Details
-
Bug
-
Resolution: Fixed
-
Major
-
7.6.0
-
OS - Debian 10
Couchbase server version - 7.6.0-1606
-
Triaged
-
Linux x86_64
-
0
-
No
Description
Steps:
- Initialise a 3 node cluster running just the KV service.
- Create a Magma bucket named 'default' with 512MB RAM per node and 1 replica.
- Load 11 million documents into the bucket. After initial loading, the disk usage on the three nodes was [1.0, 1.0, 1.0]
- Now configure maximum=0.5 for diskUsage guardrail in http://172.23.106.202:8091/settings/resourceManagement
This is done to breach the disk usage per node guardrail. - Created a new bucket, validated that mutations were blocked for both the buckets by inserting a few documents and got the error 'BUCKET_DISK_SPACE_TOO_LOW'.
- Updated the replicaNumber to 2 for bucket 'default' and rebalanced the cluster.
- Disk usage of the three nodes after replica update was [1.0, 1.0, 2.0]
- Since increasing of replicas would add more data to the disk and would cause an increase in the disk usage, the operation shouldn't be allowed.
Attachments
For Gerrit Dashboard: MB-59130 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
201570,6 | MB-59130: Disk GR must be not breached by num_replicas increase | master | ns_server | Status: NEW | 0 | 0 |
200341,5 | MB-59130: Refactor additional_bucket_params_validation | master | ns_server | Status: MERGED | +2 | +1 |
200342,19 | MB-59130: Reject increase of num_replicas if disk GR breached | master | ns_server | Status: MERGED | +2 | +1 |