
-
0
-
Unknown
Description
Build: 7.2.0-5154
Scenario:
- Single node cluster
- Magma bucket with hist. retention enabled
- Perform initial load to create doc_history using cont. updates
- Update history_retention_seconds=10
Updated bucket "bucket-0" (of type couchbase) properties:
[{ram_quota,268435456},{storage_mode,magma},{history_retention_seconds,10}] ... (11:17:54 AM)
- Perform cont. update on few set of keys again
Observation:
During the cont. updates, hist_start_seqno values got reset to '0'
For example, consider this vb_zero:
Before update:
|
[2023-02-12 11:18:01,691] 0: {'active': {'high_seqno': '853', 'history_start_seqno': '1'},
|
..
|
..
|
After update:
|
[2023-02-12 11:18:04,084] 0: {'active': {'high_seqno': '853', 'history_start_seqno': '0'},
|