Details
-
Bug
-
Resolution: Duplicate
-
Major
-
7.6.0
-
OS - Debian 10
Couchbase server version - 7.6.0-1568
-
Untriaged
-
Linux x86_64
-
0
-
No
Description
Steps:
- Initialise a 3 node cluster running just the KV service.
- Configure magmaMaximum=0.0016 in http://172.23.106.242:8091/settings/resourceManagement
-
{"bucket":{"collectionsPerQuota":{"enabled":true,"maximum":1},"dataSizePerNode":{"couchstoreMaximum":1.6,"enabled":true,"magmaMaximum":0.0016},"residentRatio":{"couchstoreMinimum":10,"enabled":true,"magmaMinimum":1}},"coresPerBucket":{"enabled":true,"minimum":0.4},"diskUsage":{"enabled":true,"maximum":85}}
- Create a Magma bucket with 256 MB RAM per node.
- Load data until it would hit the max data per node per bucket guardrail.
- Validated the above by inserting a few documents after hitting the guardrail and received the error 'BUCKET_DATA_SIZE_TOO_BIG'.
- Rebalance-out a node from the cluster.
- The rebalance-out operation was allowed and it completed successfully.
- After rebalance-out, the bucket data on the 2 nodes were 2465825479 and 2452661322 bytes, which is more than the configured guardrail limit. This was fetched from this stat kv_logical_data_size_bytes{state="active"}
- Hence, the rebalance-out operation shouldn't be allowed.
Attachments
Issue Links
- duplicates
-
MB-58931 Guard rails other than RR% are not enforced on rebalance
- Closed