Details
-
Bug
-
Resolution: Fixed
-
Major
-
2.5.0, 2.4.2
-
2.6.0 - Hopefully Done?
-
1
Description
Using CAO 2.5.0 and Couchbase Server 7.2.0
The Operator repeatedly updates bucket settings unnecessarily on the cluster when no actual configuration has changed:
{"level":"info","ts":1694691294.6770737,"logger":"cluster","msg":"Resource updated","cluster":"default/cb-example","diff":" strings.Join({\n \t... // 82 identical bytes\n \t\"Flush: false\\nenableIndexReplica: false\\nevictionPolicy: fullEvict\",\n \t\"ion\\nhistoryRetentionBytes: 0\\nhistoryRetentionCollectionDefault: \",\n- \t\"true\",\n+ \t\"null\",\n \t\"\\nhistoryRetentionSeconds: 0\\nioPriority: low\\nmaxTTL: 0\\nmemoryQuot\",\n \t\"a: 1024\\nname: magma-history-bucket\\npassword: \\\"\\\"\\nreplicas: 0\\nstor\",\n \t\"ageBackend: magma\\ntype: couchbase\\n\",\n }, \"\")\n"}
|
{"level":"info","ts":1694691294.7036622,"logger":"cluster","msg":"Bucket updated","cluster":"default/cb-example","name":"magma-history-bucket"}
|
{"level":"info","ts":1694691307.6788216,"logger":"cluster","msg":"Resource updated","cluster":"default/cb-example","diff":" strings.Join({\n \t... // 82 identical bytes\n \t\"Flush: false\\nenableIndexReplica: false\\nevictionPolicy: fullEvict\",\n \t\"ion\\nhistoryRetentionBytes: 0\\nhistoryRetentionCollectionDefault: \",\n- \t\"true\",\n+ \t\"null\",\n \t\"\\nhistoryRetentionSeconds: 0\\nioPriority: low\\nmaxTTL: 0\\nmemoryQuot\",\n \t\"a: 1024\\nname: magma-history-bucket\\npassword: \\\"\\\"\\nreplicas: 0\\nstor\",\n \t\"ageBackend: magma\\ntype: couchbase\\n\",\n }, \"\")\n"}
|
{"level":"info","ts":1694691307.705179,"logger":"cluster","msg":"Bucket updated","cluster":"default/cb-example","name":"magma-history-bucket"}
|
{"level":"info","ts":1694691320.6817575,"logger":"cluster","msg":"Resource updated","cluster":"default/cb-example","diff":" strings.Join({\n \t... // 82 identical bytes\n \t\"Flush: false\\nenableIndexReplica: false\\nevictionPolicy: fullEvict\",\n \t\"ion\\nhistoryRetentionBytes: 0\\nhistoryRetentionCollectionDefault: \",\n- \t\"true\",\n+ \t\"null\",\n \t\"\\nhistoryRetentionSeconds: 0\\nioPriority: low\\nmaxTTL: 0\\nmemoryQuot\",\n \t\"a: 1024\\nname: magma-history-bucket\\npassword: \\\"\\\"\\nreplicas: 0\\nstor\",\n \t\"ageBackend: magma\\ntype: couchbase\\n\",\n }, \"\")\n"}
|
{"level":"info","ts":1694691320.7079859,"logger":"cluster","msg":"Bucket updated","cluster":"default/cb-example","name":"magma-history-bucket"}
|
In the diffs in the above lines it looks like it's repeatedly perceiving historyRetentionCollectionDefault as transitioning from true to null.
This is visible to the end user as it spams the logs:
I've attached a minimal reproduction configuration to this ticket:
test-cluster.yaml
And Operator logs from a run of this reproduction:
operator_logs.log
Adding the annotation cao.couchbase.com/historyRetention.collectionHistoryDefault: "true" to the bucket seems to fix this.