Details
-
Bug
-
Resolution: Duplicate
-
Critical
-
7.6.2
-
Operating System : Microsoft Windows Server 2022
Couchbase Enterprise Edition 7.6.2-3715
-
Untriaged
-
Windows 64-bit
-
-
0
-
Unknown
Description
Steps taken to hit the issue
- Created a 3 node kv cluster
- Created a magma bucket with 3 collections
- Added 10M docs to each collection - dims 4096 vector
- Added 5 fts nodes
- Created 1 index for each collection
- Failed over one of the fts nodes
- Added it back to the cluster after a while during indexing
- The CPU utilization was 100% during this time and the rebalance was stuck at 5.9%.
- The rebalance was hung for 30+ minutes.
- Stopped the rebalance since there was no progress. The indexing was also very slow.
- Post this the cpu percentage never came down. The indexing rate was still very slow.
- Dropped all the indexes.
- Memory used by FTS stays at 89GB even after 120-150 minutes. The memory does not reduce at all.
- The cpu utilization is 100% even after 120-150 minutes post index drop. The cpu utilization does not decrease at all.
- The above memory and cpu issues are observed despite there being no indexes and FTS nodes being in idle state.
The CPU utilization issue is logged separately and can be tracked under MB-62392
Attachments
Issue Links
- is caused by
-
MB-62404 [Windows][Vector Search] : Data ingestion to index hung for 15+ hours
-
- Closed
-
- is duplicated by
-
MB-62392 [Windows][VectorSearch] : CPU utilization stuck at 100% even after dropping all indexes
-
- Closed
-
- relates to
-
MB-62494 [Windows][Vector Search] : Unable to create and use multiple indexes
-
- Closed
-