Details
-
Bug
-
Resolution: Fixed
-
Critical
-
7.1.0
-
7.1.0-2112
-
Untriaged
-
-
1
-
Yes
Description
Reason for logging: Haven't observed this from so many volume runs in the past. Data loading phase was always smooth without single tmo_ooms or such high dcp queues.
Build 2075 was very fine and no such issues observed.
Steps:
1. On a 4 node cluster, create a magma default bucket with 10gb per node RAM. Create 50 collections.
2. Insert 20M items per collection. During this phase DCP queue was very fine. no major peaks are observed.
3. Upsert all 20M items in each collection. Very high RAM, Peaks in DCP queue and too many tmp_ooms observed.
4. Insert new 20M items per collection. Again high RAM, Peaks in DCP queue and too many tmp_ooms observed.
5. Upsert newly added 20M items per collection. Same observations.
Attachments
Issue Links
- is caused by
-
MB-50546 Folly::UMPMCQueue leads to incorrect bucket memory usage tracking
- Closed